def each_brake(each_wait_list,L,W):
def get_sqare_rate(in_brake_sort, L,W):
brake_boat=quick_sort_brake(in_brake_sort, L, W)
s=0
for k,v in brake_boat.items():
s=s+v[1][0]*v[1][1]
sqare_rate=s/(L*W)
return sqare_rate,brake_boat
wait_list=each_wait_list
"""===============================实例化问题对象==========================="""
problem = MyProblem(wait_list, L, W) # 生成问题对象
"""=================================种群设置==============================="""
Encoding = 'P' # 编码方式
NIND = 6000 # 种群规模
# ranges还是原来的，Field会在最后一行加上1
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges, problem.borders) # 创建区域描述器
population = ea.Population(Encoding, Field, NIND) # 实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
"""===============================算法参数设置============================="""
myAlgorithm = ea.soea_SEGA_templet(problem, population) # 实例化一个算法模板对象，单目标模板
# myAlgorithm=ea.moea_NSGA2_templet(problem, population) #多目模板
myAlgorithm.MAXGEN = 15 # 13 # 最大进化代数
# myAlgorithm.recOper = ea.Xovox(XOVR=0.8) # 设置交叉算子 __init__(self, XOVR=0.7, Half=False)
# myAlgorithm.mutOper = ea.Mutinv(Pm=0.2) # 设置变异算子
myAlgorithm.logTras = 1 # 设置每多少代记录日志，若设置成0则表示不记录日志
myAlgorithm.verbose = True # 设置是否打印输出日志信息
myAlgorithm.drawing = 1 # 设置绘图方式（0：不绘图；1：绘制结果图；2：绘制目标空间过程动画；3：绘制决策空间过程动画）
"""==========================调用算法模板进行种群进化======================="""
[population, obj_trace, var_trace] = myAlgorithm.run() # 执行算法模板
population.save() # 把最后一代种群的信息保存到文件中
# 输出结果
best_gen = np.argmin(problem.maxormins * obj_trace[:, 1]) # 记录最优种群个体是在哪一代
best_ObjV = obj_trace[best_gen, 1]
print('最优的目标函数值为：%s' % (best_ObjV))
print('最优的控制变量值为：')
for i in range(var_trace.shape[1]): # (MAXGEN,Dim),进化的总代数和决策变量的维度
print(var_trace[best_gen, i])
best_sort_sequence = [int(each) for each in var_trace[best_gen]] # (4000, 12)
#best_brake_seq={i:wait_list[i] for i in best_sort_sequence}
# best_brake_seq = wait_list[best_sort_sequence]
# all_brake_boat = {}
# in_brake_sort = best_brake_seq
# sqare_rate,brake_boat=get_sqare_rate(in_brake_sort, L,W)
# brake_num=list(brake_boat.keys())
# brake_boat={best_sort_sequence[k]:v for k,v in brake_boat.items()}
# all_brake_boat[0]= {'brake_boat':brake_boat,'best_use_rate':sqare_rate}
# print('一闸最优解：组合={} \n 面积利用率={}'.format(brake_boat,sqare_rate))
#
# each2=np.delete(best_sort_sequence,brake_num,axis=0)
# in_brake_sort2=np.delete(in_brake_sort,brake_num,axis=0)
# if len(in_brake_sort2)>0:
# sqare_rate2,brake_boat2=get_sqare_rate(in_brake_sort2, L,W)
# brake_num2=list(brake_boat2.keys())
# #sqare_rate=sqare_rate+sqare_rate2
# brake_boat2={each2[k]:v for k,v in brake_boat2.items()}
# all_brake_boat[1] = {'brake_boat':brake_boat2,'best_use_rate':sqare_rate2}
# print('二闸最优解：组合={} \n 面积利用率={}'.format(brake_boat2,sqare_rate2))
#
# each3=np.delete(each2,brake_num2,axis=0)
# in_brake_sort3=np.delete(in_brake_sort2,brake_num2,axis=0)
# if len(in_brake_sort3)>0:
# sqare_rate3,brake_boat3=get_sqare_rate(in_brake_sort3, L,W)
# brake_num3=list(brake_boat3.keys())
#
# brake_boat3={each3[k]:v for k,v in brake_boat3.items()}
# all_brake_boat[2] = {'brake_boat':brake_boat3,'best_use_rate':sqare_rate3}#brake_boat3
# print('三闸最优解：组合={} \n 面积利用率={}'.format(brake_boat3,sqare_rate3))
# #sqare_rate=sqare_rate+sqare_rate3
# print('all_brake_boat00',all_brake_boat)
best_brake_seq = wait_list[best_sort_sequence]
all_brake_boat = {}
all_brake_times = 0
in_brake_sort = best_brake_seq
while True:
if len(in_brake_sort) > 0 and all_brake_times < 3:
sqare_rate, brake_boat = get_sqare_rate(in_brake_sort, L, W)
# 将闸室序号，映射到原始信号
brake_num = list(brake_boat.keys())
# 更新in_brake_sort
##each2=np.delete(each,brake_num,axis=0)
in_brake_sort = np.delete(in_brake_sort, brake_num, axis=0)
brake_boat = {best_sort_sequence[k]: v for k, v in brake_boat.items()}
best_sort_sequence = np.delete(best_sort_sequence, brake_num, axis=0)
all_brake_boat[all_brake_times] = {'brake_boat': brake_boat, 'best_use_rate': sqare_rate}
print('第{}闸最优解：组合={} \n 面积利用率={}'.format(all_brake_times, brake_boat, sqare_rate))
#绘制并保存图形
plot_save(brake_boat, all_brake_times)
all_brake_times = all_brake_times + 1
else:
break
print(' all_brake_boat11', all_brake_boat)
#brake_boat = quick_sort_brake(best_brake_seq, L, W)
##将快速入闸的顺序，对应到最优选择的顺序
#brake_boat={best_sort_sequence[k]:v for k,v in brake_boat.items()}
##将最优选择的顺序，对应到最原始的队列中的序号
#brake_boat={wait_list_num[k]:v for k,v in brake_boat.items()}
print('有效进化代数：%s' % (obj_trace.shape[0]))
print('最优的一代是第 %s 代' % (best_gen + 1))
print('评价次数：%s' % (myAlgorithm.evalsNum))
print('时间已过 %s 秒' % (myAlgorithm.passTime))
return all_brake_boat
ranges = np.vstack(x1) # 生成自变量的范围矩阵
borders = np.vstack([1, 1]) # 生成自变量的边界矩阵
"""========================遗传算法参数设置========================="""
NIND = 50 # 种群规模
MAXGEN = 10 # 最大遗传代数
GGAP = 0.8 # 代沟：子代与父代个体不相同的概率为0.8
selectStyle = 'sus'
# 遗传算法的选择方式设为"sus"——随机抽样选择
recombinStyle = 'xovdp' # 遗传算法的重组方式，设为两点交叉
recopt = 0.9 # 交叉概率
pm = 0.1 # 变异概率
SUBPOP = 1 # 设置种群数为1
maxormin = -1 # 设置最大最小化目标标记为1，表示是最小化目标，-1则表示最大化目标
"""=========================开始遗传算法进化========================"""
FieldD = ga.crtfld(ranges, borders) # 调用函数创建区域描述器
Lind = np.sum(FieldD[0, :]) # 计算编码后的染色体长度
print(FieldD)
Chrom = ga.crtbp(NIND, FieldD) # 根据区域描述器生成二进制种群
Phen = ga.bs2int(Chrom, FieldD) #对初始种群进行解码
LegV = np.ones((NIND, 1)) # 初始化种群的可行性列向量
[ObjV, LegV] = aimfuc(Phen, LegV) # 计算初始种群个体的目标函数值
# 定义进化记录器，初始值为nan
pop_trace = (np.zeros((MAXGEN, 2)) * np.nan)
# 定义种群最优个体记录器，记录每一代最优个体的染色体，初始值为nan
ind_trace = (np.zeros((MAXGEN, Lind)) * np.nan)
# 开始进化！！
start_time = time.time() # 开始计时
for gen in range(MAXGEN):
FitnV = ga.ranking(maxormin * ObjV, LegV) # 根据目标函数大小分配适应度值
SelCh = ga.selecting(selectStyle, Chrom, FitnV, GGAP, SUBPOP) # 选择
# -*- coding: utf-8 -*-
import geatpy as ea # import geatpy
from MyProblem import MyProblem # 导入自定义问题接口
if __name__ == '__main__':
"""================================实例化问题对象==========================="""
problem = MyProblem() # 生成问题对象
"""==================================种群设置=============================="""
NIND = 30 # 种群规模
# 创建区域描述器，这里需要创建两个，前1个变量用RI编码，剩余变量用排列编码
Encodings = ['RI', 'P'] # 编码方式列表
Field1 = ea.crtfld(Encodings[0], problem.varTypes[:1],
problem.ranges[:, :1],
problem.borders[:, :1]) # 创建区域描述器
Field2 = ea.crtfld(Encodings[1], problem.varTypes[1:],
problem.ranges[:, 1:], problem.borders[:, 1:])
Fields = [Field1, Field2]
population = ea.PsyPopulation(Encodings, Fields,
NIND) # 实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
"""================================算法参数设置============================"""
myAlgorithm = ea.moea_psy_NSGA2_templet(problem, population) # 实例化一个算法模板对象
myAlgorithm.MAXGEN = 200 # 最大进化代数
"""===========================调用算法模板进行种群进化======================"""
NDSet = myAlgorithm.run() # 执行算法模板，得到帕累托最优解集NDSet
NDSet.save() # 把结果保存到文件中
# 输出
print('用时：%s 秒' % (myAlgorithm.passTime))
print('非支配个体数：%s 个' % (NDSet.sizes))
print('单位时间找到帕累托前沿点个数：%s 个' % (int(NDSet.sizes // myAlgorithm.passTime)))
x1 = [-1, 2] # 自变量范围
b1 = [1, 1] # 自变量边界
codes = [1] # 变量的编码方式，2个变量均使用格雷编码
precisions = [5] # 变量的精度
scales = [0] # 采用算术刻度
ranges = np.vstack([x1]).T # 生成自变量的范围矩阵
borders = np.vstack([b1]).T # 生成自变量的边界矩阵
"""========================遗传算法参数设置========================="""
NIND = 40
# 种群个体数目
MAXGEN = 25
# 最大遗传代数
GGAP = 0.9
# 代沟：说明子代与父代的重复率为0.1
"""=========================开始遗传算法进化========================"""
FieldD = ga.crtfld(ranges, borders, precisions, codes, scales) # 调用函数创建区域描述器
Lind = np.sum(FieldD[0, :]) # 计算编码后的染色体长度
Chrom = ga.crtbp(NIND, Lind) # 根据区域描述器生成二进制种群
variable = ga.bs2rv(Chrom, FieldD) #对初始种群进行解码
ObjV = aim(variable) # 计算初始种群个体的目标函数值
pop_trace = (np.zeros((MAXGEN, 2)) * np.nan) # 定义进化记录器，初始值为nan
ind_trace = (np.zeros(
(MAXGEN, Lind)) * np.nan) # 定义种群最优个体记录器，记录每一代最优个体的染色体，初始值为nan
# 开始进化！！
for gen in range(MAXGEN):
