Python环境下基于深度判别迁移学习网络的轴承故障诊断
目前很多机器学习和数据挖掘算法都是基于训练数据和测试数据位于同一特征空间、拥有相同数据分布的假设。然而在现实应用中，该假设却未必存在。一方面，如果将利用某一领域数据训练得到的模型直接应用于新的目标领域，领域之间切实存在的数据差异可能会导致模型效果的骤然下降。另一方面，如果直接在新的目标领域中进行模型的训练，其数据的稀缺和标注的不完整可能会导致监督学习出现严重的过拟合问题，难以达到令人满意的学习效果。因此，如何对不同领域、不同来源的非结构化数据进行合理的数据迁移，实现不同领域的模型适配，成为一个亟需解决的问题。
迁移学习可以从辅助领域的现有数据中迁移知识，帮助完成目标领域的学习任务，即完成从源域到目标域的知识迁移过程，从而有效解决上述问题。特别地，随着大数据时代来临，迁移学习能够将知识从“大数据”迁移到“小数据”，解决小数据领域的知识稀缺等问题。
本代码为Python环境下基于深度判别迁移学习网络的轴承故障诊断代码，源域数据为西储大学轴承数据48kcwru_data.npy，目标域数据为为江南大学轴承数据jnudata600_data.npy，所用模块版本如下：
numpy==1.21.5
sklearn==1.0.2
pytorch_lightning==1.7.7
torch==1.10.1+cpu所用模块如下：
import numpy as np
from sklearn.preprocessing import StandardScaler
from pytorch_lightning.utilities.seed import seed_everything
import torch
import torch.nn as nn
import torch.nn.functional as F
import argparse
from sklearn.utils import shuffle
from torch.utils import data as da
from torchmetrics import MeanMetric, Accuracy部分代码如下：
#定义加载数据函数
def load_data():
source_data = np.load(args.cwru_data)
source_label = np.load(args.cwru_label).argmax(axis=-1)
target_data = np.load(args.jnu_data)
target_label = np.load(args.jnu_label).argmax(axis=-1)
source_data = StandardScaler().fit_transform(source_data.T).T
target_data = StandardScaler().fit_transform(target_data.T).T
source_data = np.expand_dims(source_data, axis=1)
target_data = np.expand_dims(target_data, axis=1)
source_data, source_label = shuffle(source_data, source_label, random_state=2)
target_data, target_label = shuffle(target_data, target_label, random_state=2)
Train_source = Dataset(source_data, source_label)
Train_target = Dataset(target_data, target_label)
return Train_source, Train_target
###############################################################
#最大均值差异MMD类
class MMD(nn.Module):
def __init__(self, m, n):
super(MMD, self).__init__()
self.m = m
self.n = n
def _mix_rbf_mmd2(self, X, Y, sigmas=(10,), wts=None, biased=True):
K_XX, K_XY, K_YY, d = self._mix_rbf_kernel(X, Y, sigmas, wts)
return self._mmd2(K_XX, K_XY, K_YY, const_diagonal=d, biased=biased)
def _mix_rbf_kernel(self, X, Y, sigmas, wts=None):
if wts is None:
wts = [1] * len(sigmas)
XX = torch.matmul(X, X.t())
XY = torch.matmul(X, Y.t())
YY = torch.matmul(Y, Y.t())
X_sqnorms = torch.diagonal(XX)
Y_sqnorms = torch.diagonal(YY)
r = lambda x: torch.unsqueeze(x, 0)
c = lambda x: torch.unsqueeze(x, 1)
K_XX, K_XY, K_YY = 0., 0., 0.
