def plt(self):
if self.value_name is None:
self.value_name = [str(i) for i in range(self.size)]
if self.size == 0:
return
if len(self.value_name) != self.size:
raise ValueError('size of value_name and size are not match')
para = copy.deepcopy(self.para)
fig = plt.figure(**para)
for pic_id in range(self.size):
legend_name = []
def color_list():
for col in ['y', 'purple', 'blue', 'red']:
yield col
color_generator = color_list()
for (key, values) in self.data.items():
values_array = np.array(values)
legend_name.append(key)
line_color = 'k' if key=='set point' else next(color_generator)
x_array = np.arange(0, values_array.shape[0], 1)
if line_color is 'k':
plt.plot(x_array/3, values_array[:, pic_id], c=line_color,linewidth=4)
else:
plt.plot(x_array/3, values_array[:, pic_id], c=line_color, linewidth=1)
if not self.value_name[pic_id] == 'Concentration(In)' \
and not self.value_name[pic_id] == 'pulp speed(Feed In)':
plt.legend(legend_name)
plt.title(self.value_name[pic_id])
plt.xlabel('time(minute)')
from Control_Exp1001.demo.thickener.common import exp_name
img_root = os.path.join('../images/', exp_name.get_exp_name()) +'/'
img_name = str(self.value_name[pic_id])+'_'.join(legend_name)
img_name = img_name.replace(' ', '_')
plt.savefig(img_root + img_name + '.png', dpi=300)
plt.show()
#"weight_matrix": [0, 0.002],
"weight_matrix": [0, 0.004],
"S": [0.0001, 0.0008],
#"S": [0.0003, 0.0024],
#"S": [0.0000, 0.000],
}
thickner_para = {
"dt": 1,
"noise_in": False,
"noise_p": 0.002,
"noise_type": 3,
'time_length': 20, # 浓密机每次仿真20秒
}
from Control_Exp1001.demo.thickener.common import exp_name
exp_name.set_exp_name('VIandHDP')
EXP_NAME = exp_name.get_exp_name()
img_path = os.path.join('../images', EXP_NAME)
if not os.path.exists(img_path):
os.mkdir(img_path)
def new_vi():
capacity = 2
predict_round = 3000
u_optim = 'sgd'
gamma = 0.6
replay_vi = ReplayBuffer(capacity=capacity)
env_VI = Thickener(
noise_p=0.03,
noise_in=True,
)