FitnV = ga.ranking(-ObjV) # 根据目标函数大小分配适应度值(由于遵循目标最小化约定，因此最大化问题要对目标函数值乘上-1)
SelCh = ga.selecting('sus', Chrom, FitnV, GGAP) # 选择，采用'sus'随机抽样选择
SelCh = ga.recombin('xovsp', SelCh, 0.7) # 重组(采用单点交叉方式，交叉概率为0.7)
SelCh = ga.mutbin(SelCh) # 二进制种群变异
variable = ga.bs2rv(SelCh, FieldD) # 对育种种群进行解码(二进制转十进制)
ObjVSel = aim(variable) # 求育种个体的目标函数值
def optimize(datasetX, datasetY, featureNamesAfterCombination,
featureOptimizationParams, datasetParams,
forecastModelParams):
featureNamesAfterOptimization = []
featureOptimizationProcess = None
isPerformOptimization = featureOptimizationParams[
"isPerformOptimization"]
if isPerformOptimization:
dim = len(featureNamesAfterCombination)
problem = FeatureOptimizationProblem(datasetX, datasetY,
featureNamesAfterCombination,
datasetParams,
forecastModelParams)
encoding = 'RI'
popNum = 10
field = ea.crtfld(encoding, problem.varTypes, problem.ranges,
problem.borders)
pop = ea.Population(encoding, field, popNum)
if dim < 7:
# 维度不高时采用穷举
Phen = []
for i in range(1, pow(2, dim)):
numStr = MathNormalUtils.toBinaryWithFixedLength(i, dim)
numArray = list(map(int, numStr))
Phen.append(numArray)
# 受维数灾影响 随机优化5个特征组合
resultlist = random.sample(range(len(Phen)), 3)
Phen = np.array(Phen)
# Phen = Phen[resultlist, :]
pop.Phen = Phen
problem.aimFunc(pop)
objTrace, varTrace = pop.ObjV, pop.Phen
objTrace = NumpyUtils.hStackByCutHead(objTrace, objTrace)
else:
algorithm = ea.soea_SEGA_templet(problem, pop)
# 算法最大进化代数
algorithm.MAXGEN = 10
# 0表示不绘图；1表示绘图；2表示动态绘图
algorithm.drawing = 1
[pop, objTrace, varTrace] = algorithm.run()
featureOptimizationProcess = {
"objTrace": objTrace,
"varTrace": varTrace
}
bestGen = np.argmin(problem.maxormins * objTrace[:, 1])
bestVar = varTrace[bestGen, :]
bestVar = [int(x) for x in bestVar]
bestVar = np.array(bestVar)
featureCombinationParams = {
"isPerformCombination": True,
"featureFlags": bestVar
}
featureCombination = FeatureCombination(
datasetX, datasetY, featureNamesAfterCombination,
featureCombinationParams)
datasetAfterOptimizationX, datasetAfterOptimizationY, featureNamesAfterOptimization = featureCombination.getDatasetAfterCombination(
)
else:
datasetAfterOptimizationX = datasetX
datasetAfterOptimizationY = datasetY
featureNamesAfterOptimization = featureNamesAfterCombination
dataset = Dataset(datasetAfterOptimizationX,
datasetAfterOptimizationY, datasetParams)
forecastTemplate = ForecastTemplate(dataset, forecastModelParams)
loss = forecastTemplate.getObjective()
if (datasetAfterOptimizationX is None) or (datasetAfterOptimizationY is
None):
raise Exception("特征优化中全部特征未被选取，请检查参数！")
return datasetAfterOptimizationX, datasetAfterOptimizationY, featureNamesAfterOptimization, featureOptimizationProcess
# -*- coding: utf-8 -*-
"""main.py"""
import numpy as np
import geatpy as ea # import geatpy
import sys, os
sys.path.insert(0,
os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
from mix_encoder.MyProblem import MyProblem # 导入自定义问题接口
"""===========================实例化问题对象========================"""
problem = MyProblem() # 生成问题对象
"""=============================种群设置==========================="""
NIND = 40 # 种群规模
# 创建区域描述器，这里需要创建两个，前2个变量用RI编码，其余用排列编码
Encodings = ['RI', 'P']
Field1 = ea.crtfld(Encodings[0], problem.varTypes[:2], problem.ranges[:, :2],
problem.borders[:, :2])
Field2 = ea.crtfld(Encodings[1], problem.varTypes[2:], problem.ranges[:, 2:],
problem.borders[:, 2:])
Fields = [Field1, Field2]
population = ea.PsyPopulation(Encodings, Fields,
NIND) # 实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
"""===========================算法参数设置=========================="""
myAlgorithm = ea.soea_psy_SEGA_templet(problem, population) # 实例化一个算法模板对象
myAlgorithm.MAXGEN = 25 # 最大进化代数
"""======================调用算法模板进行种群进化===================="""
[population, obj_trace, var_trace] = myAlgorithm.run() # 执行算法模板
population.save() # 把最后一代种群的信息保存到文件中
# 输出结果
best_gen = np.argmax(obj_trace[:, 1]) # 记录最优种群是在哪一代
best_ObjV = obj_trace[best_gen, 1]
print('最优的目标函数值为：%s' % (best_ObjV))
x = np.linspace(-1, 2, 200)
plt.plot(x, aim(x)) # 绘制目标函数图像
"""============================变量设置============================"""
x1 = [-1, 2] # 自变量范围
b1 = [1, 1] # 自变量边界
varTypes = np.array([0]) # 自变量的类型，0表示连续，1表示离散
Encoding = 'BG' # 'BG'表示采用二进制/格雷编码
codes = [1] # 变量的编码方式，2个变量均使用格雷编码
precisions = [4] # 变量的编码精度
scales = [0] # 采用算术刻度
ranges = np.vstack([x1]).T # 生成自变量的范围矩阵
borders = np.vstack([b1]).T # 生成自变量的边界矩阵
"""=========================遗传算法参数设置========================="""
NIND = 40 # 种群个体数目
MAXGEN = 25 # 最大遗传代数
FieldD = ea.crtfld(Encoding, varTypes, ranges, borders, precisions, codes,
scales) # 调用函数创建区域描述器
Lind = int(np.sum(FieldD[0, :])) # 计算编码后的染色体长度
obj_trace = np.zeros((MAXGEN, 2)) # 定义目标函数值记录器
var_trace = np.zeros((MAXGEN, Lind)) # 定义染色体记录器，记录每一代最优个体的染色体
"""=========================开始遗传算法进化========================"""
start_time = time.time() # 开始计时
Chrom = ea.crtbp(NIND, Lind) # 生成种群染色体矩阵
variable = ea.bs2real(Chrom, FieldD) # 对初始种群进行解码
ObjV = aim(variable) # 计算初始种群个体的目标函数值
best_ind = np.argmax(ObjV) # 计算当代最优个体的序号
# 开始进化
for gen in range(MAXGEN):
FitnV = ea.ranking(-ObjV) # 根据目标函数大小分配适应度值(由于遵循目标最小化约定，因此最大化问题要对目标函数值乘上-1)
SelCh = Chrom[ea.selecting('rws', FitnV, NIND - 1), :] # 选择，采用'rws'轮盘赌选择
SelCh = ea.recombin('xovsp', SelCh, 0.7) # 重组(采用两点交叉方式，交叉概率为0.7)
SelCh = ea.mutbin(Encoding, SelCh) # 二进制种群变异
'font.serif': 'Times New Roman',
# 'figure.dpi': 300,
# 'savefig.dpi': 300,
# 'font.style':'normal', # ‘normal’, ‘italic’ or ‘oblique’.
# 'font.weight':'normal', #or 'blod'
'font.size': 12, #or large,small
}
rcParams.update(params)
"""================================实例化问题对象============================"""
problem = MyProblem() # 生成问题对象
"""==================================种群设置==============================="""
NIND = 60 # 种群规模
# 创建区域描述器，这里需要创建两个，前25个变量用P编码，剩余变量用RI编码
Encodings = ['P', 'RI']
Field1 = ea.crtfld(Encodings[0], problem.varTypes[:problem.num],
problem.ranges[:, :problem.num],
problem.borders[:, :problem.num])
Field2 = ea.crtfld(Encodings[1], problem.varTypes[problem.num:],
problem.ranges[:, problem.num:],
problem.borders[:, problem.num:])
Fields = [Field1, Field2]
population = ea.PsyPopulation(Encodings, Fields,
NIND) # 实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
# prophetPop = ea.PsyPopulation(Encodings, Fields, 2) # 实例化先知种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
# from MyFunction import Chroms_pro
# prophetPop.initChrom( NIND = 2)
# prophetPop.Chroms=Chroms_pro
"""================================算法参数设置============================="""
myAlgorithm = My_soea_psy_EGA_templet(problem, population) # 实例化一个算法模板对象
myAlgorithm.MAXGEN = 400 # 最大进化代数
本案例通过降维的方法，将等式约束化成了不等式约束，大大拓宽了可行解的空间，方便遗传算法求解
此外，本案例展示了利用多种群竞争的进化算法模板sga_mpc_real_templet了解决该问题。
"""
import numpy as np
import geatpy as ga
# 获取函数接口地址
AIM_M = __import__('aimfuc')
PUN_M = __import__('punishing')
# 变量设置
x1 = [0, 1] # 自变量1的范围
x2 = [0, 1] # 自变量2的范围
b1 = [1, 1] # 自变量1是否包含下界
b2 = [1, 1] # 自变量2是否包含上界
ranges = np.vstack([x1, x2]).T # 生成自变量的范围矩阵
borders = np.vstack([b1, b2]).T # 生成自变量的边界矩阵
precisions = [
2
] * 2 # 在二进制/格雷码编码中代表自变量的编码精度，当控制变量是连续型时，根据crtfld参考资料，该变量只表示边界精度，故设置为一定的正数即可
newRanges = ga.meshrng(ranges, gridnum=2) # 对控制变量范围进行网格化，网格边长为2
# 生成网格化后的区域描述器集合
FieldDRs = []
for i in range(len(newRanges)):
FieldDRs.append(ga.crtfld(newRanges[i], borders, precisions))
# 调用编程模板
[pop_trace, var_trace, times] = ga.sga_mpc_real_templet(AIM_M, 'aimfuc', PUN_M,\
'punishing', FieldDRs, problem = 'R', maxormin = -1, MAXGEN = 50, NIND = 50,\
SUBPOP = 1, GGAP = 0.9, selectStyle = 'tour', recombinStyle = 'xovdprs',\
recopt = 0.9, pm = 0.3, distribute = True, drawing = 1)
def each_brake(each_wait_list, L, W, brake_num, wait_list_num):
wait_list = each_wait_list
"""===============================实例化问题对象==========================="""