for sigma, wt in zip(sigmas, wts):
gamma = 1 / (2 * sigma ** 2)
K_XX += wt * torch.exp(-gamma * (-2 * XX + c(X_sqnorms) + r(X_sqnorms)))
K_XY += wt * torch.exp(-gamma * (-2 * XY + c(X_sqnorms) + r(Y_sqnorms)))
K_YY += wt * torch.exp(-gamma * (-2 * YY + c(Y_sqnorms) + r(Y_sqnorms)))
return K_XX, K_XY, K_YY, torch.sum(torch.tensor(wts))运行结果如下
Epoch1, train_loss is 2.03397,test_loss is 4.93007, train_accuracy is 0.44475,test_accuracy is 0.18675,train_all_loss is 41.71445,target_cla_loss is 1.61769,source_cla_loss is 3.70468,cda_loss is 6.74935,mda_loss is 37.17306
Epoch2, train_loss is 0.54279,test_loss is 6.41293, train_accuracy is 0.83325,test_accuracy is 0.20800,train_all_loss is 6.87837,target_cla_loss is 1.71905,source_cla_loss is 1.94874,cda_loss is 1.58677,mda_loss is 4.59905
Epoch3, train_loss is 0.18851,test_loss is 5.60176, train_accuracy is 0.93775,test_accuracy is 0.29850,train_all_loss is 5.26101,target_cla_loss is 0.66165,source_cla_loss is 0.54253,cda_loss is 1.02360,mda_loss is 4.54996
Epoch4, train_loss is 0.14104,test_loss is 4.58690, train_accuracy is 0.94850,test_accuracy is 0.30800,train_all_loss is 4.09870,target_cla_loss is 0.54025,source_cla_loss is 0.38254,cda_loss is 0.88701,mda_loss is 3.57343
Epoch5, train_loss is 0.11775,test_loss is 5.07279, train_accuracy is 0.95900,test_accuracy is 0.28300,train_all_loss is 3.27498,target_cla_loss is 0.52239,source_cla_loss is 0.29470,cda_loss is 0.87684,mda_loss is 2.84035
Epoch6, train_loss is 0.08998,test_loss is 5.02790, train_accuracy is 0.97300,test_accuracy is 0.29825,train_all_loss is 3.21299,target_cla_loss is 0.39788,source_cla_loss is 0.21586,cda_loss is 0.76452,mda_loss is 2.88089
Epoch7, train_loss is 0.07695,test_loss is 4.51329, train_accuracy is 0.97975,test_accuracy is 0.31800,train_all_loss is 2.92297,target_cla_loss is 0.40623,source_cla_loss is 0.16808,cda_loss is 0.76679,mda_loss is 2.63759
Epoch8, train_loss is 0.07421,test_loss is 4.26603, train_accuracy is 0.97900,test_accuracy is 0.32400,train_all_loss is 2.65909,target_cla_loss is 0.39052,source_cla_loss is 0.17180,cda_loss is 0.72834,mda_loss is 2.37541
Epoch9, train_loss is 0.05102,test_loss is 3.63495, train_accuracy is 0.98950,test_accuracy is 0.35800,train_all_loss is 2.62192,target_cla_loss is 0.41614,source_cla_loss is 0.11351,cda_loss is 0.77280,mda_loss is 2.38951
Epoch10, train_loss is 0.06574,test_loss is 3.52261, train_accuracy is 0.98850,test_accuracy is 0.33100,train_all_loss is 2.59112,target_cla_loss is 0.48041,source_cla_loss is 0.14138,cda_loss is 0.73876,mda_loss is 2.32783
Epoch11, train_loss is 0.05876,test_loss is 3.86388, train_accuracy is 0.99050,test_accuracy is 0.33475,train_all_loss is 2.45829,target_cla_loss is 0.45235,source_cla_loss is 0.11720,cda_loss is 0.73355,mda_loss is 2.22250
Epoch12, train_loss is 0.05688,test_loss is 3.82415, train_accuracy is 0.99350,test_accuracy is 0.32075,train_all_loss is 2.43463,target_cla_loss is 0.41805,source_cla_loss is 0.10393,cda_loss is 0.68400,mda_loss is 2.22049
Epoch13, train_loss is 0.05224,test_loss is 3.78473, train_accuracy is 0.99300,test_accuracy is 0.32225,train_all_loss is 2.26402,target_cla_loss is 0.42708,source_cla_loss is 0.09958,cda_loss is 0.68476,mda_loss is 2.05325
