def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
# p1_VM_Output : lambda_pls 반영 10 runs act결과 반영 vm
# p1_ACT_Output : lambda_pls 반영 10 runs act결과 반영
# p1_ez_run : prd - act (실제)
# p1_y_act : lambda_pls 반영 안한 act
# p1_y_prd : lambda_pls 반영 안하고 10 runs act결과 반영하지 않은 예측 값
p1_VM_Output, p1_ACT_Output, p1_ez_run, p1_y_act, p1_y_prd = fwc_p1_vm.VM_Run(lamda_PLS=v1_PLS, Z=Z_VM, M=M)
# p1_q1_mape_Queue = []
#
# # metrology 마다 보여주는 MAPE 값이 의미가 없다.
# for z in np.arange(0, Z_VM, 1):
# mape = fdh_graph.mean_absolute_percentage_error(z + 1, p1_y_act[((z + 1) * M) - 1][0], p1_y_prd[((z + 1) * M) - 1][0])
# p1_q1_mape_Queue.append(mape)
#
# print('Process-1 q1 Every Metrology MAPE : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
# p1_q1_mape_Queue = []
#
# for i in range(Z_VM * M):
# mape = fdh_graph.mean_absolute_percentage_error(i + 1, p1_y_act[i][0], p1_y_prd[i][0])
# p1_q1_mape_Queue.append(mape)
#
# print('Process-1 q1 All MAPE : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
#
# fdh_graph.plt_show1(Z_VM * M, p1_y_act[:, 0:1], p1_y_prd[:, 0:1])
# fdh_graph.plt_show2(Z_VM, p1_ez_run[:, 0:1], p1_ez_run[:, 1:2])
#
# np.savetxt("output/p1_VM_Output.csv", p1_VM_Output, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_ACT_Output.csv", p1_ACT_Output, delimiter=",", fmt="%.4f")
min_mape = 100
t = 1000
for m in range(1000):
SEED2 = t + m
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd, p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M, f=p1_y_act) #DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd = fwc_p2_act.VM_Run(lamda_PLS=v2_PLS, Z=Z_VM, M=M)
p2_q2_mape_Queue = []
for i in range(Z_VM * M):
mape = fdh_graph.mean_absolute_percentage_error(i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
mape = np.mean(p2_q2_mape_Queue)
print('Process-2 q2 All MAPE : {0:.2f}%'.format(mape))
if mape < min_mape:
min_mape = mape
seed_right = SEED2
print('SEED2 : ', SEED2, ' , min_mape = ', min_mape)
print('seed_right : ', seed_right, ' , min_mape = ', min_mape)
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_DynamicSampling_노이즈시뮬레이터(A_p1, d_p1, C_p1, dM,
dStart, SEED)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) #DoE Run
VM_Output, ACT_Output, ez_run, y_act, y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=RUNS_CNT, M=M)
np.savetxt("output/VM_Output.csv", VM_Output, delimiter=",", fmt="%.4f")
np.savetxt("output/ACT_Output.csv", ACT_Output, delimiter=",", fmt="%.4f")
np.savetxt("output/ez_run.csv", ez_run, delimiter=",", fmt="%.4f")
np.savetxt("output/y_act.csv", y_act, delimiter=",", fmt="%.4f")
np.savetxt("output/y_prd.csv", y_prd, delimiter=",", fmt="%.4f")
fdh_graph.plt_show1(N, y_act[:, 0:1], y_prd[:, 0:1])
fdh_graph.plt_show2(RUNS_CNT, ez_run[:, 0:1], ez_run[:, 1:2], Noise=True)
ez_run_out = []
runM = M
ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, RUNS_CNT + 1):
if z == dStart:
runM = dM
for k in np.arange(z * runM, (z + 1) * runM):
ez_run_out.append(ez_run[z])
ez_run_out = np.array(ez_run_out)
np.savetxt("output/ez_run2.csv", ez_run_out, delimiter=",", fmt="%.4f")
fdh_graph.plt_show5(ez_run_out, N, M, dM, S1, Noise=True)
p1_q1_mape_Queue = []
#metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, 40, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, y_act[((z + 1) * M) - 1][0], y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, N, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, y_act[i][0], y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) #DoE Run
VM_Output, ACT_Output, ez_run, y_act, y_prd, ez_all_run = fwc_p1_vm.VM_Run(lamda_PLS=v_PLS, Z=Z_VM, M=M)
fdh_graph.plt_show1(Z_VM * M, y_act[:, 0:1], y_prd[:, 0:1], 'Process-1 Normal Case', '1')
fdh_graph.plt_show2(Z_VM, ez_run[:, 0:1], ez_run[:, 1:2], 'Process-1 Normal Case')
p1_q1_mape_Queue = []
#metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(0, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(z + 1, y_act[((z + 1) * M) - 1][0], y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in range(Z_VM * M):
mape = fdh_graph.mean_absolute_percentage_error(i + 1, y_act[i][0], y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
# np.savetxt("output/Normal_mape_Queue_Output.csv", p1_mape_Queue, delimiter=",", fmt="%.4f")
# np.savetxt("output/Normal_VM_Output.csv", VM_Output, delimiter=",", fmt="%.4f")
np.savetxt("output/ez_all_run.csv", ez_all_run, delimiter=",", fmt="%.4f")
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) #DoE Run
VM_Output, ACT_Output, ez_run, y_act, y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=Z_VM, M=M)
fdh_graph.plt_show1(Z_VM * M, y_act[:, 0:1], y_prd[:, 0:1])
fdh_graph.plt_show2(Z_VM, ez_run[:, 0:1], ez_run[:, 1:2])
p1_q1_mape_Queue = []
#metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(0, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, y_act[((z + 1) * M) - 1][0], y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in range(Z_VM * M):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, y_act[i][0], y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_DynamicSampling_시뮬레이터(A_p1, d_p1, C_p1, dM, dStart,