problem = MyProblem(wait_list, L, W) # 生成问题对象
"""=================================种群设置==============================="""
Encoding = 'P' # 编码方式
NIND = 6000 #4000,6000 种群规模
# ranges还是原来的，Field会在最后一行加上1
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders) # 创建区域描述器
population = ea.Population(Encoding, Field,
NIND) # 实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
"""===============================算法参数设置============================="""
myAlgorithm = ea.soea_SEGA_templet(problem,
population) # 实例化一个算法模板对象，单目标模板
# myAlgorithm=ea.moea_NSGA2_templet(problem, population) #多目模板
myAlgorithm.MAXGEN = 3 # 13 # 最大进化代数
# myAlgorithm.recOper = ea.Xovox(XOVR=0.8) # 设置交叉算子 __init__(self, XOVR=0.7, Half=False)
# myAlgorithm.mutOper = ea.Mutinv(Pm=0.2) # 设置变异算子
myAlgorithm.logTras = 1 # 设置每多少代记录日志，若设置成0则表示不记录日志
myAlgorithm.verbose = False # 设置是否打印输出日志信息
myAlgorithm.drawing = 0 # 设置绘图方式（0：不绘图；1：绘制结果图；2：绘制目标空间过程动画；3：绘制决策空间过程动画）
"""==========================调用算法模板进行种群进化======================="""
[population, obj_trace, var_trace] = myAlgorithm.run() # 执行算法模板
# population.save() # 把最后一代种群的信息保存到文件中
# 输出结果
best_gen = np.argmin(problem.maxormins * obj_trace[:, 1]) # 记录最优种群个体是在哪一代
best_ObjV = obj_trace[best_gen, 1]
# print('最优的目标函数值为：%s' % (best_ObjV))
# print('最优的控制变量值为：')
for i in range(var_trace.shape[1]): # (MAXGEN,Dim),进化的总代数和决策变量的维度
print(var_trace[best_gen, i])
best_sort_sequence = [int(each)
for each in var_trace[best_gen]] # (4000, 12)
#best_brake_seq={i:wait_list[i] for i in best_sort_sequence}
best_brake_seq = wait_list[best_sort_sequence]
#闸次原始顺序
brake_boat = quick_sort_brake(best_brake_seq, L, W)
# #将快速入闸的顺序，对应到最优选择的顺序
# brake_boat={best_sort_sequence[k]:v for k,v in brake_boat.items()}
#
# #将最优选择的顺序，对应到最原始的队列中的序号
# brake_boat={wait_list_num[k]:v for k,v in brake_boat.items()}
# N_e = len(brake_boat)
# plot_example(X, Y, li_e, wi_e, N_e,brake_num=brake_num)
best_use_rate = plot_save(brake_boat, brake_num)
print('plot finished')
# print('有效进化代数：%s' % (obj_trace.shape[0]))
# print('最优的一代是第 %s 代' % (best_gen + 1))
# print('评价次数：%s' % (myAlgorithm.evalsNum))
# print('时间已过 %s 秒' % (myAlgorithm.passTime))
return {
'brake_boat': brake_boat,
'best_sort_sequence': best_sort_sequence,
'best_use_rate': best_use_rate
}
def run_wind_turbine(problem):
"""
optimize the wind turbines solely
"""
# settings for genetic algorithm, most parameters can be specified through the .ini file
Encoding = "RI" # encoding of the individuals, RI means use real numbers to encode
NIND = int(
problem.settings["global"]["num_individual"]) # number of individuals
Field = ga.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders)
population = ga.Population(Encoding, Field,
NIND) # generate the population instance
myAlgo = ga.soea_DE_best_1_L_templet(problem,
population) # initialize the problem
myAlgo.MAXGEN = int(
problem.settings["global"]["max_generation"]) # number of generations
myAlgo.mutOper.F = float(
problem.settings["global"]["mut_factor"]) # mutation factor
myAlgo.mutOper.Pm = float(
problem.settings["global"]["mut_prob"]) # mutation probability
myAlgo.recOper.XOVR = float(
problem.settings["global"]["cor_factor"]) # crossover factor
myAlgo.drawing = 0 # 0: no drawing; 1: drawing at the end of process; 2: animated drawing
myAlgo.logTras = int(
problem.settings["global"]["log_trace"]) # record log every * steps
myAlgo.verbose = bool(int(
problem.settings["global"]["log_by_print"])) # whether print log
# display running information for debug
print(TIP + """
Genetic algorithm will start with:
Number of individuals: {},
Number of generations: {},
Mutation factor: {},
Mutation Probability: {},
Crossover factor: {}
""".format(NIND, myAlgo.MAXGEN, myAlgo.mutOper.F, myAlgo.mutOper.Pm,
myAlgo.recOper.XOVR) + RESET)
# record computational time
start_t = time.time()
# run genetic algorithm
[best_ind, population] = myAlgo.run()
# delete progress bar in problem
problem.pbar.close()
# time interval
end_t = time.time()
interval_t = end_t - start_t
# judge whether genetic algorithm failed
if len(myAlgo.log["gen"]) == 0:
print(ERROR + "Genetic algorithm failed!!!" + RESET)
print(
TIP +
"You can re-execute these codes by amplifying your target wave farm."
+ RESET)
exit(ga_failed)
# otherwise, genetic algorithm succeed
# output: optimization results of wind turbines
best_ind.save(os.path.join(problem.settings["proj_name"], "record_array"))
file = open(
os.path.join(problem.settings["proj_name"], "record_array",
"record.txt"), "w")
file.write("# Used time: {} s\n".format(interval_t))
file.write("# average_objv, maximum_objv\n")
if myAlgo.log:
file.write("\n".join([
"{:.6f}, {:.6f}".format(myAlgo.log["f_avg"][gen],
myAlgo.log["f_max"][gen])
for gen in range(len(myAlgo.log["gen"]))
]))
file.close()
# output: plot the objv curves
fig = plt.figure(figsize=(10, 10))
plt.plot(myAlgo.log["f_max"], label="max")
plt.plot(myAlgo.log["f_avg"], label="avg")
plt.legend()
plt.savefig(
os.path.join(problem.settings["proj_name"], "record_array",
"objvs.png"))
# output: plot the wind turbine array figure
fig = plt.figure(figsize=(16, 8))
file = open(
os.path.join(problem.settings["proj_name"], "record_array",
"Phen.csv"))
# read out the best individual
var_trace = [float(item) for item in file.readline().split(",")]
# plot the best individual in tow sub plots
for sub in range(1, 3):
plt.subplot(1, 2, sub)
for i in range(len(var_trace) // 2):
# plt.scatter([var_trace[i]], [var_trace[i + len(var_trace) // 2]], s=100, c="k", zorder=2)
plt.plot([var_trace[i], var_trace[i]], [
var_trace[i + len(var_trace) // 2] -
float(problem.settings["wind_turbine"]["rotor_diameter"]),
var_trace[i + len(var_trace) // 2] +
float(problem.settings["wind_turbine"]["rotor_diameter"])
],
"k-",
linewidth=5,
zorder=2)
# tune the sub plots limits
plt.xlim((0, problem.ub[0]))
plt.ylim((0, problem.ub[-1]))
# compute flow fields of the best individual
x_array, y_array, wind_vels, turb_ints = problem.plot_field(var_trace)
# plot wind velocity and turbulence intensity
plt.subplot(1, 2, 1)
plt.contourf(x_array, y_array, wind_vels, zorder=1, cmap="bwr")
plt.subplot(1, 2, 2)
plt.contourf(x_array, y_array, turb_ints, zorder=1, cmap="bwr")
# output flow fields into a file
x_array = x_array.flatten()
y_array = y_array.flatten()
wind_vels = wind_vels.flatten()
turb_ints = turb_ints.flatten()
file = open(
os.path.join(problem.settings["proj_name"], "record_array",
"flow_fields.txt"), "w")
file.write("# x, y, wind_velocity, turb_intensity\n")
for i in range(x_array.size):
file.write("{}, {}, {}, {}\n".format(x_array[i], y_array[i],
wind_vels[i], turb_ints[i]))
file.close()
plt.savefig(
os.path.join(problem.settings["proj_name"], "record_array",
"array.png"))
def main_1(filepath_1):
filepath_1 = 'input_1.json'
input_1 = _get_json_(filepath_1)
filepath_2 = input_1['filePath'] #读入csv路径
data = pd.read_csv(filepath_2)
# 必须添加header=None，否则默认把第一行数据处理成列名导致缺失
data_1 = pd.DataFrame(data)
num = len(data_1) # 病人总的数目
morning_time = int(input_1['startTime'].split(':')[0]) * 60 + int(
input_1['startTime'].split(':')[1])
afternoon_time = int(input_1['endTime'].split(':')[0]) * 60 + int(
input_1['endTime'].split(':')[1])
n_x = int(input_1['orNum']) #手术室数目
n_y = int(input_1['recoverNum']) #复苏室数目
t_s = int(input_1['minRecoverTime']) #最小复苏时间
calculte_r = calculte(data, n_x, n_y, t_s, morning_time,
afternoon_time) # 实例化
list_doctID, list_patientID, list_operation, list_sleepy, list_clean, list_start, list_index_or, list_of_all = calculte_r._process_data_(
num)
# print(list_doctID, list_sleepy, list_operation, list_clean, list_patientID, list_start, list_index_or)
Encoding = 'RI' # 编码方式为实数
conordis = 1 # 染色体解码后得到的变量是离散的
NIND = 50 # 种群规模
problem = MyProblem(num, n_x, n_y, NIND, list_of_all, morning_time,