Epoch14, train_loss is 0.06636,test_loss is 3.89151, train_accuracy is 0.98675,test_accuracy is 0.32200,train_all_loss is 2.42129,target_cla_loss is 0.42966,source_cla_loss is 0.12657,cda_loss is 0.66312,mda_loss is 2.18544
Epoch15, train_loss is 0.05342,test_loss is 3.78424, train_accuracy is 0.99575,test_accuracy is 0.32200,train_all_loss is 2.33275,target_cla_loss is 0.42920,source_cla_loss is 0.10599,cda_loss is 0.64312,mda_loss is 2.11953
Epoch16, train_loss is 0.04968,test_loss is 3.67101, train_accuracy is 0.99750,test_accuracy is 0.32100,train_all_loss is 2.23092,target_cla_loss is 0.43684,source_cla_loss is 0.09945,cda_loss is 0.63847,mda_loss is 2.02393
Epoch17, train_loss is 0.05957,test_loss is 3.67722, train_accuracy is 0.99525,test_accuracy is 0.33000,train_all_loss is 2.27638,target_cla_loss is 0.45589,source_cla_loss is 0.12059,cda_loss is 0.63428,mda_loss is 2.04677
Epoch18, train_loss is 0.05812,test_loss is 3.59771, train_accuracy is 0.99600,test_accuracy is 0.32350,train_all_loss is 2.23080,target_cla_loss is 0.47418,source_cla_loss is 0.11620,cda_loss is 0.61432,mda_loss is 2.00575
Epoch19, train_loss is 0.06287,test_loss is 3.43253, train_accuracy is 0.99350,test_accuracy is 0.32950,train_all_loss is 2.40593,target_cla_loss is 0.48567,source_cla_loss is 0.12299,cda_loss is 0.64320,mda_loss is 2.17005
Epoch20, train_loss is 0.06316,test_loss is 3.51056, train_accuracy is 0.99575,test_accuracy is 0.32475,train_all_loss is 2.31846,target_cla_loss is 0.46683,source_cla_loss is 0.13138,cda_loss is 0.63335,mda_loss is 2.07705
Epoch21, train_loss is 0.05934,test_loss is 3.84920, train_accuracy is 0.99425,test_accuracy is 0.32475,train_all_loss is 2.24664,target_cla_loss is 0.43869,source_cla_loss is 0.11981,cda_loss is 0.62297,mda_loss is 2.02066
Epoch22, train_loss is 0.05423,test_loss is 3.69176, train_accuracy is 0.99500,test_accuracy is 0.34025,train_all_loss is 2.28318,target_cla_loss is 0.46334,source_cla_loss is 0.11229,cda_loss is 0.64396,mda_loss is 2.06016
Epoch23, train_loss is 0.04934,test_loss is 3.35009, train_accuracy is 0.99775,test_accuracy is 0.33525,train_all_loss is 2.29074,target_cla_loss is 0.46118,source_cla_loss is 0.10556,cda_loss is 0.63410,mda_loss is 2.07565
Epoch24, train_loss is 0.05903,test_loss is 3.62366, train_accuracy is 0.99275,test_accuracy is 0.33475,train_all_loss is 2.19408,target_cla_loss is 0.42396,source_cla_loss is 0.12135,cda_loss is 0.61378,mda_loss is 1.96895
Epoch25, train_loss is 0.06076,test_loss is 3.72236, train_accuracy is 0.99475,test_accuracy is 0.33100,train_all_loss is 2.17257,target_cla_loss is 0.41436,source_cla_loss is 0.12382,cda_loss is 0.60274,mda_loss is 1.94704
Epoch26, train_loss is 0.05901,test_loss is 3.59237, train_accuracy is 0.99600,test_accuracy is 0.33225,train_all_loss is 2.34868,target_cla_loss is 0.47314,source_cla_loss is 0.12100,cda_loss is 0.60417,mda_loss is 2.11995
Epoch27, train_loss is 0.05911,test_loss is 3.82265, train_accuracy is 0.99500,test_accuracy is 0.34000,train_all_loss is 2.25363,target_cla_loss is 0.43506,source_cla_loss is 0.12174,cda_loss is 0.58978,mda_loss is 2.02940
Epoch28, train_loss is 0.05933,test_loss is 3.65749, train_accuracy is 0.99750,test_accuracy is 0.34050,train_all_loss is 2.22602,target_cla_loss is 0.47101,source_cla_loss is 0.12723,cda_loss is 0.61243,mda_loss is 1.99045