SEED)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) # DoE Run
VM_Output, ACT_Output, ez_run, y_act, y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=RUNS_CNT, M=M)
np.savetxt("output/VM_Output.csv", VM_Output, delimiter=",", fmt="%.4f")
np.savetxt("output/ACT_Output.csv", ACT_Output, delimiter=",", fmt="%.4f")
np.savetxt("output/ez_run.csv", ez_run, delimiter=",", fmt="%.4f")
np.savetxt("output/y_act.csv", y_act, delimiter=",", fmt="%.4f")
np.savetxt("output/y_prd.csv", y_prd, delimiter=",", fmt="%.4f")
fdh_graph.plt_show1(N, y_act[:, 0:1], y_prd[:, 0:1])
fdh_graph.plt_show2(RUNS_CNT, ez_run[:, 0:1], ez_run[:, 1:2], Noise=False)
ez_run_out = []
runM = M
ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, RUNS_CNT + 1):
if z == dStart:
runM = dM
for k in np.arange(z * runM, (z + 1) * runM):
ez_run_out.append(ez_run[z])
#ez_run_out[0] = np.array([0, 0])
ez_run_out = np.array(ez_run_out)
np.savetxt("output/ez_run2.csv", ez_run_out, delimiter=",", fmt="%.4f")
fdh_graph.plt_show5(ez_run_out, N, M, dM, S1, Noise=False)
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_노이즈시뮬레이터(A_p1, d_p1, C_p1, SEED)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) #DoE Run
VM_Output, ACT_Output, ez_run, y_act, y_prd, abnormal_p1_ez_all_run_q1 = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=Z_VM, M=M)
# fdh_graph.plt_show1(Z_VM * M, y_act[:, 0:1], y_prd[:, 0:1])
# fdh_graph.plt_show2(Z_VM, ez_run[:, 0:1], ez_run[:, 1:2], Noise=True)
fdh_graph.plt_show1(
Z_VM * M, y_act[:, 0:1], y_prd[:, 0:1],
'Process-1 Abnormal Case (' + str(Z_VM) + ' Metrology Runs)', '1')
#fdh_graph.plt_show2_1(Z_VM * M, ez_run[:, 0:1], ez_run[:, 1:2], 'Process-1 Abnormal Case', color1='bx--', color2='rx-')
#np.savetxt("output/p1_ez_run.csv", ez_run, delimiter=",", fmt="%.4f")
# for z in np.arange(22, 34, 1):
# mape = fdh_graph.mean_absolute_percentage_error(z + 1, y_act[((z + 1) * M) - 1][0], y_prd[((z + 1) * M) - 1][0])
# p1_mape_Queue.append(mape)
# for z in np.arange(21, 32, 1):
# mape_sum = 0
# for k in np.arange(z * M + 1, ((z + 1) * M) + 1):
# mape = fdh_graph.mean_absolute_percentage_error(k, y_act[k][0], y_prd[k][0])
# mape_sum += mape
# p1_mape_Queue.append(mape_sum / M)
p1_q1_mape_Queue = []
p1_q1_mape_Queue_temp = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN_1, Z_VM, 1):
act = y_act[((z + 1) * M) - 1][0]
prd = y_prd[((z + 1) * M) - 1][0]
mape = fdh_graph.mean_absolute_percentage_error(z + 1, act, prd)
if z >= 15 and z < 20:
print('z = ', z, ', act = ', act, ', prd = ', prd, ', ez = ',
np.abs(act - prd), ', MAPE = ', mape)
p1_q1_mape_Queue_temp.append(np.abs(act - prd))
else:
p1_q1_mape_Queue.append(mape)
print(
'Process-1 q1 Actual Metrology MAPE between 5 Lots (15 ~ 20 Lots) : {0:.2f}'
.format(np.mean(p1_q1_mape_Queue_temp)))
print('Process-1 q1 Every Metrology MAPE After 20 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN_2 * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, y_act[i][0], y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 20 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
# np.savetxt("output/noise_mape_Queue.csv", p1_mape_Queue, delimiter=",", fmt="%.4f")
# np.savetxt("output/abNormal_VM_Output.csv", VM_Output, delimiter=",", fmt="%.4f")
np.savetxt("output/noise_ez_run.csv", ez_run, delimiter=",", fmt="%.4f")
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_노이즈시뮬레이터(A_p1, d_p1, C_p1, SEED)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) #DoE Run
VM_Output, ACT_Output, ez_run, y_act, y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=Z_VM, M=M)
fdh_graph.plt_show1(Z_VM * M, y_act[:, 0:1], y_prd[:, 0:1])
fdh_graph.plt_show12(Z_VM, ez_run[:, 0:1], ez_run[:, 1:2])
#np.savetxt("output/p1_ez_run.csv", ez_run, delimiter=",", fmt="%.4f")
# for z in np.arange(22, 34, 1):
# mape = fdh_graph.mean_absolute_percentage_error(z + 1, y_act[((z + 1) * M) - 1][0], y_prd[((z + 1) * M) - 1][0])
# p1_mape_Queue.append(mape)
# for z in np.arange(21, 32, 1):
# mape_sum = 0
# for k in np.arange(z * M + 1, ((z + 1) * M) + 1):
# mape = fdh_graph.mean_absolute_percentage_error(k, y_act[k][0], y_prd[k][0])
# mape_sum += mape
# p1_mape_Queue.append(mape_sum / M)
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, y_act[((z + 1) * M) - 1][0], y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, y_act[i][0], y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_DynamicSampling_노이즈시뮬레이터(A_p1, d_p1, C_p1, dM,
dStart, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) # DoE Run
Error_VMresult, Error_ACTResult, ez_run, p1_y_act, p1_y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=RUNS_CNT, M=M)
ez_run_out = []
runM = M
ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, RUNS_CNT + 1):