afternoon_time)
# n_o为手术室数量，n_r为复苏室数量, chrom为染色体[1,3,2]表示第一台
# 手术在1号手术室在1号手术室内做，o_time[30,100,60]表示第一台手术时长30分钟，
# c_time表示清洁时长，r_time表示复苏时长（0或自定义最小复苏时长默认为60min）
# 生成问题对象
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders) # 创建区域描述器
population = ea.Population(Encoding, Field, NIND) # 创建种群对象
x_chuandai = 20
id_trace = (np.zeros(
(x_chuandai, num)) * np.nan) # 定义变量记录器，记录决策变量值，初始值为nan
Algorithm_1 = Algorithm(problem, population, id_trace) # 实例化一个算法模板对象
Algorithm_1.MAXGEN = x_chuandai # 最大遗传代数
[population, obj_trace, var_trace,
id_trace] = Algorithm_1.run() # 执行算法模板 #如何返回最优解
best_gen = np.argmin(obj_trace[:, 1]) # 记录最优种群是在哪一代
best_ObjV = obj_trace[best_gen, 1] # 目标函数值最优
#mean_ObjV = obj_trace[best_gen, 0] # 均值
best_paixu = var_trace[best_gen, :] # 最优解
ARRAY = id_trace[best_gen, :]
print('最优的目标函数值为：%s' % (best_ObjV))
print('有效进化代数：%s' % (obj_trace.shape[0]))
print('最优的一代是第 %s 代' % (best_gen + 1))
print('时间已过 %s 秒' % (Algorithm_1.passTime))
# id_trace = (np.zeros((self.MAXGEN, NVAR)) * np.nan) # 定义变量记录器，记录决策变量值，初始值为nan
#返回一个ARRAY将list调整并传给best_result函数
sel_data = list_of_all[:, list(ARRAY.astype(np.int))]
#list_doctID, list_patientID, list_operation, list_sleepy, list_clean, list_start, list_index_or
list_doctID_2 = sel_data[0]
list_patientID_2 = sel_data[1]
list_operation_2 = sel_data[2]
list_sleepy_2 = sel_data[3]
list_clean_2 = sel_data[4]
list_start_2 = sel_data[5]
list_index_or_2 = sel_data[6]
ARRAY_1 = ARRAY.astype(np.int)
best_paixu_1 = best_paixu.astype(np.int)
print('最优解的index:', ARRAY_1)
print('最优解的手术室号:', best_paixu_1)
o_total_time, o_total_r_time, o_total_empty_time, overtime_work, result = calculte_r._best_result_(
best_paixu_1, num, list_sleepy_2, list_operation_2, list_clean_2)
# o_total_time是手术室工作的总时长// o_total_r_time是手术室总复苏时长// o_total_empty_time是手术室总空闲时长// overtime_work是手术室总超时工作时长//
index_or_1, list_clean_1, list_sleepy_1, list_start_1 = calculte_r._get_list_(
num, result, ARRAY_1, list_clean, list_operation)
#用于返回数据
data['复苏时间'] = list_sleepy_1 # 手术室内的复苏时间
data['清洁时间'] = list_clean_1 # 手术室内的清洁时间
data['手术开始时间'] = list_start_1 # 手术室每一台手术的开始时间
data['手术室编码'] = index_or_1 # 手术室内的手术的编码
#存入csv
data.to_csv('output.csv', sep=',', header=True)
#json输出内容
orRatio = str((o_total_time - o_total_r_time - list_clean_1.sum()) /
(o_total_time + o_total_empty_time))
cleanRatio = str(list_clean_1.sum() / (o_total_time + o_total_empty_time))
recoverRoomratio = str(o_total_r_time /
(o_total_time + o_total_empty_time))
emptyRatio = str(o_total_empty_time / (o_total_time + o_total_empty_time))
extraHours = overtime_work
extraHoursRatio = (extraHours / o_total_time)
overtimeRatio = str(overtime_work.sum() / o_total_time)
dict_2 = {}
dict_2["filePath"] = "output.csv"
dict_2["orRatio"] = orRatio # 手术室利用率
dict_2["recoverRatio"] = recoverRoomratio # 复苏室利用率
dict_2["cleanRatio"] = cleanRatio # 用于清洁的时间
dict_2["emptyRatio"] = emptyRatio # 闲置时间比例
dict_2["extraHours"] = extraHours.tolist() # 加班总时间(分钟)
dict_2["extraHoursRatio"] = extraHoursRatio.tolist() # 每一个元素
dict_2["overtimeRatio"] = overtimeRatio # 额外加班时间
# 用于饼图展示的比例：
return dict_2
import numpy as np
import geatpy as ea
from hsdn_near_optimal_performance import SOHybridNetTEOptimizeProblem
max_val = float('inf')
if __name__ == "__main__":
# 初始化连接图
graph = np.array([[0, 1, 1, max_val], [1, 0, 1, 1], [1, 1, 0, 1], [max_val, 1, 1, 0]])
band_width = np.array([[0.0, 100.0, 100.0, 0.0], [0.0, 0.0, 100.0, 100.0], [0.0, 0.0, 0.0, 100.0],
[0.0, 0.0, 0.0, 0.0]])
traffic = np.array([[0, 10, 15, 10], [10, 0, 9, 8], [10, 10, 0, 20], [10, 10, 10, 0]])
sdn_node_count = 1
problem = SOHybridNetTEOptimizeProblem(graph, sdn_node_count, traffic, band_width)
Encodings = ['P', 'RI']
Field1 = ea.crtfld(Encodings[0], problem.varTypes[:sdn_node_count],
problem.ranges[:, :sdn_node_count], problem.borders[:, :sdn_node_count]) # 创建区域描述器
Field2 = ea.crtfld(Encodings[1], problem.varTypes[sdn_node_count:],
problem.ranges[:, sdn_node_count:], problem.borders[:, sdn_node_count:]) # 创建区域描述器
Fields = [Field1, Field2]
# 种群规模
NIND = 50
population = ea.PsyPopulation(Encodings, Fields, NIND)
myAlgorithm = ea.moea_psy_NSGA3_templet(problem, population)
myAlgorithm.MAXGEN = 20
myAlgorithm.drawing = 2
NDSet = myAlgorithm.run()
NDSet.save() # 把结果保存到文件中
# 输出
print('用时：%s 秒' % myAlgorithm.passTime)
print('非支配个体数：%s 个' % NDSet.sizes)
def Scheduler(input_json, input_config):
"""
调度器
用法：
from MultiScheduler import Scheduler
output_json, output_overall = Scheduler(input_json, input_config)
输入：
input_json:患者信息
input_json = [
{
"age": "70",
"anaesthetic": "全身麻醉",
"department": "心血管科",
"doctorName": "李四",
"gender": "男",
"id": "1",
"key": "0",
"name": "王小舟",
"operatingName": "心脏搭桥手术",
"orId": "",
"predTime": "120",
"rank": "2",
"startTime": ""
},{
"age": "23",
"anaesthetic": "全身麻醉",
"department": "妇产科",
"doctorName": "王五",
"gender": "女",
"id": "1",
"key": "0",
"name": "张小臻",
"operatingName": "剖腹产",
"orId": "",
"predTime": "120",
"rank": "2",
"startTime": ""
}
]
input_config:环境变量，手术室数量、复苏室数量等信息
input_config = {
"recover_min": 60, #最小恢复时间
"end_time": "3:30", #下班时间
"operRoom": 3, #手术室数目
"recover": 3, #恢复室数目
"start_time": "08:30" #上班时间
}
输出：
output_json:添加手术室号、开始时间等的病人信息
output_json = [
{
"key": "0",
"id": "1",
"name": "尹小帆",
"gender": "男",
"age": "70",
"department": "心血管科",
"operatingName": "心脏搭桥手术",
"doctorName": "李四",
"predTime": "120",
"anaesthetic": "全身麻醉",
"rank": "2",
"orId": "10",
"startTime": "8:00",
"recoverDuration": 15,
"cleanDuration": 20
}, {
"key": "1",
"id": "2",
"name": "司徒",
"gender": "女",
"age": "23",
"department": "妇产科",
"operatingName": "剖腹产手术",
"doctorName": "王小二",
"predTime": "100",
"anaesthetic": "局部麻醉",
"rank": "1",
"orId": "7",
"startTime": "9:00",
"recoverDuration": 15,
"cleanDuration": 20
}
]
output_overall:手术室利用比例，恢复室利用比例、各手术室超出工作时间和比例
output_overall = {
"orRatio": "0.99999",
"recoverRoomRatio": "0.8",
"extraHours": [4, 5, 6],
"extraHourRatio": [0.4, 0.2, 0.9]
}
"""
#===================以下为程序部分===================#
if (not input_json) or (not input_config): #判断空字符串输入，返回空字符串
return '', ''
data = pd.DataFrame(input_json)
input_1 = input_config
num = len(data) # 病人总的数目
morning_time = int(input_1['start_time'].split(':')[0]) * 60 + int(
input_1['start_time'].split(':')[1])
afternoon_time = int(input_1['end_time'].split(':')[0]) * 60 + int(
input_1['end_time'].split(':')[1])
n_x = int(input_1['operRoom']) #手术室数目
n_y = int(input_1['recover']) #复苏室数目
t_s = int(input_1['doctor']) #最小复苏时间
calculte_r = calculte(data, n_x, n_y, t_s, morning_time,
afternoon_time) # 实例化
list_doctID, list_patientID, list_operation, list_sleepy, list_clean, list_start, list_index_or, list_of_all = calculte_r._process_data_(
num)
list_of_all_1 = list_of_all[:] # 复制list_of_all为list_of_all_1
list_start_temple = list_start
list_index_or_temple = list_index_or
list_start_3 = np.zeros((num, ), dtype=np.int)
list_index_or_3 = np.zeros((num, 1), dtype=np.int)
list_clean = list_clean.astype(np.int)
# print(list_clean)
temple_for_startime = np.where(list_start_temple != '')[0]
for value in temple_for_startime:
neirong = (int((list_start[value].split(':')[0]))) * 60 + int(
(list_start[value].split(':')[1])) - morning_time
list_start_3[value] = neirong # list_start_3中存入的是有开始时间病人的开始时间
temple_for_orNum = np.where(list_index_or_temple != '')[0]
index_index_or_temple_1 = np.array(
[], dtype=np.int) # list_start_3中存储的是转换后的list_start值(分钟)
for value in temple_for_orNum:
list_index_or_3[value] = list_index_or[
value] # list_index_or_3中存入的是有手术室号病人的手术室号
# print(list_index_or_3[value][0])
set_jiao = set(temple_for_startime) & set(temple_for_orNum)
list_cha_start_time = list(set(temple_for_startime) - set_jiao)
list_cha_index_or = list(set(temple_for_orNum) - set_jiao)
list_jiao = list(set_jiao)
