机械轴承故障诊断之连续小波变换（cwt）——转换与导入resnet

 

 右图为通过连续小波变换将一维振动信号转化为二维时频图

看来很多文章大篇幅的再讲理论，有意义，但在下朽木，只能贴代码和图片

#导入包
from read_picture import read_directory
from tensorflow import keras
from sklearn.metrics import confusion_matrix
import matplotlib.pyplot as plt
import tensorflow.keras as keras
import tensorflow.keras.layers as layers
import random
from datetime import datetime
import numpy as np

num_classes = 10
height = 52
width = 52

# 小波时频图---2D-CNN输入
x_train, y_train = read_directory(r'cwt_picture\train', height, width, normal=1)
x_valid, y_valid = read_directory(r'cwt_picture\valid', height, width, normal=1)
x_test, y_test = read_directory(r'cwt_picture\test', height, width, normal=1)

x_train = np.squeeze(x_train)
x_valid = np.squeeze(x_valid)
x_test = np.squeeze(x_test)

x_train = np.expand_dims(x_train, axis=3)
x_valid = np.expand_dims(x_valid, axis=3)
x_test = np.expand_dims(x_test, axis=3)
y_train = [int(i) for i in y_train]
y_valid = [