if z == dStart:
runM = dM
for k in np.arange(z * runM, (z + 1) * runM):
ez_run_out.append(ez_run[z])
ez_run_out = np.array(ez_run_out)
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN_1, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN_1 * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
min_mape = 100
t = 1000
for m in range(1000):
SEED2 = t + m
fwc_p2_act = VM_Process2_노이즈시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS,
p1_y_prd, p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M,
f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd = fwc_p2_act.VM_Run(
lamda_PLS=v2_PLS, Z=Z_VM, M=M)
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN_2 * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
mape = np.mean(p2_q2_mape_Queue)
print('Process-2 q2 All MAPE After 20 Lot : {0:.2f}%'.format(mape))
if mape < min_mape:
min_mape = mape
seed_right = SEED2
print('SEED2 : ', SEED2, ' , min_mape = ', min_mape)
print('seed_right : ', seed_right, ' , min_mape = ', min_mape)
p2_q2_mape_Queue = []
for z in np.arange(15, 20, 1): # 15 ~ 20 5개 Metrology만 본 상태
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p2_y_act[((z + 1) * M) - 1][1],
p2_y_prd[((z + 1) * M) - 1][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 Every Metrology MAPE After 5 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd = fwc_p1_vm.VM_Run(lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_DynamicSampling_노이즈시뮬레이터(A_p1, d_p1, C_p1, dM, dStart, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) # DoE Run
VM_Output, ACT_Output, ez_run, p1_y_act, p1_y_prd = fwc_p1_vm.VM_Run(lamda_PLS=v_PLS, Z=RUNS_CNT, M=M)
fdh_graph.plt_show1(N, p1_y_act[:, 0:1], p1_y_prd[:, 0:1])
ez_run_out = []
runM = M
ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, RUNS_CNT + 1):
if z == dStart:
runM = dM
for k in np.arange(z * runM, (z + 1) * runM):
ez_run_out.append(ez_run[z])
ez_run_out = np.array(ez_run_out)
fdh_graph.plt_show5(ez_run_out, N, M, dM, S1, Noise=True)
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(z + 1, p1_y_act[((z + 1) * M) - 1][0], p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
np.savetxt("output/p1_mape.csv", p1_q1_mape_Queue, delimiter=",", fmt="%.4f")
np.savetxt("output/Error_VMresult1.csv", VM_Output, delimiter=",", fmt="%.4f")
np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
np.savetxt("output/Error_ACTResult.csv", ACT_Output, delimiter=",", fmt="%.4f")
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd, p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M, f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd = fwc_p2_act.VM_Run(lamda_PLS=v2_PLS, Z=Z_VM, M=M)
fdh_graph.plt_show1(Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2])
fdh_graph.plt_show2(Z_VM, p2_ez_run[:, 0:1], p2_ez_run[:, 1:2], Noise=True)
p2_q2_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(z + 1, p2_y_act[((z + 1) * M) - 1][1], p2_y_prd[((z + 1) * M) - 1][1])
#print("으아아악 : ", p2_y_act[((z + 1) * M) - 1][1], p2_y_prd[((z + 1) * M) - 1][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(np.mean(p2_q2_mape_Queue)))
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 All MAPE After 15 Lot : {0:.2f}%'.format(np.mean(p2_q2_mape_Queue)))
def main():
noise_ez_run = np.loadtxt('output/noise_ez_run.csv',
delimiter=",",
dtype=np.float32)
abnormal_p2_ez_all_run_q2 = np.loadtxt(
'D:/10. 대학원/04. Source/09. VM_Source/04. VMOnly/output/abnormal_p2_ez_all_run_q2.csv',
delimiter=",",
dtype=np.float32)
before_normal_p2_ez_all_run_q2 = np.loadtxt(
'D:/10. 대학원/04. Source/09. VM_Source/04. VMOnly/output/before_normal_p2_ez_all_run_q2.csv',
delimiter=",",
dtype=np.float32)
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd, _ = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_DynamicSampling_노이즈시뮬레이터(A_p1, d_p1, C_p1, dM,
dStart, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) # DoE Run
VM_Output, ACT_Output, ez_run, p1_y_act, p1_y_prd, p1_ez_all_run_q1 = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=RUNS_CNT, M=M)
#fdh_graph.plt_show1(N, p1_y_act[:, 0:1], p1_y_prd[:, 0:1])
ez_run_out = []
noise_ez_run_out = []
runM = M
noise_ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, old_N + 1):
for k in np.arange(z * M, (z + 1) * M):
noise_ez_run_out.append(noise_ez_run[z])
noise_ez_run_out = np.array(noise_ez_run_out)
ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, RUNS_CNT + 1):
if z == dStart:
runM = dM
for k in np.arange(z * runM, (z + 1) * runM):
ez_run_out.append(ez_run[z])
ez_run_out = np.array(ez_run_out)
#fdh_graph.plt_show5(ez_run_out, N, M, dM, S1, Noise=True)
#fdh_graph.plt_show5_1(noise_ez_run_out, ez_run_out, N, M, dM, S1, type=1)
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd,
p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M,
f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd, p2_ez_all_run_q2 = fwc_p2_act.VM_Run(
lamda_PLS=v2_PLS, Z=Z_VM, M=M)