list_cha_index_or_1 = list_cha_index_or[:] # 复制仅绑定手术室号的值
dict_for_xunhuan = {} # 存入需要找到并确定的dict的值
# 对仅绑定手术室号的情况进行序列调整, 这一列需要被传递
for index, value in enumerate(list_cha_index_or):
a = 0
for j in temple_for_startime:
if value > j:
a -= 1
list_cha_index_or_1[index] = value + a
dict_for_xunhuan[value + a] = [list_index_or_3[value][0], value]
# dict_for_xunhuan中存入的是每一个人对应的手术室号0
# 和它实际的位
# ARRAY对应的字典
dict_for_result = {} # 存入需要找到并确定的dict的值
for i in range(num):
a = 0
for j in temple_for_startime:
if i > j:
a -= 1
dict_for_result[i] = [i + a]
for k in temple_for_startime:
del dict_for_result[k]
length_sum = len(temple_for_startime) # 绑定了时间的人数
Encoding = 'RI' # 编码方式为实数 # 染色体解码后得到的变量是离散的
NIND = 50 # 种群规模
problem = MyProblem(num - length_sum, n_x, n_y, NIND, list_of_all_1,
morning_time, afternoon_time, dict_for_xunhuan)
# n_o为手术室数量，n_r为复苏室数量, chrom为染色体[1,3,2]表示第一台
# 手术在1号手术室在1号手术室内做，o_time[30,100,60]表示第一台手术时长30分钟，
# c_time表示清洁时长，r_time表示复苏时长（0或自定义最小复苏时长默认为60min）
# 生成问题对象
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders) # 创建区域描述器
population = ea.Population(Encoding, Field, NIND) # 创建种群对象
x_chuandai = 30
id_trace = (np.zeros(
(x_chuandai, num - length_sum)) * np.nan) # 定义变量记录器，记录决策变量值，初始值为nan
# problem, population, id_trace, n_x, list_jiao, list_start_3, list_operation, list_sleepy, list_index_or_3, list_doctID, list_clean, list_cha_start_time
Algorithm_1 = Algorithm(problem, population, id_trace, n_x, list_jiao,
list_start_3, list_operation, list_sleepy,
list_index_or_3, list_doctID, list_clean,
list_cha_start_time) # 实例化一个算法模板对象
Algorithm_1.MAXGEN = x_chuandai # 最大遗传代数
[population, obj_trace, var_trace, id_trace,
new_fixed_dict] = Algorithm_1.run() # 执行算法模板 #如何返回最优解
best_gen = np.argmin(obj_trace[:, 1]) # 记录最优种群是在哪一代
best_ObjV = obj_trace[best_gen, 1] # 目标函数值最优
# mean_ObjV = obj_trace[best_gen, 0] # 均值
best_paixu = var_trace[best_gen, :] # 最优解
ARRAY = id_trace[best_gen, :]
sel_data = list_of_all_1[:, list(ARRAY.astype(np.int))]
# list_doctID, list_patientID, list_operation, list_sleepy, list_clean, list_start, list_index_or
list_doctID_2 = sel_data[0]
list_patientID_2 = sel_data[1]
list_operation_2 = sel_data[2]
list_sleepy_2 = sel_data[3]
list_clean_2 = sel_data[4]
list_start_2 = sel_data[5]
list_index_or_2 = sel_data[6]
ARRAY_1 = ARRAY.astype(np.int)
best_paixu_1 = best_paixu.astype(np.int)
o_total_time, o_total_r_time, o_total_empty_time, overtime_work, result, fixed_result, o_c_time = calculte_r._best_result_(
best_paixu_1, new_fixed_dict, list_sleepy_2, list_operation_2,
list_clean_2, list_doctID_2)
# o_total_time是手术室工作的总时长// o_total_r_time是手术室总复苏时长// o_total_empty_time是手术室总空闲时长// overtime_work是手术室总超时工作时长//
index_or_1, list_clean_1, list_sleepy_1, list_start_1 = calculte_r._get_list_(
num, result, fixed_result, ARRAY_1, list_clean, list_operation,
dict_for_xunhuan, list_start, dict_for_result)
data['recoverDuration'] = list_sleepy_1 # 手术室内的复苏时间
data['cleanDuration'] = list_clean_1 # 手术室内的清洁时间
data['startTime'] = list_start_1 # 手术室每一台手术的开始时间
data['orId'] = index_or_1 # 手术室内的手术的编码
data['predTime'] = list_operation
#存入csv
#json输出内容
orRatio = str(
(o_total_time.sum() - o_total_r_time.sum() - list_clean_1.sum()) /
(o_total_time.sum() + o_total_empty_time.sum()))
everyorRatio = (o_total_time - o_total_r_time -
o_c_time) / (o_total_time + o_total_empty_time)
cleanRatio = str(list_clean_1.sum() /
(o_total_time.sum() + o_total_empty_time.sum()))
recoverRoomratio = str(o_total_r_time.sum() /
(o_total_time.sum() + o_total_empty_time.sum()))
emptyRatio = o_total_empty_time / (o_total_time + o_total_empty_time)
extraHours = overtime_work
extraHoursRatio = (extraHours / o_total_time)
overtimeRatio = str(overtime_work.sum() / o_total_time.sum())
# dict_2 = {}
# dict_2["filePath"] = "output.csv"
# dict_2["orRatio"] = orRatio # 手术室利用率
# dict_2["recoverRatio"] = recoverRoomratio # 复苏室利用率
# dict_2["cleanRatio"] = cleanRatio # 用于清洁的时间
# dict_2["emptyRatio"] = emptyRatio # 闲置时间比例
# dict_2["extraHours"] = extraHours.tolist() # 加班总时间(分钟)
# dict_2["extraHoursRatio"] = extraHoursRatio.tolist() # 每一个元素
# dict_2["overtimeRatio"] = overtimeRatio # 额外加班时间
# 用于饼图展示的比例：
data.to_csv('output.csv', sep=',', header=True)
output_json = data.to_json(orient='records', force_ascii=False)
output_overall = {
"orRatio": orRatio,
"everyorRatio": everyorRatio.tolist(),
"recoverRoomRatio": recoverRoomratio,
"emptyRatio": emptyRatio.tolist(),
"cleanRatio": cleanRatio,
"extraHours": extraHours.tolist(),
"extraHourRatio": extraHoursRatio.tolist(),
"overtimeRatio": overtimeRatio
}
return output_json, output_overall
x3 - x1 <= 0
x4 - x2 <= 0
x1,x2,x3,x4均为0或1
本案例调用了“sga_new_real_templet”这个算法模板，其详细用法可利用help命令查看，或是在github下载并查看源码
调用算法模板时可以设置drawing=2，此时算法模板将在种群进化过程中绘制动画，但注意执行前要在Python控制台执行命令matplotlib qt5。
"""
import numpy as np
import geatpy as ga
# 获取函数接口地址
AIM_M = __import__('aimfuc')
# 变量设置
ranges = np.vstack([np.zeros((1, 4)), np.ones((1, 4))]) # 生成自变量的范围矩阵
borders = np.vstack([np.ones((1, 4)), np.ones((1, 4))]) # 生成自变量的边界矩阵
FieldDR = ga.crtfld(ranges, borders) # 生成区域描述器
# 调用编程模板
[pop_trace, var_trace, times] = ga.sga_new_real_templet(AIM_M,
'aimfuc',
None,
None,
FieldDR,
problem='I',
maxormin=-1,
MAXGEN=50,
NIND=10,
SUBPOP=1,
GGAP=0.9,
selectStyle='sus',
recombinStyle='xovdp',
recopt=0.9,
Use GA algorithm to search the sentence: Tom is a little boy.
Use 'help' go get more tutorials about the algorithm templet 'sga_new_real_templet' we use there,
or read the source codes in github.
You can set 'drawing = 2' to get a dynamic graph. (But remember to run the command 'matplotlib qt5' in Python console before.)
"""
import numpy as np
import geatpy as ga
strs = 'Tom is a little boy.' # define the sentence
words = []
for c in strs:
words.append(ord(c)) # change the words to ascii code.
# Attention: you had better put the aim-function in another file.
def aim(Phen, LegV): # define the aim function
real = words
diff = np.sum((Phen - real)**2, 1)
return [np.array([diff]).T, LegV]
if __name__ == "__main__":
AIM_M = __import__('main') # get the handle of aim-function
# variables setting
ranges = np.vstack([32 * np.ones((1, len(words))), 122 * np.ones((1, len(words)))]) # ranges of variables
borders = np.vstack([np.ones((1, len(words))), 122 * np.ones((1, len(words)))]) # borders of variables
FieldDR = ga.crtfld(ranges, borders) # create FieldDR
# call the GEA algorithm template
[pop_trace, var_trace, times] = ga.sga_new_real_templet(AIM_M, 'aim', None, None, FieldDR, problem = 'I', maxormin = 1, MAXGEN = 2000, NIND = 50, SUBPOP = 1, GGAP = 0.9, selectStyle = 'etour', recombinStyle = 'xovdprs', recopt = 0.9, pm = None, distribute = True, drawing = 1)
# output results
for num in var_trace[np.argmin(pop_trace[:, 1]), :]:
print(chr(int(num)), end = '')
import numpy as np
import geatpy as ea
from MyProblem import MyProblem #导入自定义问题接口
'''=========== 实例化问题对象==========='''
steps = [3,1,4,2,3,2,2,2,2,1]
problem = MyProblem() #生成问题对象
'''=========== 种群设置==========='''
NIND = 40 #种群规模
# 创建区域描述器，这里需要创建两个，分别是工序编码 机器编码
Encodings = ['RI', 'RI']
# crtfld (生成译码矩阵，俗称“区域描述器”)
Field1 = ea.crtfld(Encodings[0], problem.varTypes[:sum(steps)], problem.ranges[:,:sum(steps)], problem.borders[:,:sum(steps)]) # 创建区域描述器
Field2 = ea.crtfld(Encodings[1], problem.varTypes[sum(steps):], problem.ranges[:,sum(steps):], problem.borders[:,sum(steps):])
Fields = [Field1, Field2]
#实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
population = ea.PsyPopulation(Encodings, Fields, NIND)
'''=================算法参数设置==================='''
myAlgorithm = ea.moea_psy_NSGA2_templet(problem, population) # 实例化一个算法模板对象
myAlgorithm.MAXGEN = 700 # 最大进化代数
myAlgorithm.drawing = 2 # 设置绘图方式（0：不绘图；1：绘制结果图；2：绘制过程动画）
"""===========================调用算法模板进行种群进化======================"""
NDSet = myAlgorithm.run() # 执行算法模板，得到帕累托最优解集NDSet