Epoch29, train_loss is 0.03960,test_loss is 3.72375, train_accuracy is 0.99900,test_accuracy is 0.34075,train_all_loss is 2.20123,target_cla_loss is 0.44257,source_cla_loss is 0.10232,cda_loss is 0.60967,mda_loss is 1.99369
Epoch30, train_loss is 0.05384,test_loss is 3.61430, train_accuracy is 0.99350,test_accuracy is 0.33100,train_all_loss is 2.12823,target_cla_loss is 0.43999,source_cla_loss is 0.11275,cda_loss is 0.60491,mda_loss is 1.91099
Epoch31, train_loss is 0.04751,test_loss is 3.67043, train_accuracy is 0.99650,test_accuracy is 0.35875,train_all_loss is 2.09769,target_cla_loss is 0.39988,source_cla_loss is 0.11007,cda_loss is 0.61966,mda_loss is 1.88566
Epoch32, train_loss is 0.05494,test_loss is 3.66357, train_accuracy is 0.99325,test_accuracy is 0.35325,train_all_loss is 2.16350,target_cla_loss is 0.42613,source_cla_loss is 0.12841,cda_loss is 0.61385,mda_loss is 1.93109
Epoch33, train_loss is 0.04867,test_loss is 3.86881, train_accuracy is 0.99600,test_accuracy is 0.34925,train_all_loss is 2.09730,target_cla_loss is 0.43453,source_cla_loss is 0.11316,cda_loss is 0.60486,mda_loss is 1.88020
Epoch34, train_loss is 0.04144,test_loss is 3.81459, train_accuracy is 0.99775,test_accuracy is 0.34900,train_all_loss is 2.03613,target_cla_loss is 0.43036,source_cla_loss is 0.09037,cda_loss is 0.58916,mda_loss is 1.84382
Epoch35, train_loss is 0.04441,test_loss is 3.66703, train_accuracy is 0.99775,test_accuracy is 0.35975,train_all_loss is 2.19538,target_cla_loss is 0.42232,source_cla_loss is 0.09971,cda_loss is 0.58817,mda_loss is 1.99462
Epoch36, train_loss is 0.05367,test_loss is 3.85576, train_accuracy is 0.99750,test_accuracy is 0.34825,train_all_loss is 2.16997,target_cla_loss is 0.42345,source_cla_loss is 0.12474,cda_loss is 0.59072,mda_loss is 1.94381
Epoch37, train_loss is 0.04486,test_loss is 3.92458, train_accuracy is 0.99500,test_accuracy is 0.34375,train_all_loss is 2.10291,target_cla_loss is 0.42692,source_cla_loss is 0.10405,cda_loss is 0.61382,mda_loss is 1.89478
Epoch38, train_loss is 0.04253,test_loss is 3.61390, train_accuracy is 0.99775,test_accuracy is 0.35300,train_all_loss is 1.99870,target_cla_loss is 0.41496,source_cla_loss is 0.10335,cda_loss is 0.61797,mda_loss is 1.79206
Epoch39, train_loss is 0.04100,test_loss is 3.58168, train_accuracy is 0.99825,test_accuracy is 0.35125,train_all_loss is 2.21055,target_cla_loss is 0.43278,source_cla_loss is 0.09059,cda_loss is 0.58880,mda_loss is 2.01780
Epoch40, train_loss is 0.04859,test_loss is 3.62033, train_accuracy is 0.99700,test_accuracy is 0.35875,train_all_loss is 2.10599,target_cla_loss is 0.39430,source_cla_loss is 0.10756,cda_loss is 0.59355,mda_loss is 1.89964
Epoch41, train_loss is 0.05128,test_loss is 3.67334, train_accuracy is 0.99550,test_accuracy is 0.35000,train_all_loss is 2.04580,target_cla_loss is 0.41525,source_cla_loss is 0.11839,cda_loss is 0.60271,mda_loss is 1.82561
Epoch42, train_loss is 0.05321,test_loss is 3.69000, train_accuracy is 0.99450,test_accuracy is 0.34000,train_all_loss is 2.08382,target_cla_loss is 0.37200,source_cla_loss is 0.10866,cda_loss is 0.57336,mda_loss is 1.88063