#fdh_graph.plt_show1(Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2])
#fdh_graph.plt_show2(Z_VM, p2_ez_run[:, 0:1], p2_ez_run[:, 1:2], Noise=True)
fdh_graph.plt_show1(
Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2],
'Process-2 Abnormal Case with Process-1 Dynamic Sampling', '2')
fdh_graph.plt_show2_2(
Z_VM * M,
abnormal_p2_ez_all_run_q2[:, 1:2],
p2_ez_all_run_q2[:, 1:2],
'Process-2 UPStream Rule X with Process-1 Dynamic Sampling',
'2',
color1='bx-',
color2='rx--')
#fdh_graph.plt_show2_2(Z_VM * M, abnormal_p2_ez_all_run_q2[:, 1:2], p2_ez_all_run_q2[:, 1:2], before_normal_p2_ez_all_run_q2[:, 1:2], 'Process-2 Abnormal Case', '2', color1='bx-', color2='rx--', color3='gx--')
p2_q2_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p2_y_act[((z + 1) * M) - 1][1],
p2_y_prd[((z + 1) * M) - 1][1])
#print("으아아악 : ", p2_y_act[((z + 1) * M) - 1][1], p2_y_prd[((z + 1) * M) - 1][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
def main():
fdh_graph = FDC_Graph()
## Metrology마다 차이를 보여줌.. 근데.. Process-2의 변동량이 크지 않아 바꾸기로 결정 ('19.09.25)
normal_p1_ez_all_run_q1 = np.loadtxt('output/p1_ez_all_run_q1.csv',
delimiter=",",
dtype=np.float32)
normal_p2_ez_all_run_q2 = np.loadtxt('output/p2_ez_all_run_q2.csv',
delimiter=",",
dtype=np.float32)
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd, _ = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_노이즈시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Error_VMresult, Error_ACTResult, ez_run, p1_y_act, p1_y_prd, abnormal_p1_ez_all_run_q1 = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fdh_graph.plt_show1(Z_VM * M, p1_y_act[:, 0:1], p1_y_prd[:, 0:1],
'Process-1 Abnormal Case', '1')
fdh_graph.plt_show2_2(Z_VM * M,
normal_p1_ez_all_run_q1[:, 0:1],
abnormal_p1_ez_all_run_q1[:, 0:1],
'Process-1 Abnormal Case',
'1',
color1='bx-',
color2='rx--')
#fdh_graph.plt_show2_1(Z_VM * M, p1_ez_run_q1, ez_run[:, 0:1], 'Process-1 Abnormal Case', color1='bx--', color2='rx-')
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
# np.savetxt("output/p1_mape.csv", p1_q1_mape_Queue, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_VMresult1.csv", Error_VMresult, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_ACTResult.csv", Error_ACTResult, delimiter=",", fmt="%.4f")
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd,
p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M,
f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd, abnormal_p2_ez_all_run_q2 = fwc_p2_act.VM_Run(
lamda_PLS=v2_PLS, Z=Z_VM, M=M)
np.savetxt("output/abnormal_p2_ez_all_run_q2.csv",
abnormal_p2_ez_all_run_q2,
delimiter=",",
fmt="%.4f")
np.savetxt("output/before_normal_p2_ez_all_run_q2.csv",
normal_p2_ez_all_run_q2,
delimiter=",",
fmt="%.4f")
fdh_graph.plt_show1(Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2],
'Process-2 Abnormal Case', '2')
fdh_graph.plt_show2_2(Z_VM * M,
normal_p2_ez_all_run_q2[:, 1:2],
abnormal_p2_ez_all_run_q2[:, 1:2],
'Process-2 Abnormal Case',
'2',
color1='bx-',
color2='rx--')
#fdh_graph.plt_show2_2(Z_VM * M, normal_p2_ez_all_run_q2[:, 0:1], abnormal_p2_ez_all_run_q2[:, 0:1], 'Process-2 Abnormal Case', color1='bx-', color2='rx--')
#fdh_graph.plt_show2(Z_VM, p2_ez_run_q2, p2_ez_run[:, 1:2], 'Process-2 Abnormal Case')
p2_q2_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p2_y_act[((z + 1) * M) - 1][1],
p2_y_prd[((z + 1) * M) - 1][1])
#print("으아아악 : ", p2_y_act[((z + 1) * M) - 1][1], p2_y_prd[((z + 1) * M) - 1][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
# p1_VM_Output : lambda_pls 반영 10 runs act결과 반영 vm
# p1_ACT_Output : lambda_pls 반영 10 runs act결과 반영
# p1_ez_run : prd - act (실제)
# p1_y_act : lambda_pls 반영 안한 act
# p1_y_prd : lambda_pls 반영 안하고 10 runs act결과 반영하지 않은 예측 값
p1_VM_Output, p1_ACT_Output, p1_ez_run, p1_y_act, p1_y_prd, p1_ez_all_run = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(0, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in range(Z_VM * M):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
fdh_graph.plt_show1(Z_VM * M, p1_y_act[:, 0:1], p1_y_prd[:, 0:1],
'Process-1 Normal Case', '1')
fdh_graph.plt_show2(Z_VM, p1_ez_run[:, 0:1], p1_ez_run[:, 1:2],
'Process-1 Normal Case')
# np.savetxt("output/p1_VM_Output.csv", p1_VM_Output, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_ACT_Output.csv", p1_ACT_Output, delimiter=",", fmt="%.4f")
np.savetxt("output/p1_ez_all_run_q1.csv",
p1_ez_all_run,
delimiter=",",
fmt="%.4f")
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd,
p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M,
f=p1_y_act) #DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd, p2_ez_all_run = fwc_p2_act.VM_Run(
lamda_PLS=v2_PLS, Z=Z_VM, M=M)
p2_q2_mape_Queue = []
for z in range(Z_VM):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p2_y_act[((z + 1) * M) - 1][1],
p2_y_prd[((z + 1) * M) - 1][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 Every Metrology MAPE : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