NDSet.save() # 把结果保存到文件中
# 输出
print('用时：%s 秒'%(myAlgorithm.passTime))
print('非支配个体数：%s 个'%(NDSet.sizes))
print('单位时间找到帕累托前沿点个数：%s 个'%(int(NDSet.sizes // myAlgorithm.passTime)))
def searchMoea(net_path, save_path, upper_limit=None, lower_limit=None):
"""多目标优化搜索,输入为预训练模型的权重所在路径，输出为遗传算法搜索结果的.csv保存的目录路径"""
"""==============================环境参数设置和问题类的初始化==========================="""
# get options
opt = BaseOptions().parse()
if not (upper_limit is None):
opt.max_rate = upper_limit
if not (lower_limit is None):
opt.min_rate = lower_limit
# basic settings
os.environ["CUDA_VISIBLE_DEVICES"] = str(opt.gpu_ids)[1:-1]
if torch.cuda.is_available():
device = "cuda"
torch.backends.cudnn.benchmark = False
else:
device = "cpu"
##################### Get Dataloader ####################
dataloader_train, dataloader_val = custom_get_dataloaders(opt)
# dummy_input is sample input of dataloaders
if hasattr(dataloader_val, "dataset"):
dummy_input = dataloader_val.dataset.__getitem__(0)
dummy_input = dummy_input[0]
dummy_input = dummy_input.unsqueeze(0)
else:
# for imagenet dali loader
dummy_input = torch.rand(1, 3, 224, 224)
##################### Create Baseline Model ####################
net = ModelWrapper(opt)
# 预训练模型的权重导入
load(net, net_path)
flops_before, params_before = model_summary(net.get_compress_part(),
dummy_input)
compression_scheduler = distiller.file_config(net.get_compress_part(),
net.optimizer,
opt.compress_schedule_path)
num_layer = len(compression_scheduler.policies[1])
##################### 定义问题，并进行遗传算法 ####################
args = dict()
args["dataloader_train"] = dataloader_train
args["dataloader_val"] = dataloader_val
args["dummy_input"] = dummy_input
args["flops_before"] = flops_before
args["device"] = device
args["num_layer"] = num_layer
problem = MoeaProblem(opt, net_path, args)
"""==============================种群设置==========================="""
Encoding = 'RI' #编码方式,此处为整数实数混合编码
NIND = 10 #种群规模
Field = gp.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders) #创建区域描述器
population = gp.Population(Encoding, Field, NIND)
"""===========================算法参数设置=========================="""
#myAlgorithm = gp.moea_awGA_templet(problem, population)
#myAlgorithm = gp.moea_MOEAD_DE_templet(problem, population)
#myAlgorithm = gp.moea_RVEA_RES_templet(problem, population)
myAlgorithm = gp.moea_NSGA2_archive_templet(problem, population)
#实例化算法模板
myAlgorithm.MAXGEN = 100 # 最大进化代数
#myAlgorithm.mutOper.F = 0.5 # 差分进化中的参数F
#myAlgorithm.recOper.XOVR = 0.7 # 设置交叉概率
myAlgorithm.logTras = 10 # 设置每隔多少代记录日志，若设置成0则表示不记录日志
myAlgorithm.verbose = True # 设置是否打印输出日志信息
myAlgorithm.drawing = 1 # 设置绘图方式（0：不绘图；1：绘制结果图；2：绘制目标空间过程动画)
"""==========================根据先验知识创建先知种群==============="""
#data = pd.read_csv('/eagle_eye/search_results/old_dataset/pop_get_search_result_moea2/Phen.csv')
#well_config=np.array(data)[5:,:]
#prophetpop=gp.Population(Encoding,Field,5,well_config)# 实例化种群对象
#myAlgorithm.call_aimFunc(prophetpop)# 计算先知种群的目标函数值及约束（假如有约束）
#myAlgorithm.call_aimFunc()# 计算先知种群的目标函数值及约束（假如有约束）
"""==========================调用算法模板进行种群进化==============="""
#[BestIndi, population] = myAlgorithm.run(prophetpop) # 执行带先验种群的算法模板，得到最优个体以及最后一代种群
[BestIndi, population] = myAlgorithm.run() # 执行算法模板，得到最优个体以及最后一代种群
BestIndi.save(os.path.join(save_path, "BestIndi")) # 把最优个体的信息保存到文件中
population.save(os.path.join(save_path, "Population")) # save population
"""=================================输出结果======================="""
print('评价次数：%s' % myAlgorithm.evalsNum)
print('时间已过%s秒' % myAlgorithm.passTime)
if BestIndi.sizes != 0:
print('最优的目标函数值为： ')
print(BestIndi.ObjV[0, :])
print('最优的控制变量值为：')
print(BestIndi.Phen[0, :])
return BestIndi.ObjV[0, :], BestIndi.Phen[0, :]
else:
print('没找到可行解。')
return -1, -1
# -*- coding: utf-8 -*-
import numpy as np
import geatpy as ea
from MyProblem import MyProblem
import matplotlib.pyplot as plt
if __name__ == '__main__':
problem = MyProblem()
Encoding = 'RI' # Encoding
NIND = 20 # Population size
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges, problem.borders) # Create a region descriptor
population = ea.Population(Encoding, Field, NIND) # Instantiate population object
"""================================Algorithm parameter settings============================="""
myAlgorithm = ea.moea_NSGA3_DE_templet(problem, population) # Instantiate an algorithm template object
myAlgorithm.MAXGEN = 25 # Maximum number of evolutions
myAlgorithm.mutOper.F = 0.5 # The parameter F in differential evolution
myAlgorithm.recOper.XOVR = 0.7 # Reorganization Probability
myAlgorithm.drawing = 0
"""===========================Call algorithm template for population evolution======================="""
NDSet = myAlgorithm.run () #
NDSet.save () #
#drawing
print ( 'Seconds: %s' % (myAlgorithm.passTime) )
print ( 'Non-dominated individuals：%s ' % (NDSet.sizes) )
print ( 'Decision variable value：%s ' % (NDSet.Phen) )
print ( 'Decision target value：%s ' % (NDSet.ObjV) )
ax1 = plt.Circle ( (NDSet.Phen[-1,2], 2.5), NDSet.Phen[-1,0], fill=False )
ax2 = plt.Rectangle ( (1, 2), 18, 1, fill=False )
ax3 = plt.Circle ( (NDSet.Phen[-1,3], 13.5), NDSet.Phen[-1,1], fill=False )
def awGA_templet(AIM_M,
AIM_F,
PUN_M,
PUN_F,
ranges,
borders,
precisions,
maxormin,
MAXGEN,
MAXSIZE,
NIND,
SUBPOP,
GGAP,
selectStyle,
recombinStyle,
recopt,
pm,
drawing=1):
"""
awGA_templet.py - 基于awGA的多目标优化编程模板
语法：
该函数除了参数drawing外，不设置可缺省参数。
当某个参数需要缺省时，在调用函数时传入None即可。
比如当没有罚函数时，则在调用编程模板时将第3、4个参数设置为None即可，如：
awGA_templet(AIM_M, 'aimfuc', None, None, ..., maxormin)
输入参数：
AIM_M - 目标函数的地址，传入该函数前通常由AIM_M = __import__('目标函数名')语句得到
AIM_F : str - 目标函数名
PUN_M - 罚函数的地址，传入该函数前通常由PUN_M = __import__('罚函数名')语句得到
PUN_F : str - 罚函数名
ranges : array - 代表自变量的范围矩阵，要求上界必须大于下界
例如：[[1, 2, 3],
[3, 4, 5]]
表示有3个控制变量，其范围分别是1-3, 2-4, 3-5
borders : list -（可选参数）代表是否包含变量范围的边界，为1代表控制变量的范围包含该边界
当为None时，默认设置为全是1的矩阵
例如：[[1, 0, 1],
[0, 1, 1]]
表示上面的三个控制变量的范围分别是：[1, 3)、(2, 4]、[3, 5]
precisions : list -（可选参数）代表控制变量的精度，
如等于4，表示对应的控制变量的编码可以精确到小数点后4位。
当precisions为None时，默认precision为1*n的0矩阵(此时表示种群是离散编码的)
precision的元素必须不小于0
maxormin int - 最小最大化标记，1表示目标函数最小化；-1表示目标函数最大化
MAXGEN : int - 最大遗传代数
MAXSIZE : int - 帕累托最优集最大规模
NIND : int - 种群规模，即种群中包含多少个个体
SUBPOP : int - 子种群数量，即对一个种群划分多少个子种群
GGAP : float - 代沟，表示子代与父代染色体及性状不相同的概率
selectStyle : str - 指代所采用的低级选择算子的名称，如'rws'(轮盘赌选择算子)
recombinStyle: str - 指代所采用的低级重组算子的名称，如'xovsp'(单点交叉)
recopt : float - 交叉概率
pm : float - 重组概率
drawing : int - (可选参数)，0表示不绘图，1表示绘制最终结果图，2表示绘制进化过程的动画。
默认drawing为1
算法描述:
本模板实现了基于适应性权重聚合法(awGA)的多目标优化搜索，
通过维护一个全局帕累托最优集来实现帕累托前沿的搜索，故并不需要保证种群所有个体都是非支配的
"""
#==========================初始化配置===========================
# 获取目标函数和罚函数
aimfuc = getattr(AIM_M, AIM_F) # 获得目标函数
FieldDR = ga.crtfld(ranges, borders, precisions)
#=========================开始遗传算法进化=======================
Chrom = ga.crtrp(NIND, FieldDR) # 创建简单离散种群
ObjV = aimfuc(Chrom) # 计算种群目标函数值
NDSet = np.zeros((0, Chrom.shape[1])) # 定义帕累托最优解记录器
NDSetObjV = np.zeros((0, ObjV.shape[1])) # 定义帕累托最优解的目标函数值记录器
ax = None
start_time = time.time() # 开始计时
# 开始进化！！
for gen in range(MAXGEN):
if NDSet.shape[0] > MAXSIZE:
break
[CombinObjV, weight] = ga.awGA(maxormin * ObjV) # 计算适应性权重以及多目标的加权单目标
FitnV = ga.ranking(maxormin * CombinObjV) # 根据加权单目标计算适应度
# 更新帕累托最优集以及种群非支配个体的适应度
[FitnV, NDSet, NDSetObjV,
repnum] = ga.upNDSet(Chrom, maxormin * ObjV, FitnV, NDSet,
maxormin * NDSetObjV)
# 进行遗传操作！！
SelCh = ga.selecting(selectStyle, Chrom, FitnV, GGAP, SUBPOP) # 选择
SelCh = ga.recombin(recombinStyle, SelCh, recopt, SUBPOP) #交叉
SelCh = ga.mutbga(SelCh, FieldDR, pm) # 变异
if repnum > Chrom.shape[0] * 0.1: # 进行一次高斯变异
SelCh = ga.mutgau(SelCh, FieldDR, pm) # 高斯变异
ObjVSel = aimfuc(SelCh) # 求育种个体的目标函数值
[CombinObjV, weight] = ga.awGA(maxormin * ObjVSel)
FitnVSel = ga.ranking(maxormin * CombinObjV)