Epoch43, train_loss is 0.04858,test_loss is 3.74816, train_accuracy is 0.99575,test_accuracy is 0.34375,train_all_loss is 2.07542,target_cla_loss is 0.42002,source_cla_loss is 0.10167,cda_loss is 0.57270,mda_loss is 1.87447
Epoch44, train_loss is 0.04634,test_loss is 3.87925, train_accuracy is 0.99550,test_accuracy is 0.34175,train_all_loss is 2.08382,target_cla_loss is 0.37761,source_cla_loss is 0.10194,cda_loss is 0.58080,mda_loss is 1.88604
Epoch45, train_loss is 0.05111,test_loss is 3.70675, train_accuracy is 0.99375,test_accuracy is 0.35175,train_all_loss is 2.13403,target_cla_loss is 0.39479,source_cla_loss is 0.11404,cda_loss is 0.58542,mda_loss is 1.92198
Epoch46, train_loss is 0.05028,test_loss is 3.58449, train_accuracy is 0.99525,test_accuracy is 0.35550,train_all_loss is 2.08147,target_cla_loss is 0.41587,source_cla_loss is 0.10975,cda_loss is 0.57340,mda_loss is 1.87279
Epoch47, train_loss is 0.04632,test_loss is 3.54860, train_accuracy is 0.99750,test_accuracy is 0.35175,train_all_loss is 2.09232,target_cla_loss is 0.41676,source_cla_loss is 0.10310,cda_loss is 0.56944,mda_loss is 1.89061
Epoch48, train_loss is 0.05131,test_loss is 3.71509, train_accuracy is 0.99600,test_accuracy is 0.34425,train_all_loss is 2.03649,target_cla_loss is 0.37779,source_cla_loss is 0.11077,cda_loss is 0.57039,mda_loss is 1.83090
Epoch49, train_loss is 0.04696,test_loss is 3.79712, train_accuracy is 0.99675,test_accuracy is 0.36775,train_all_loss is 1.96412,target_cla_loss is 0.38354,source_cla_loss is 0.11067,cda_loss is 0.59040,mda_loss is 1.75606
Epoch50, train_loss is 0.04041,test_loss is 3.61169, train_accuracy is 0.99750,test_accuracy is 0.36175,train_all_loss is 2.00069,target_cla_loss is 0.37321,source_cla_loss is 0.09218,cda_loss is 0.58409,mda_loss is 1.81278
Epoch51, train_loss is 0.04318,test_loss is 3.74727, train_accuracy is 0.99750,test_accuracy is 0.35675,train_all_loss is 2.00346,target_cla_loss is 0.35602,source_cla_loss is 0.09815,cda_loss is 0.57770,mda_loss is 1.81194
Epoch52, train_loss is 0.04124,test_loss is 3.54835, train_accuracy is 0.99700,test_accuracy is 0.35150,train_all_loss is 2.03243,target_cla_loss is 0.36050,source_cla_loss is 0.09370,cda_loss is 0.57510,mda_loss is 1.84517
Epoch53, train_loss is 0.04390,test_loss is 3.82567, train_accuracy is 0.99725,test_accuracy is 0.35275,train_all_loss is 1.99703,target_cla_loss is 0.34785,source_cla_loss is 0.10884,cda_loss is 0.56020,mda_loss is 1.79739
Epoch54, train_loss is 0.04348,test_loss is 3.86695, train_accuracy is 0.99750,test_accuracy is 0.35850,train_all_loss is 1.94795,target_cla_loss is 0.38242,source_cla_loss is 0.10408,cda_loss is 0.56894,mda_loss is 1.74873
Epoch55, train_loss is 0.03718,test_loss is 3.56571, train_accuracy is 0.99775,test_accuracy is 0.36575,train_all_loss is 1.97170,target_cla_loss is 0.37277,source_cla_loss is 0.08584,cda_loss is 0.56435,mda_loss is 1.79215
Epoch56, train_loss is 0.04152,test_loss is 3.39552, train_accuracy is 0.99850,test_accuracy is 0.36575,train_all_loss is 2.02024,target_cla_loss is 0.36317,source_cla_loss is 0.10414,cda_loss is 0.58045,mda_loss is 1.82174
Epoch57, train_loss is 0.03893,test_loss is 3.68062, train_accuracy is 0.99875,test_accuracy is 0.35975,train_all_loss is 1.97693,target_cla_loss is 0.34846,source_cla_loss is 0.09432,cda_loss is 0.57112,mda_loss is 1.79064