p2_q2_mape_Queue = []
for i in range(Z_VM * M):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 All MAPE : {0:.2f}%'.format(np.mean(p2_q2_mape_Queue)))
# for i in np.arange(22 * M, 34 * M, 1):
# mape = fdh_graph.mean_absolute_percentage_error(i + 1, p2_y_act[i][1], p2_y_prd[i][1])
# p2_mape_Queue.append(mape)
fdh_graph.plt_show1(Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2],
'Process-2 Normal Case', '2')
fdh_graph.plt_show2(Z_VM, p2_ez_run[:, 0:1], p2_ez_run[:, 1:2],
'Process-2 Normal Case')
# np.savetxt("output/p2_VM_Output.csv", p1_VM_Output, delimiter=",", fmt="%.4f")
# np.savetxt("output/p2_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
# np.savetxt("output/p2_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
# np.savetxt("output/p2_ACT_Output.csv", p1_ACT_Output, delimiter=",", fmt="%.4f")
np.savetxt("output/p2_ez_all_run_q2.csv",
p2_ez_all_run,
delimiter=",",
fmt="%.4f")
def main():
fdh_graph = FDC_Graph()
normal_p2_ez_all_run_q2 = np.loadtxt('output/p2_ez_all_run_q2.csv',
delimiter=",",
dtype=np.float32)
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd, _ = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_노이즈시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Error_VMresult, Error_ACTResult, ez_run, p1_y_act, p1_y_prd, _ = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
# fdh_graph.plt_show1(Z_VM * M, p1_y_act[:, 0:1], p1_y_prd[:, 0:1])
# fdh_graph.plt_show2(Z_VM, ez_run[:, 0:1], ez_run[:, 1:2])
# np.savetxt("output/Error_VMresult2.csv", Error_VMresult, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_ACTResult2.csv", Error_ACTResult, delimiter=",", fmt="%.4f")
# for z in np.arange(21, 32, 1):
# for k in np.arange(z * M, (z + 1) * M):
# Error_VMresult[k] = Error_VMresult[((z + 1) * M) - 1]
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for z in np.arange(0, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
# np.savetxt("output/p1_mape.csv", p1_q1_mape_Queue, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_VMresult1.csv", Error_VMresult, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_ACTResult.csv", Error_ACTResult, delimiter=",", fmt="%.4f")
for z in np.arange(Nz_RUN, Z_VM, 1):
if p1_q1_mape_Queue[z] >= Upstream_Rule_MAPE:
x1 = p1_y_act[(z * M) - 1]
x2 = p1_y_act[((z + 1) * M) - 1]
s = (x2 - x1) / M
for k in np.arange(z * M, (z + 1) * M - 1, 1):
i = k % (z * M) + 1
p1_y_prd[k] = x1 + s * i
#print('test : ', x1, x2, p1_y_prd[k])
#Error_VMresult[k] = Error_VMresult[((z + 1) * M) - 1]
# for z in np.arange(15, 40, 1):
# if p1_mape_Queue[z] >= 10:
# for k in np.arange(z * M, (z + 1) * M - 1):
# Error_VMresult[k] = Error_VMresult[((z + 1) * M) - 1]
np.savetxt("output/Error_VMresult_OK.csv",
p1_y_prd,
delimiter=",",
fmt="%.4f")
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd,
p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M,
f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd, reactive_p2_ez_all_run_q2 = fwc_p2_act.VM_Run(
lamda_PLS=v2_PLS, Z=Z_VM, M=M)
np.savetxt("output/reactive_p2_ez_all_run_q2.csv",
reactive_p2_ez_all_run_q2,
delimiter=",",
fmt="%.4f")
np.savetxt("output/after_normal_p2_ez_all_run_q2.csv",
normal_p2_ez_all_run_q2,
delimiter=",",
fmt="%.4f")
fdh_graph.plt_show1(Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2],
'Reactive Process-2 Event Processing Case', '2')
fdh_graph.plt_show2_2(Z_VM * M,
normal_p2_ez_all_run_q2[:, 1:2],
reactive_p2_ez_all_run_q2[:, 1:2],
'Reactive Process-2 Event Processing Case',
'2',
color1='bx-',
color2='rx--')
#fdh_graph.plt_show2(Z_VM, p2_ez_run[:, 0:1], p2_ez_run[:, 1:2])
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p2_y_act[((z + 1) * M) - 1][1],
p2_y_prd[((z + 1) * M) - 1][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
p2_q2_mape_Queue = []
np.savetxt("output/p2_mape.csv",
p2_q2_mape_Queue,
delimiter=",",
fmt="%.4f")
np.savetxt("output/p2_ACT_Output.csv",
Error_VMresult,
delimiter=",",
fmt="%.4f")
np.savetxt("output/p1_y_prd.csv", p2_y_prd, delimiter=",", fmt="%.4f")
np.savetxt("output/p1_y_act.csv", p2_y_act, delimiter=",", fmt="%.4f")
np.savetxt("output/p2_ACT_Output.csv",
Error_ACTResult,
delimiter=",",
fmt="%.4f")
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_DynamicSampling_노이즈시뮬레이터(A_p1, d_p1, C_p1, dM,
dStart, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) # DoE Run
VM_Output, ACT_Output, ez_run, p1_y_act, p1_y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=RUNS_CNT, M=M)
fdh_graph.plt_show1(Z_VM * M, p1_y_act[:, 0:1], p1_y_prd[:, 0:1])
ez_run_out = []
runM = M
ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, RUNS_CNT + 1):
if z == dStart:
runM = dM
for k in np.arange(z * runM, (z + 1) * runM):
ez_run_out.append(ez_run[z])
ez_run_out = np.array(ez_run_out)
fdh_graph.plt_show5(ez_run_out, N, M, dM, S1, Noise=True)
np.savetxt("output/Error_VMresult2.csv",
VM_Output,
delimiter=",",
