西储大学轴承故障数据预处理_西储大学轴承数据集预处理python

西储大学轴承各类型故障数据准备

导入python处理包看下

import scipy.io # To use the '.mat' files
import seaborn as sns
import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt

看下文件下文件

for root, dirs, files in os.walk("0hp_load_48_KHz", topdown=False):
    for file_name in files:
        path = os.path.join(root, file_name)
        print(path)

看下数据mat内部概况

path = r'0hp_load_48_KHz\7_OR3.mat'
mat = scipy.io.loadmat(path)
mat.items()

list(mat.keys())

 

获取风扇端数据

key_name = list(mat.keys())[3]
DE_data = mat.get(key_name)
fault = np.full((len(DE_data), 1), file_name[:-4])
df_temp = pd.DataFrame({'DE_data':np.ravel(DE_data) , 'fault':np.ravel(fault)})
df_temp

 

plt.figure(figsize=(15,5))
plt.plot(df_temp.iloc[:,0])
plt.show()

 画下这个数据的波形

 处理数据集

df=pd.DataFrame(columns=['DE_data','fault'])
 
for root, dirs, files in os.walk("0hp_load_48_KHz", topdown=False):
    for file_name in files:
        path = os.path.join(root, file_name)
        print(path)
 
        mat = scipy.io.loadmat(path)
 
        key_name = list(mat.keys())[3]
        DE_data = mat.get(key_name)
        fault = np.full((len(DE_data), 1), file_name[:-4])
 
        df_temp = pd.DataFrame({'DE_data':np.ravel(DE_data) , 'fault':np.ravel(fault)})
        
        df = pd.concat([df,df_temp],axis=0)
        print(df['fault'].unique())
        
df.to_csv('0hp_all_faults.csv',index=False)   

看下数据集情况

df = pd.read_csv('0hp_all_faults.csv')

faults = df['fault'].unique()
for  f in faults:
    plt.figure(figsize=(10,3))
    plt.plot(df[df['fault']==f].iloc[:,0])
    plt.title(f)
    plt.show()

 

 

 

 

 

 

等等。。。。

 拼接看下

plt.figure(figsize=(15,5))
sns.scatterplot(data=df.iloc[::100,:],y='DE_data',x=np.arange(0,len(df),100),hue='fault')
plt.show()

 生成数据集文件

load=3
 
df=pd.DataFrame(columns=['DE_data','fault'])
 
for root, dirs, files in os.walk("{}hp_load_48_KHz".format(load), topdown=False):
    for file_name in files:
        path = os.path.join(root, file_name)
        print(path)
 
        mat = scipy.io.loadmat(path)
 
        for k in list(mat.keys()):
            if k[5:]== 'DE_time':
                key_name=k
        DE_data = mat.get(key_name)
        fault = np.full((len(DE_data), 1), file_name[:-4])
 
        df_temp = pd.DataFrame({'DE_data':np.ravel(DE_data) , 'fault':np.ravel(fault)})
        
        df = pd.concat([df,df_temp],axis=0)
        print(df['fault'].unique())
        
df.to_csv('{}hp_all_faults.csv'.format(load),index=False)   

欢迎加微信公众号：轴承故障诊断和寿命预测