Epoch58, train_loss is 0.04239,test_loss is 3.59319, train_accuracy is 0.99750,test_accuracy is 0.36800,train_all_loss is 1.99725,target_cla_loss is 0.37591,source_cla_loss is 0.10367,cda_loss is 0.56670,mda_loss is 1.79933
Epoch59, train_loss is 0.04245,test_loss is 3.66916, train_accuracy is 0.99800,test_accuracy is 0.35800,train_all_loss is 1.99187,target_cla_loss is 0.34185,source_cla_loss is 0.10355,cda_loss is 0.57694,mda_loss is 1.79644
Epoch60, train_loss is 0.04063,test_loss is 3.69465, train_accuracy is 0.99700,test_accuracy is 0.35850,train_all_loss is 1.92894,target_cla_loss is 0.34705,source_cla_loss is 0.09351,cda_loss is 0.56955,mda_loss is 1.74377
Epoch61, train_loss is 0.04399,test_loss is 3.56587, train_accuracy is 0.99650,test_accuracy is 0.35900,train_all_loss is 2.02587,target_cla_loss is 0.37329,source_cla_loss is 0.09890,cda_loss is 0.57672,mda_loss is 1.83196
Epoch62, train_loss is 0.03630,test_loss is 3.75333, train_accuracy is 0.99850,test_accuracy is 0.36125,train_all_loss is 2.06062,target_cla_loss is 0.33665,source_cla_loss is 0.09386,cda_loss is 0.58990,mda_loss is 1.87410
Epoch63, train_loss is 0.04550,test_loss is 3.58529, train_accuracy is 0.99850,test_accuracy is 0.35700,train_all_loss is 2.03326,target_cla_loss is 0.37309,source_cla_loss is 0.11314,cda_loss is 0.55576,mda_loss is 1.82723
Epoch64, train_loss is 0.03636,test_loss is 3.52662, train_accuracy is 0.99900,test_accuracy is 0.36075,train_all_loss is 1.94026,target_cla_loss is 0.39102,source_cla_loss is 0.09922,cda_loss is 0.58333,mda_loss is 1.74360
Epoch65, train_loss is 0.03628,test_loss is 3.86440, train_accuracy is 0.99850,test_accuracy is 0.35875,train_all_loss is 1.99657,target_cla_loss is 0.32306,source_cla_loss is 0.09738,cda_loss is 0.56515,mda_loss is 1.81036
Epoch66, train_loss is 0.04234,test_loss is 3.57190, train_accuracy is 0.99775,test_accuracy is 0.36675,train_all_loss is 1.92315,target_cla_loss is 0.36443,source_cla_loss is 0.10319,cda_loss is 0.55765,mda_loss is 1.72775
Epoch67, train_loss is 0.03892,test_loss is 3.82084, train_accuracy is 0.99650,test_accuracy is 0.35025,train_all_loss is 1.97754,target_cla_loss is 0.32409,source_cla_loss is 0.10061,cda_loss is 0.57595,mda_loss is 1.78693
Epoch68, train_loss is 0.04147,test_loss is 3.64863, train_accuracy is 0.99775,test_accuracy is 0.35175,train_all_loss is 2.04404,target_cla_loss is 0.33164,source_cla_loss is 0.09546,cda_loss is 0.55667,mda_loss is 1.85975
Epoch69, train_loss is 0.03997,test_loss is 3.96786, train_accuracy is 0.99550,test_accuracy is 0.35550,train_all_loss is 1.87499,target_cla_loss is 0.33787,source_cla_loss is 0.09773,cda_loss is 0.55921,mda_loss is 1.68755
Epoch70, train_loss is 0.03543,test_loss is 3.93736, train_accuracy is 0.99975,test_accuracy is 0.35850,train_all_loss is 1.96610,target_cla_loss is 0.34311,source_cla_loss is 0.08989,cda_loss is 0.56598,mda_loss is 1.78530
Epoch71, train_loss is 0.03870,test_loss is 4.00044, train_accuracy is 0.99825,test_accuracy is 0.34775,train_all_loss is 1.97678,target_cla_loss is 0.33252,source_cla_loss is 0.10044,cda_loss is 0.55977,mda_loss is 1.78711
Epoch72, train_loss is 0.03661,test_loss is 4.20446, train_accuracy is 0.99850,test_accuracy is 0.33975,train_all_loss is 1.91947,target_cla_loss is 0.32661,source_cla_loss is 0.09100,cda_loss is 0.55292,mda_loss is 1.74052