[Chrom, ObjV] = ga.reins(Chrom, SelCh, SUBPOP, 1, 0.9, FitnV, FitnVSel,
ObjV, ObjVSel) #重插入
if drawing == 2:
ax = ga.frontplot(NDSetObjV, False, ax, gen + 1) # 绘制动态图
end_time = time.time() # 结束计时
#=========================绘图及输出结果=========================
if drawing != 0:
ga.frontplot(NDSetObjV, True)
times = end_time - start_time
print('用时：', times, '秒')
print('帕累托前沿点个数：', NDSet.shape[0], '个')
print('单位时间找到帕累托前沿点个数：', int(NDSet.shape[0] // times), '个')
# 返回帕累托最优集以及执行时间
return [ObjV, NDSet, NDSetObjV, end_time - start_time]
def optimize(self):
"""===============================实例化问题对象==========================="""
problem = MyProblem(self.M, self.maxormins, self.Dim, self.varTypes,
self.lb, self.ub, self.lbin, self.ubin,
self.f_list) # 生成问题对象
"""=================================种群设置==============================="""
Encoding = 'BG' # 编码方式,使用二进制/格雷编码
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders) # 创建区域描述器
population = ea.Population(
Encoding, Field, self.NIND) # 实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
"""===============================算法参数设置============================="""
if self.M == 1:
myAlgorithm = ea.soea_SGA_templet(
problem, population) # 实例化一个算法模板对象, single objective
else:
myAlgorithm = ea.moea_NSGA3_templet(
problem, population) # 实例化一个算法模板对象, multi objective
myAlgorithm.MAXGEN = self.MAXGEN # 最大进化代数
myAlgorithm.recOper.XOVR = 0.7 # 交叉概率
myAlgorithm.mutOper.Pm = 1 # 变异概率(换算成标准变异概率为0.025)
myAlgorithm.drawing = 0 # 设置绘图方式（0：不绘图；1：绘制结果图；2：绘制目标空间过程动画；3：绘制决策空间过程动画）
"""==========================调用算法模板进行种群进化======================="""
if self.M == 1: # single objective
[population, obj_trace, var_trace] = myAlgorithm.run() # 执行算法模板
self.obj_trace = obj_trace # save obj_trace for plot
population.save() # 把最后一代种群的信息保存到文件中
# 输出结果
best_gen = np.argmin(problem.maxormins *
obj_trace[:, 1]) # 记录最优种群个体是在哪一代
best_ObjV = np.array([obj_trace[best_gen, 1]])
best_var = np.array([var_trace[best_gen]])
if os.path.exists('best_ObjV.csv') == False:
np.savetxt('best_ObjV.csv', best_ObjV, delimiter=',')
np.savetxt('best_var.csv', best_var, delimiter=',')
else:
best_ObjV_old = self.read_csv_to_np('best_ObjV.csv')
best_var_old = self.read_csv_to_np('best_var.csv')
best_ObjV = np.vstack((best_ObjV_old, best_ObjV))
best_var = np.vstack((best_var_old, best_var))
np.savetxt('best_ObjV.csv', best_ObjV, delimiter=',')
np.savetxt('best_var.csv', best_var, delimiter=',')
# print('最优的目标函数值为：%s' % (best_ObjV))
# print('最优的控制变量值为：')
# for i in range(var_trace.shape[1]):
# print(var_trace[best_gen, i])
# print('有效进化代数：%s' % (obj_trace.shape[0]))
# print('最优的一代是第 %s 代' % (best_gen + 1))
# print('评价次数：%s' % (myAlgorithm.evalsNum))
# print('时间已过 %s 秒' % (myAlgorithm.passTime))
return population
else: # multi objective
NDSet = myAlgorithm.run() # 执行算法模板，得到帕累托最优解集NDSet
NDSet.save() # 把结果保存到文件中
# 输出
# print('用时：%s 秒' % (myAlgorithm.passTime))
# print('非支配个体数：%s 个' % (NDSet.sizes))
# print('单位时间找到帕累托前沿点个数：%s 个' % (int(NDSet.sizes // myAlgorithm.passTime)))
return NDSet
def main(aim, n, Dim, phase):
start = time.perf_counter()
"""===============================实例化问题对象==========================="""
PoolType = 'Thread'
problem = MyProblem_multiprocess(aim, n, Dim, phase, PoolType) # 生成问题对象
"""=================================种群设置==============================="""
Encoding = 'RI' # 编码方式
NIND = 1000 # 种群规模
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders) # 创建区域描述器
population = ea.Population(Encoding, Field,
NIND) # 实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
"""===============================算法参数设置============================="""
myAlgorithm = ea.soea_SEGA_templet(problem, population) # 实例化一个算法模板对象
# myAlgorithm.mutOper.Pm = 1 # 修改变异算子的变异概率(原模板中breeder GA变异算子的Pm定义为1 / Dim)
# myAlgorithm.recOper.XOVR = 0.9 # 修改交叉算子的交叉概率
# myAlgorithm.recOper.XOVR = 0.7 # 重组概率
# myAlgorithm.mutOper.F = 0.5 # 差分进化中的参数F
# if n == 0:
# myAlgorithm.MAXGEN = 20 # 最大进化代数
# myAlgorithm.maxTrappedCount = 10 # 进化停滞计数器最大上限值，如果连续maxTrappedCount代被判定进化陷入停滞，则终止进化
# else:
# myAlgorithm.MAXGEN = 20
myAlgorithm.MAXGEN = 40
myAlgorithm.trappedValue = 1e-6 # “进化停滞”判断阈值
myAlgorithm.maxTrappedCount = 10 # 进化停滞计数器最大上限值，如果连续maxTrappedCount代被判定进化陷入停滞，则终止进化
myAlgorithm.drawing = 2 # 设置绘图方式（0：不绘图；1：绘制结果图；2：绘制目标空间过程动画；3：绘制决策空间过程动画）
"""==========================调用算法模板进行种群进化======================="""
[population, obj_trace, var_trace] = myAlgorithm.run() # 执行算法模板
population.save() # 把最后一代种群的信息保存到文件中
# problem.pool.close() # 及时关闭问题类中的池，否则在采用多进程运算后内存得不到释放
# 输出结果
best_gen = np.argmin(problem.maxormins * obj_trace[:, 1]) # 记录最优种群个体是在哪一代
best_ObjV = obj_trace[best_gen, 1]
print('最优的目标函数值为：%.8smm' % (best_ObjV))
if phase == None:
phase = [0] + [i * 10 for i in var_trace[best_gen, :]] # 第一次算出的所有相位值
res = aim.bias_n_li_fast(phase)
find_index = np.where(max(res[n + 1:, 0]) == res[
n + 1:,
0])[0][0] + 1 # 第一次最大偏心值的索引位置（此处res内的n+1和最后面的+1表示：不考虑第一级，整体向后偏移一位）
else:
phase = phase + [i * 10
for i in var_trace[best_gen, :]] # 所有相位值(后续几级相位值有更新)
res = aim.bias_n_li_fast(phase)
find_index = np.where(max(res[n:, 0]) == res[n:,
0])[0][0] # 每次的最大偏心值的索引位置
print('所在位置及相位偏斜角为：第%s级零件末端; %.8s°' %
(find_index + 1 + n, res[find_index + n, 1]))
if n == 0:
for i in range(len(phase)):
print('第%s级相位、偏心值及相位偏斜角为：%s°; %.8smm; %.8s°' %
(i + 1, phase[i], res[i, 0], res[i, 1]))
else:
for i in range(var_trace.shape[1]):
print('第%s级相位、偏心值及相位偏斜角为：%s°; %.8smm; %.8s°' %
(i + (n + 1), phase[i + n], res[i + n, 0], res[i + n, 1]))
print('有效进化代数：%s' % (obj_trace.shape[0]))
print('最优的一代是第 %s 代' % (best_gen + 1))
print('评价次数：%s' % (myAlgorithm.evalsNum))
# print('时间已过 %s 分 %.8s 秒' % (int(myAlgorithm.passTime // 60), myAlgorithm.passTime % 60))
n = n + int(
(find_index + 1)) # 计算累计固定好的级数(此处find_index类型为int64,改为int类型防止后面报错！)
end = time.perf_counter()
print(f'时间已过 {int((end - start) // 60)} 分 {(end - start) % 60:.8f} 秒')
return population, myAlgorithm, obj_trace, var_trace, best_gen, best_ObjV, phase, res, n
# -*- coding: utf-8 -*-
import numpy as np
import geatpy as ea # import geatpy
from MyProblem import MyProblem # 导入自定义问题接口
"""==================================实例化问题对象================================"""
problem = MyProblem() # 生成问题对象
"""==================================种群设置================================"""
Encoding = 'R' # 编码方式
conordis = 0 # 表示染色体解码后得到的变量是连续的
NIND = 100 # 种群规模
Field = ea.crtfld(Encoding, conordis, problem.ranges, problem.borders) # 创建区域描述器
population = ea.Population(Encoding, conordis, Field, NIND) # 实例化种群对象（此时种群还没被真正初始化）
"""==================================算法参数设置================================"""
myAlgorithm = ea.soea_SEGA_templet(problem, population) # 实例化一个算法模板对象
myAlgorithm.MAXGEN = 500 # 最大遗传代数
"""=======================调用算法模板进行种群进化=============================="""
[population, obj_trace, var_trace] = myAlgorithm.run() # 执行算法模板
# 输出结果
best_gen = np.argmax(obj_trace[:, 1]) # 记录最优种群是在哪一代
best_ObjV = obj_trace[best_gen, 1]
print('最优的目标函数值为：%s'%(best_ObjV))
print('最优的控制变量值为：')
for i in range(var_trace.shape[1]):
print(var_trace[best_gen, i])
print('有效进化代数：%s'%(obj_trace.shape[0]))
print('最优的一代是第 %s 代'%(best_gen + 1))
print('评价次数：%s'%(myAlgorithm.evalsNum))
print('时间已过 %s 秒'%(myAlgorithm.passTime))
"""
import numpy as np
import geatpy as ga # import geatpy
AIM_M = __import__('aimfuc') # get the address of objective function
AIM_F = 'DTLZ1' # You can set DTL1,2,3 or 4
"""==================================variables setting================================"""
ranges = np.vstack([np.zeros((1, 7)), np.ones(
(1, 7))]) # define the ranges of variables in DTLZ1
borders = np.vstack([np.ones((1, 7)), np.ones(
(1, 7))]) # define the borders of variables in DTLZ1
precisions = [
4
] * 7 # set any precision that is bigger than 0, or 'crtfld' would create a Field for integer-code.