fmt="%.4f")
np.savetxt("output/Error_ACTResult2.csv",
ACT_Output,
delimiter=",",
fmt="%.4f")
# for z in np.arange(21, 32, 1):
# for k in np.arange(z * M, (z + 1) * M):
# Error_VMresult[k] = Error_VMresult[((z + 1) * M) - 1]
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for z in np.arange(0, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
np.savetxt("output/p1_mape.csv",
p1_q1_mape_Queue,
delimiter=",",
fmt="%.4f")
np.savetxt("output/Error_VMresult1.csv",
VM_Output,
delimiter=",",
fmt="%.4f")
np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
np.savetxt("output/Error_ACTResult.csv",
ACT_Output,
delimiter=",",
fmt="%.4f")
for z in np.arange(Nz_RUN, Z_VM, 1):
if p1_q1_mape_Queue[z] >= Upstream_Rule_MAPE:
x1 = p1_y_act[(z * M) - 1]
x2 = p1_y_act[((z + 1) * M) - 1]
s = (x2 - x1) / M
act_q1 = np.abs(x2[0] - x1[0])
act_q2 = np.abs(x2[1] - x1[1])
if act_q1 <= 4 or act_q2 <= 4:
for k in np.arange(z * M, (z + 1) * M - 1, 1):
i = k % (z * M) + 1
#p1_y_prd[k] = x1 + s * i
p1_y_prd[k] = x2
#print('test : ', x1, x2, p1_y_prd[k])
#Error_VMresult[k] = Error_VMresult[((z + 1) * M) - 1]
# for z in np.arange(15, 40, 1):
# if p1_mape_Queue[z] >= 10:
# for k in np.arange(z * M, (z + 1) * M - 1):
# Error_VMresult[k] = Error_VMresult[((z + 1) * M) - 1]
np.savetxt("output/Error_VMresult_OK.csv",
p1_y_prd,
delimiter=",",
fmt="%.4f")
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd,
p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M,
f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd = fwc_p2_act.VM_Run(
lamda_PLS=v2_PLS, Z=Z_VM, M=M)
fdh_graph.plt_show1(Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2])
fdh_graph.plt_show2(Z_VM, p2_ez_run[:, 0:1], p2_ez_run[:, 1:2], Noise=True)
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p2_y_act[((z + 1) * M) - 1][1],
p2_y_prd[((z + 1) * M) - 1][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
p2_q2_mape_Queue = []
def main():
fdh_graph = FDC_Graph()
noise_ez_run = np.loadtxt('output/noise_ez_run.csv',
delimiter=",",
dtype=np.float32)
fwc_p1_vm = VM_Process1_DynamicSampling_노이즈시뮬레이터(A_p1, d_p1, C_p1, dM,
dStart, SEED)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) #DoE Run
VM_Output, ACT_Output, ez_run, y_act, y_prd, abnormal_p1_ez_all_run_q1 = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=RUNS_CNT, M=M)
# np.savetxt("output/VM_Output.csv", VM_Output, delimiter=",", fmt="%.4f")
# np.savetxt("output/ACT_Output.csv", ACT_Output, delimiter=",", fmt="%.4f")
# np.savetxt("output/ez_run.csv", ez_run, delimiter=",", fmt="%.4f")
# np.savetxt("output/y_act.csv", y_act, delimiter=",", fmt="%.4f")
# np.savetxt("output/y_prd.csv", y_prd, delimiter=",", fmt="%.4f")
fdh_graph.plt_show1(
N, y_act[:, 0:1], y_prd[:, 0:1], 'Process-1 Dynamic Sampling Case (' +
str(RUNS_CNT) + ' Metrology Runs)', '1')
#fdh_graph.plt_show2(RUNS_CNT, ez_run[:, 0:1], ez_run[:, 1:2], Noise=True)
ds_ez_run_out = []
noise_ez_run_out = []
runM = M
noise_ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, old_N + 1):
for k in np.arange(z * M, (z + 1) * M):
noise_ez_run_out.append(noise_ez_run[z])
noise_ez_run_out = np.array(noise_ez_run_out)
ds_ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, RUNS_CNT + 1):
if z == dStart:
runM = dM
for k in np.arange(z * runM, (z + 1) * runM):
ds_ez_run_out.append(ez_run[z])
ez_run_out = np.array(ds_ez_run_out)
#np.savetxt("output/ez_run2.csv", ez_run_out, delimiter=",", fmt="%.4f")
#fdh_graph.plt_show5_1(noise_ez_run_out, ez_run_out, N, M, dM, S1, type=1) #60 Metrology Runs
fdh_graph.plt_show5_1(noise_ez_run_out, ez_run_out, N, M, dM, S1,
type=2) # 100 Metrology Runs
# p1_q1_mape_Queue = []
# #metrology 마다 보여주는 MAPE 값이 의미가 없다. ==> 로직 고쳐야 함.. 40 runs가 아니라 60 Runs, 80 Runs 변경 계산 필요
# for z in np.arange(Nz_RUN, old_N, 1):
# mape = fdh_graph.mean_absolute_percentage_error(z + 1, y_act[((z + 1) * M) - 1][0], y_prd[((z + 1) * M) - 1][0])
# p1_q1_mape_Queue.append(mape)
#
# print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(S1 * M, N, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, y_act[i][0], y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 20 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_노이즈시뮬레이터(A_p1, d_p1, C_p1, SEED)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Error_VMresult, Error_ACTResult, ez_run, p1_y_act, p1_y_prd = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
# fdh_graph.plt_show1(Z_VM * M, p1_y_act[:, 0:1], p1_y_prd[:, 0:1])
# fdh_graph.plt_show2(Z_VM, ez_run[:, 0:1], ez_run[:, 1:2])
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
np.savetxt("output/p1_mape.csv",
p1_q1_mape_Queue,
delimiter=",",
fmt="%.4f")
np.savetxt("output/Error_VMresult1.csv",
Error_VMresult,
delimiter=",",
fmt="%.4f")
np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
np.savetxt("output/Error_ACTResult.csv",
Error_ACTResult,
delimiter=",",
fmt="%.4f")
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd,