Epoch73, train_loss is 0.03299,test_loss is 4.03290, train_accuracy is 0.99975,test_accuracy is 0.35475,train_all_loss is 1.89665,target_cla_loss is 0.32557,source_cla_loss is 0.09065,cda_loss is 0.57111,mda_loss is 1.71633
Epoch74, train_loss is 0.03557,test_loss is 3.74976, train_accuracy is 0.99875,test_accuracy is 0.35050,train_all_loss is 1.94581,target_cla_loss is 0.33794,source_cla_loss is 0.09797,cda_loss is 0.57739,mda_loss is 1.75631
Epoch75, train_loss is 0.03606,test_loss is 3.90655, train_accuracy is 0.99900,test_accuracy is 0.35175,train_all_loss is 1.91040,target_cla_loss is 0.36908,source_cla_loss is 0.09017,cda_loss is 0.56595,mda_loss is 1.72673
Epoch76, train_loss is 0.02996,test_loss is 4.31643, train_accuracy is 0.99850,test_accuracy is 0.35225,train_all_loss is 1.87109,target_cla_loss is 0.35043,source_cla_loss is 0.08601,cda_loss is 0.58625,mda_loss is 1.69141
Epoch77, train_loss is 0.03729,test_loss is 4.11600, train_accuracy is 0.99675,test_accuracy is 0.36625,train_all_loss is 1.98549,target_cla_loss is 0.30353,source_cla_loss is 0.09079,cda_loss is 0.59306,mda_loss is 1.80504
Epoch78, train_loss is 0.03488,test_loss is 4.00549, train_accuracy is 0.99900,test_accuracy is 0.36025,train_all_loss is 1.94518,target_cla_loss is 0.31716,source_cla_loss is 0.09354,cda_loss is 0.56666,mda_loss is 1.76326
Epoch79, train_loss is 0.03735,test_loss is 3.74691, train_accuracy is 0.99775,test_accuracy is 0.36200,train_all_loss is 1.98189,target_cla_loss is 0.36152,source_cla_loss is 0.10226,cda_loss is 0.55654,mda_loss is 1.78782
Epoch80, train_loss is 0.03132,test_loss is 3.66145, train_accuracy is 0.99925,test_accuracy is 0.37600,train_all_loss is 1.92512,target_cla_loss is 0.32335,source_cla_loss is 0.09147,cda_loss is 0.55130,mda_loss is 1.74618
Epoch81, train_loss is 0.04329,test_loss is 3.67632, train_accuracy is 0.99775,test_accuracy is 0.36875,train_all_loss is 2.01323,target_cla_loss is 0.36775,source_cla_loss is 0.12122,cda_loss is 0.53975,mda_loss is 1.80126
Epoch82, train_loss is 0.03796,test_loss is 3.88163, train_accuracy is 0.99800,test_accuracy is 0.36575,train_all_loss is 1.94328,target_cla_loss is 0.34789,source_cla_loss is 0.09737,cda_loss is 0.53522,mda_loss is 1.75760
Epoch83, train_loss is 0.03361,test_loss is 3.93112, train_accuracy is 0.99850,test_accuracy is 0.35125,train_all_loss is 1.94797,target_cla_loss is 0.34964,source_cla_loss is 0.09108,cda_loss is 0.56788,mda_loss is 1.76514
Epoch84, train_loss is 0.03604,test_loss is 3.92195, train_accuracy is 0.99900,test_accuracy is 0.37450,train_all_loss is 1.89947,target_cla_loss is 0.35758,source_cla_loss is 0.09633,cda_loss is 0.56305,mda_loss is 1.71108
Epoch85, train_loss is 0.03087,test_loss is 3.79357, train_accuracy is 0.99850,test_accuracy is 0.37225,train_all_loss is 1.93883,target_cla_loss is 0.33742,source_cla_loss is 0.08896,cda_loss is 0.57650,mda_loss is 1.75848
Epoch86, train_loss is 0.03754,test_loss is 3.96986, train_accuracy is 0.99875,test_accuracy is 0.36800,train_all_loss is 1.87951,target_cla_loss is 0.33196,source_cla_loss is 0.10165,cda_loss is 0.56499,mda_loss is 1.68817