FieldDR = ga.crtfld(ranges, borders, precisions) # create the FieldDR
"""=======================use sga2_templet to find the Pareto front==================="""
[ObjV, NDSet, NDSetObjV,
times] = ga.moea_nsga2_templet(AIM_M,
AIM_F,
None,
None,
FieldDR,
problem='R',
maxormin=1,
MAXGEN=500,
MAXSIZE=2000,
NIND=50,
SUBPOP=1,
GGAP=1,
selectStyle='tour',
import geatpy as ga
# 获取函数接口地址
AIM_M = __import__('aimfuc')
# 变量设置
x1 = [-3, 12.1] # 自变量1的范围
x2 = [4.1, 5.8] # 自变量2的范围
b1 = [1, 1] # 自变量1是否包含下界
b2 = [1, 1] # 自变量2是否包含上界
codes = [0, 0] # 自变量的编码方式，0表示采用标准二进制编码
precisions = [4, 4] # 自变量的精度(精度不宜设置太高，否则影响搜索性能和效果)
scales = [0, 0] # 是否采用对数刻度
ranges = np.vstack([x1, x2]).T # 生成自变量的范围矩阵
borders = np.vstack([b1, b2]).T # 生成自变量的边界矩阵
# 生成区域描述器
FieldD = ga.crtfld(ranges, borders, precisions, codes)
# 调用编程模板
[pop_trace, var_trace, times] = ga.sga_new_code_templet(AIM_M,
'aimfuc',
None,
None,
FieldD,
problem='R',
maxormin=-1,
MAXGEN=1000,
NIND=100,
SUBPOP=1,
GGAP=0.8,
selectStyle='sus',
recombinStyle='xovdp',
# -*- coding: utf-8 -*-
import geatpy as ea # import geatpy
from MyProblem import MyProblem # 导入自定义问题接口
"""==================================实例化问题对象================================"""
problem = MyProblem() # 生成问题对象，实例化MyProblem的对象
"""==================================种群设置================================"""
Encoding = 'I' # 编码方式 ，这种编码方式是整数
conordis = 1 # 表示染色体解码后得到的变量是离散的 表示解码后的变量是离散的//1表示的是连续
NIND = 50 # 种群规模，总变量个数是50，解空间个数是50
Field = ea.crtfld(Encoding, conordis, problem.ranges,
problem.borders) # 创建区域描述器 根据编码方式/解码后的变量类型/决策边界/
## 包含边界来确定区域描述
population = ea.Population(Encoding, conordis, Field,
NIND) # 实例化种群对象（此时种群还没被真正初始化）
#用编码方式/解码后的变量类型/解空间/种群规模来初始化population
"""==================================算法参数设置================================"""
myAlgorithm = ea.moea_NSGA2_templet(
problem, population) # 实例化一个算法模板对象,传入problem/population，
# 也就是生成问题对象和种群对象
myAlgorithm.MAXGEN = 200 # 最大遗传代数
"""=======================调用算法模板进行种群进化=============================="""
NDSet = myAlgorithm.run() # 执行算法模板，得到帕累托最优解集NDSet，直接输出解集
print(population.FitnV)
# 输出
#print('用时：%s 秒'%(myAlgorithm.passTime))
#print('非支配个体数：%s 个'%(NDSet.sizes))
#print(NDSet.Phen.shape)
#print(NDSet.Phen)
#patore
#print('单位时间找到帕累托前沿点个数：%s 个'%(int(NDSet.sizes // myAlgorithm.passTime)))
# -*- coding: utf-8 -*-
import geatpy as ea # import geatpy
from MyProblem import MyProblem # 导入自定义问题接口
if __name__ == '__main__':
"""===============================实例化问题对象============================"""
problem = MyProblem() # 生成问题对象
"""==================================种群设置==============================="""
Encoding = 'RI' # 编码方式
NIND = 50 # 种群规模
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders) # 创建区域描述器
population = ea.Population(Encoding, Field,
NIND) # 实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
"""=================================算法参数设置============================"""
myAlgorithm = ea.moea_NSGA2_templet(problem, population) # 实例化一个算法模板对象
myAlgorithm.MAXGEN = 200 # 最大进化代数
myAlgorithm.logTras = 0 # 设置每多少代记录日志，若设置成0则表示不记录日志
myAlgorithm.verbose = False # 设置是否打印输出日志信息
myAlgorithm.drawing = 1 # 设置绘图方式（0：不绘图；1：绘制结果图；2：绘制目标空间过程动画；3：绘制决策空间过程动画）
"""==========================调用算法模板进行种群进化=========================
调用run执行算法模板，得到帕累托最优解集NDSet以及最后一代种群。NDSet是一个种群类Population的对象。
NDSet.ObjV为最优解个体的目标函数值；NDSet.Phen为对应的决策变量值。
详见Population.py中关于种群类的定义。
"""
[NDSet, population] = myAlgorithm.run() # 执行算法模板，得到非支配种群以及最后一代种群
NDSet.save() # 把非支配种群的信息保存到文件中
"""==================================输出结果=============================="""
print('用时：%s 秒' % myAlgorithm.passTime)
print('非支配个体数：%d 个' %
NDSet.sizes) if NDSet.sizes != 0 else print('没有找到可行解！')
def __init__(self, Encodings, Fields=None, NIND=None, Chroms=None, ObjV=None, FitnV=None, CV=None, Phen=None, EncoIdxs=None):
"""
描述: 种群类的构造函数，用于实例化种群对象，例如：
import geatpy as ea
population = ea.PsyPopulation(Encodings, Fields, NIND)，
NIND为所需要的个体数，
此时得到的population还没被真正初始化，仅仅是完成种群对象的实例化。
该构造函数必须传入Chroms，才算是完成种群真正的初始化。
因此一开始可以只传入Encodings, Fields以及NIND来完成种群对象的实例化，
其他属性可以后面再通过计算进行赋值。
特殊用法1：
可以利用ea.PsyPopulation(Encodings, Fields, 0)来创建一个“空种群”,即不含任何个体的种群对象。
特殊用法2：
直接用ea.PsyPopulation(Encodings)构建一个只包含编码信息的空种群。
"""
if NIND is None:
NIND = 0
if isinstance(NIND, int) and NIND >= 0:
self.sizes = NIND
else:
raise RuntimeError('error in PysPopulation: Size error. (种群规模设置有误，必须为非负整数。)')
self.EncoIdxs = EncoIdxs
self.Encodings = Encodings
if Encodings is None:
self.Fields = None
self.Chroms = None
self.Linds = 0
self.ChromNum = 2 # 随意设置一个大于1的值
else:
self.ChromNum = len(Encodings)
if self.ChromNum == 1:
raise RuntimeError(
'error in PysPopulation: ChromNum must be bigger than 1. ('
'使用PysPopulation类时，染色体数目必须大于1，否则应该使用Population类。)')
if Fields is None:
self.Fields = None
else:
if type(Fields) == tuple:
self.Fields = []
if len(EncoIdxs) != len(Encodings):
raise RuntimeError('error in PysPopulation: Dimension disagree. (EncoIdxs的元素个数与Encodings的不一致。)')
for i, Encoding in enumerate(Encodings):
params = []
for item in Fields:
if item.ndim == 1:
params.append(item[EncoIdxs[i]])
elif item.ndim == 2:
params.append(item[:, EncoIdxs[i]])
else:
raise RuntimeError('error in PysPopulation: Parameter error. (参数错误，详见文件头的注释。)')
params = tuple(params)
self.Fields.append(ea.crtfld(Encoding, *params))
else:
self.Fields = Fields.copy()
self.Chroms = [None] * self.ChromNum # 初始化Chroms为元素是None的列表
self.Linds = []
if Chroms is None:
self.Linds = [0] * self.ChromNum
else:
for i in range(self.ChromNum):
if Chroms[i] is not None:
self.Linds.append(Chroms[i].shape[1])
self.Chroms[i] = Chroms[i].copy() if Chroms[i] is not None else None
else:
self.Linds.append(0)
self.ObjV = ObjV.copy() if ObjV is not None else None
self.FitnV = FitnV.copy() if FitnV is not None else None
self.CV = CV.copy() if CV is not None else None
self.Phen = Phen.copy() if Phen is not None else None
# 注意：不建议把目标函数放在执行脚本内，建议放在另一个文件中
def aimfuc(x, LegV): # 定义目标函数
f = np.array([np.sum(x**2, 1)]).T # 注意计算结果要符合Geatpy对种群目标函数矩阵的数据结构
return [f, LegV] # 注意返回的参数要符合Geatpy数据结构
if __name__ == "__main__":
AIM_M = __import__('sga_test_Sphere') # 获取函数所在文件的地址
# 变量设置
ranges = np.array([[-5.12, -5.12, -5.12], [5.12, 5.12,
5.12]]) # 生成自变量的范围矩阵
borders = np.array([[1, 1, 1], [1, 1, 1]]) # 生成自变量的边界矩阵（1表示变量的区间是闭区间）
precisions = [
1, 1, 1
] # 根据crtfld的函数特性，这里需要设置精度为任意正值，否则在生成区域描述器时会默认为整数编码，并对变量范围作出一定调整
FieldDR = ga.crtfld(ranges, borders, precisions) # 生成区域描述器
# 调用编程模板
[pop_trace, var_trace,
times] = ga.sga_new_real_templet(AIM_M,
'aimfuc',
None,
None,
FieldDR,
problem='R',
maxormin=1,
MAXGEN=100,
NIND=50,
SUBPOP=1,
GGAP=0.9,
selectStyle='sus',
recombinStyle='xovsp',
def geaty_func(func,
dim,
lb,
ub,
NIND=400,
MAXGEN=25,
fig_dir="",
njobs=0,
method="SEGA",
n_populations=5,
opt_res=None):
"""
将整个遗传算法过程进行整合，编写成一整个函数，便于之后使用。当前此函数，只能处理参数是连续
型的最优化问题。
args：
func: 用于最小化的函数。
dim: 需要优化的params的个数。
lb，ub：array或list，是params的上下限，这里默认都是闭区间。
NIND: 种群规模
MAXGEN: 最大进化代数
fig_dir: 保存图片的地址，注意，需要在最后加/
njobs: 0，1使用单核、否则使用多核，其中-1表示使用所有的核
method: 使用的进化算法模板：
SEGA：增强精英保留的遗传算法模板
multiSEGA：增强精英保留的多种群协同遗传算法
psySEGA：增强精英保留的多染色体遗传算法
rand1lDE: 差分进化DE/rand/1/L算法
n_populations: 如果是multiSEGA，此表示的是种群数量
"""
""" 实例化问题对象 """
problem = MyProblem(func, dim, lb, ub, njobs=njobs) # 生成问题对象
""" 种群设置 """
# 创建区域描述器
# (不同的方法使用不同的编码方式)
# if "DE" in method:
Encoding = "RI"
# else:
# Encoding = "BG"
# 实例化种群对象（此时种群还没被初始化，仅仅是完成种群对象的实例化）
if method == "multiSEGA":
one_ind = NIND // n_populations
res_ind = NIND % n_populations
NINDS = [one_ind] * n_populations
NINDS[-1] = NINDS[-1] + res_ind
population = []
for nind in NINDS:
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders)
population.append(ea.Population(Encoding, Field, nind))
# --------------------------------------------
# opt_res = load("./RESULTS2/test0418_5/opt_res.pkl")
if opt_res is not None:
prophetChrom = np.expand_dims(opt_res["BestParam"], axis=0)
prophetChrom = np.tile(opt_res["BestParam"], (2, 1))
prophetPops = []
for i in range(len(population)):
prophetPop = ea.Population(Encoding, Field, 2, prophetChrom)
prophetPop.Phen = prophetPop.decoding()
problem.aimFunc(prophetPop)
prophetPops.append(prophetPop)
else:
prophetPops = None
# --------------------------------------------
else:
# --------------------------------------------
# prophetPops = None
# --------------------------------------------
Field = ea.crtfld(Encoding, problem.varTypes, problem.ranges,
problem.borders)
if method == "psySEGA":
raise NotImplementedError
population = ea.PsyPopulation([Encoding], [Field], NIND)
else:
population = ea.Population(Encoding, Field, NIND)
""" 算法参数设置 """
if method == "SEGA":
algorithm = ea.soea_SEGA_templet
elif method == "multiSEGA":
algorithm = ea.soea_multi_SEGA_templet
elif method == "psySEGA":
algorithm = ea.soea_psy_SEGA_templet
elif method == "rand1lDE":
algorithm = ea.soea_DE_rand_1_L_templet
else:
raise NotImplementedError
# 实例化一个算法模板对象
myAlgorithm = algorithm(problem, population)
# 最大进化代数
myAlgorithm.MAXGEN = MAXGEN
# "进化停滞"判断阈值
# myAlgorithm.trappedValue = 1e-6
# 进化停滞计数器最大上限值，如果连续maxTrappedCount代被判定进化陷入停滞，则终止进化
# myAlgorithm.maxTrappedCount = 100
# 控制是否绘制图片
myAlgorithm.drawing = 1
# 控制绘图的路径（自己改的源码）
myAlgorithm.drawing_file = fig_dir
""" 调用算法模板进行种群进化 """
[population, obj_trace, var_trace] = myAlgorithm.run(prophetPops) # 执行算法模板
# population.save() # 把最后一代种群的信息保存到文件中
""" 输出结果 """
# 记录最优种群个体是在哪一代
best_gen = np.argmin(problem.maxormins * obj_trace[:, 1])
best_ObjV = obj_trace[best_gen, 1]
print('最优的目标函数值为：%s' % (best_ObjV))
# print('最优的控制变量值为：')
# for i in range(var_trace.shape[1]):
# print(var_trace[best_gen, i])
print('有效进化代数：%s' % (obj_trace.shape[0]))
print('最优的一代是第 %s 代' % (best_gen + 1))
print('评价次数：%s' % (myAlgorithm.evalsNum))
print('时间已过 %s 秒' % (myAlgorithm.passTime))
return {
"BestObjv": best_ObjV,
"BestParam": var_trace[best_gen, :],
"AvailableEvoluion": obj_trace.shape[0],
"EvaluationCount": myAlgorithm.evalsNum,
"PassTime": myAlgorithm.passTime,
"VarTrace": var_trace,
"ObjTrace": obj_trace,
"population": population,
}