p1_y_act, SEED)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M,
f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd = fwc_p2_act.VM_Run(
lamda_PLS=v2_PLS, Z=Z_VM, M=M)
fdh_graph.plt_show1(Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2])
fdh_graph.plt_show2(Z_VM, p2_ez_run[:, 0:1], p2_ez_run[:, 1:2])
p2_q2_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p2_y_act[((z + 1) * M) - 1][1],
p2_y_prd[((z + 1) * M) - 1][1])
#print("으아아악 : ", p2_y_act[((z + 1) * M) - 1][1], p2_y_prd[((z + 1) * M) - 1][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
np.savetxt("output/p2_mape.csv",
p2_q2_mape_Queue,
delimiter=",",
fmt="%.4f")
np.savetxt("output/p2_ACT_Output.csv",
Error_VMresult,
delimiter=",",
fmt="%.4f")
np.savetxt("output/p1_y_prd.csv", p2_y_prd, delimiter=",", fmt="%.4f")
np.savetxt("output/p1_y_act.csv", p2_y_act, delimiter=",", fmt="%.4f")
np.savetxt("output/p2_ACT_Output.csv",
Error_ACTResult,
delimiter=",",
fmt="%.4f")
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd, _ = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_노이즈시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Error_VMresult, Error_ACTResult, ez_run, p1_y_act, p1_y_prd, _ = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
# fdh_graph.plt_show1(Z_VM * M, p1_y_act[:, 0:1], p1_y_prd[:, 0:1])
# fdh_graph.plt_show2(Z_VM, ez_run[:, 0:1], ez_run[:, 1:2])
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for z in np.arange(0, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
for z in np.arange(Nz_RUN, Z_VM, 1):
if p1_q1_mape_Queue[z] >= Upstream_Rule_MAPE:
x1 = p1_y_act[(z * M) - 1]
x2 = p1_y_act[((z + 1) * M) - 1]
s = (x2 - x1) / M
for k in np.arange(z * M, (z + 1) * M - 1, 1):
i = k % (z * M) + 1
p1_y_prd[k] = x1 + s * i
# np.savetxt("output/p1_mape.csv", p1_q1_mape_Queue, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_VMresult1.csv", Error_VMresult, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_ACTResult.csv", Error_ACTResult, delimiter=",", fmt="%.4f")
max_mape = 0
min_mape = 1000000000000
t = 0
for m in range(1000):
SEED2 = t + m
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS,
p1_y_prd, p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M,
f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd, _ = fwc_p2_act.VM_Run(
lamda_PLS=v2_PLS, Z=Z_VM, M=M)
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
mape = np.mean(p2_q2_mape_Queue)
print('SEED2 : ', SEED2,
'Process-2 q2 All MAPE After 15 Lot : {0:.2f}%'.format(mape))
if mape > max_mape:
max_mape = mape
seed_right = SEED2
print('SEED2 : ', SEED2, ' , max_mape = ', max_mape)
if mape < min_mape:
min_mape = mape
seed_right1 = SEED2
print('SEED2 : ', SEED2, ' , min_mape = ', min_mape)
print('seed_right : ', seed_right, ' , max_mape = ', max_mape)
print('seed_right1 : ', seed_right1, ' , min_mape = ', min_mape)
def main():
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd = fwc_p1_vm.VM_Run(lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_노이즈시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Error_VMresult, Error_ACTResult, ez_run, p1_y_act, p1_y_prd = fwc_p1_vm.VM_Run(lamda_PLS=v1_PLS, Z=Z_VM, M=M)
np.savetxt("output/ez_run.csv", ez_run, delimiter=",", fmt="%.4f")
# fdh_graph.plt_show1(Z_VM * M, p1_y_act[:, 0:1], p1_y_prd[:, 0:1])
# fdh_graph.plt_show2(Z_VM, ez_run[:, 0:1], ez_run[:, 1:2], Noise=True)
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN_1, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(z + 1, p1_y_act[((z + 1) * M) - 1][0], p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
np.savetxt("output/p1_mape.csv", p1_q1_mape_Queue, delimiter=",", fmt="%.4f")
np.savetxt("output/Error_VMresult1.csv", Error_VMresult, delimiter=",", fmt="%.4f")
np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
np.savetxt("output/Error_ACTResult.csv", Error_ACTResult, delimiter=",", fmt="%.4f")
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN_1 * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
#print('Process-1 q1 All MAPE After 20 Lot : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(np.mean(p1_q1_mape_Queue)))
# fdh_graph.plt_show1(Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2])
# fdh_graph.plt_show2(Z_VM, p2_ez_run[:, 0:1], p2_ez_run[:, 1:2], Noise=False)
max_mape = 0
t = 0
for m in range(1000):
SEED2 = t + m
fwc_p2_act = VM_Process2_노이즈시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd, p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M, f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd = fwc_p2_act.VM_Run(lamda_PLS=v2_PLS, Z=Z_VM, M=M)
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN_2 * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
mape = np.mean(p2_q2_mape_Queue)
print('Process-2 q2 All MAPE After 20 Lot : {0:.2f}%'.format(mape))
if mape > max_mape:
max_mape = mape
seed_right = SEED2
print('SEED2 : ', SEED2, ' , max_mape = ', max_mape)