Epoch87, train_loss is 0.03479,test_loss is 4.27059, train_accuracy is 0.99875,test_accuracy is 0.36100,train_all_loss is 1.86776,target_cla_loss is 0.34986,source_cla_loss is 0.10001,cda_loss is 0.55966,mda_loss is 1.67679
Epoch88, train_loss is 0.03385,test_loss is 4.07302, train_accuracy is 0.99900,test_accuracy is 0.36325,train_all_loss is 1.98173,target_cla_loss is 0.32548,source_cla_loss is 0.08992,cda_loss is 0.56596,mda_loss is 1.80266
Epoch89, train_loss is 0.03606,test_loss is 3.76652, train_accuracy is 0.99825,test_accuracy is 0.36950,train_all_loss is 1.99725,target_cla_loss is 0.36634,source_cla_loss is 0.10286,cda_loss is 0.54637,mda_loss is 1.80312
Epoch90, train_loss is 0.03380,test_loss is 3.84020, train_accuracy is 0.99900,test_accuracy is 0.36500,train_all_loss is 1.91125,target_cla_loss is 0.31314,source_cla_loss is 0.08440,cda_loss is 0.53862,mda_loss is 1.74168
Epoch91, train_loss is 0.03329,test_loss is 3.78597, train_accuracy is 0.99875,test_accuracy is 0.36275,train_all_loss is 1.84015,target_cla_loss is 0.35008,source_cla_loss is 0.08478,cda_loss is 0.53946,mda_loss is 1.66642
Epoch92, train_loss is 0.03170,test_loss is 3.90322, train_accuracy is 0.99925,test_accuracy is 0.36850,train_all_loss is 1.84773,target_cla_loss is 0.31886,source_cla_loss is 0.07877,cda_loss is 0.55678,mda_loss is 1.68140
Epoch93, train_loss is 0.02658,test_loss is 4.19532, train_accuracy is 0.99925,test_accuracy is 0.36575,train_all_loss is 1.83341,target_cla_loss is 0.28239,source_cla_loss is 0.06854,cda_loss is 0.56702,mda_loss is 1.67993
Epoch94, train_loss is 0.02931,test_loss is 4.24633, train_accuracy is 0.99950,test_accuracy is 0.36750,train_all_loss is 1.84162,target_cla_loss is 0.28099,source_cla_loss is 0.08070,cda_loss is 0.54899,mda_loss is 1.67792
Epoch95, train_loss is 0.03792,test_loss is 4.27938, train_accuracy is 0.99750,test_accuracy is 0.37175,train_all_loss is 1.89878,target_cla_loss is 0.30210,source_cla_loss is 0.10248,cda_loss is 0.54445,mda_loss is 1.71165
Epoch96, train_loss is 0.02876,test_loss is 3.81267, train_accuracy is 0.99925,test_accuracy is 0.37500,train_all_loss is 1.88203,target_cla_loss is 0.32059,source_cla_loss is 0.07481,cda_loss is 0.55576,mda_loss is 1.71959
Epoch97, train_loss is 0.03724,test_loss is 3.74088, train_accuracy is 0.99775,test_accuracy is 0.37925,train_all_loss is 1.93946,target_cla_loss is 0.33789,source_cla_loss is 0.10656,cda_loss is 0.56191,mda_loss is 1.74292
Epoch98, train_loss is 0.03961,test_loss is 4.07593, train_accuracy is 0.99600,test_accuracy is 0.36750,train_all_loss is 1.90211,target_cla_loss is 0.31981,source_cla_loss is 0.10458,cda_loss is 0.57807,mda_loss is 1.70774
Epoch99, train_loss is 0.02847,test_loss is 4.25489, train_accuracy is 0.99975,test_accuracy is 0.35350,train_all_loss is 1.84025,target_cla_loss is 0.30272,source_cla_loss is 0.07335,cda_loss is 0.59257,mda_loss is 1.67737
Epoch100, train_loss is 0.02855,test_loss is 4.00182, train_accuracy is 0.99850,test_accuracy is 0.36100,train_all_loss is 1.82983,target_cla_loss is 0.28872,source_cla_loss is 0.07271,cda_loss is 0.56430,mda_loss is 1.67182
工学博士，担任《Mechanical System and Signal Processing》审稿专家，担任
《中国电机工程学报》优秀审稿专家，《控制与决策》，《系统工程与电子技术》，《电力系统保护与控制》，《宇航学报》等EI期刊审稿专家。
擅长领域：现代信号处理，机器学习，深度学习，数字孪生，时间序列分析，设备缺陷检测、设备异常检测、设备智能故障诊断与健康管理PHM等。