print('seed_right : ', seed_right, ' , max_mape = ', max_mape)
def main():
noise_ez_run = np.loadtxt('output/noise_ez_run.csv',
delimiter=",",
dtype=np.float32)
reactive_p2_ez_all_run_q2 = np.loadtxt(
'D:/10. 대학원/04. Source/09. VM_Source/04. VMOnly/output/reactive_p2_ez_all_run_q2.csv',
delimiter=",",
dtype=np.float32)
after_normal_p2_ez_all_run_q2 = np.loadtxt(
'D:/10. 대학원/04. Source/09. VM_Source/04. VMOnly/output/after_normal_p2_ez_all_run_q2.csv',
delimiter=",",
dtype=np.float32)
fdh_graph = FDC_Graph()
fwc_p1_vm = VM_Process1_시뮬레이터(A_p1, d_p1, C_p1, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v1_PLS, Z=Z_DoE, M=M) # DoE Run
Normal_VMResult, Normal_ACTResult, ez_run, o_y_act, o_y_prd, _ = fwc_p1_vm.VM_Run(
lamda_PLS=v1_PLS, Z=Z_VM, M=M)
fwc_p1_vm = VM_Process1_DynamicSampling_노이즈시뮬레이터(A_p1, d_p1, C_p1, dM,
dStart, SEED1)
fwc_p1_vm.DoE_Run(lamda_PLS=v_PLS, Z=Z_DoE, M=M) # DoE Run
VM_Output, ACT_Output, ez_run, p1_y_act, p1_y_prd, p1_ez_all_run_q1 = fwc_p1_vm.VM_Run(
lamda_PLS=v_PLS, Z=RUNS_CNT, M=M)
#fdh_graph.plt_show1(Z_VM * M, p1_y_act[:, 0:1], p1_y_prd[:, 0:1])
ez_run_out = []
noise_ez_run_out = []
runM = M
noise_ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, old_N + 1):
for k in np.arange(z * M, (z + 1) * M):
noise_ez_run_out.append(noise_ez_run[z])
noise_ez_run_out = np.array(noise_ez_run_out)
ez_run_out.append(np.array([0, 0]))
for z in np.arange(1, RUNS_CNT + 1):
if z == dStart:
runM = dM
for k in np.arange(z * runM, (z + 1) * runM):
ez_run_out.append(ez_run[z])
ez_run_out = np.array(ez_run_out)
#fdh_graph.plt_show5(ez_run_out, N, M, dM, S1, Noise=True)
##fdh_graph.plt_show5_1(noise_ez_run_out, ez_run_out, N, M, dM, S1, type=1)
# np.savetxt("output/Error_VMresult2.csv", VM_Output, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_ACTResult2.csv", ACT_Output, delimiter=",", fmt="%.4f")
# for z in np.arange(21, 32, 1):
# for k in np.arange(z * M, (z + 1) * M):
# Error_VMresult[k] = Error_VMresult[((z + 1) * M) - 1]
p1_q1_mape_Queue = []
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p1_y_act[i][0], p1_y_prd[i][0])
p1_q1_mape_Queue.append(mape)
print('Process-1 q1 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p1_q1_mape_Queue)))
p1_q1_mape_Queue = []
for z in np.arange(0, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p1_y_act[((z + 1) * M) - 1][0],
p1_y_prd[((z + 1) * M) - 1][0])
p1_q1_mape_Queue.append(mape)
# np.savetxt("output/p1_mape.csv", p1_q1_mape_Queue, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_VMresult1.csv", VM_Output, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_prd.csv", p1_y_prd, delimiter=",", fmt="%.4f")
# np.savetxt("output/p1_y_act.csv", p1_y_act, delimiter=",", fmt="%.4f")
# np.savetxt("output/Error_ACTResult.csv", ACT_Output, delimiter=",", fmt="%.4f")
for z in np.arange(Nz_RUN, Z_VM, 1):
if p1_q1_mape_Queue[z] >= Upstream_Rule_MAPE:
x1 = p1_y_act[(z * M) - 1]
x2 = p1_y_act[((z + 1) * M) - 1]
s = (x2 - x1) / M
# act_q1 = np.abs(x2[0] - x1[0])
# act_q2 = np.abs(x2[1] - x1[1])
# if act_q1 <= 4 or act_q2 <= 4:
for k in np.arange(z * M, (z + 1) * M - 1, 1):
i = k % (z * M) + 1
p1_y_prd[k] = x1 + s * i
#p1_y_prd[k] = x2
#print('test : ', x1, x2, p1_y_prd[k])
#Error_VMresult[k] = Error_VMresult[((z + 1) * M) - 1]
# for z in np.arange(15, 40, 1):
# if p1_mape_Queue[z] >= 10:
# for k in np.arange(z * M, (z + 1) * M - 1):
# Error_VMresult[k] = Error_VMresult[((z + 1) * M) - 1]
np.savetxt("output/Error_VMresult_OK.csv",
p1_y_prd,
delimiter=",",
fmt="%.4f")
fwc_p2_act = VM_Process2_시뮬레이터(A_p2, d_p2, C_p2, F_p2, v1_PLS, p1_y_prd,
p1_y_act, SEED2)
fwc_p2_act.DoE_Run(lamda_PLS=v2_PLS, Z=Z_DoE, M=M,
f=o_y_act) # DoE Run ACT값으로 가능
p2_VM_Output, p2_ACT_Output, p2_ez_run, p2_y_act, p2_y_prd, p2_ez_all_run_q2 = fwc_p2_act.VM_Run(
lamda_PLS=v2_PLS, Z=Z_VM, M=M)
#fdh_graph.plt_show1(Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2])
#fdh_graph.plt_show2(Z_VM, p2_ez_run[:, 0:1], p2_ez_run[:, 1:2], Noise=True)
fdh_graph.plt_show1(
Z_VM * M, p2_y_act[:, 1:2], p2_y_prd[:, 1:2],
'Process-2 UpStream Rule Case with Process-1 Dynamic Sampling', '2')
fdh_graph.plt_show2_4(
Z_VM * M,
reactive_p2_ez_all_run_q2[:, 1:2],
p2_ez_all_run_q2[:, 1:2],
'Process-2 UPStream Rule X with Process-1 Dynamic Sampling',
'2',
color1='bx-',
color2='rx--')
p2_q2_mape_Queue = []
for i in np.arange(Nz_RUN * M, Z_VM * M, 1):
mape = fdh_graph.mean_absolute_percentage_error(
i + 1, p2_y_act[i][1], p2_y_prd[i][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 All MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
# metrology 마다 보여주는 MAPE 값이 의미가 없다.
for z in np.arange(Nz_RUN, Z_VM, 1):
mape = fdh_graph.mean_absolute_percentage_error(
z + 1, p2_y_act[((z + 1) * M) - 1][1],
p2_y_prd[((z + 1) * M) - 1][1])
p2_q2_mape_Queue.append(mape)
print('Process-2 q2 Every Metrology MAPE After 15 Lot : {0:.2f}%'.format(
np.mean(p2_q2_mape_Queue)))
p2_q2_mape_Queue = []