基于传感器时间序列数据的机器故障预测(第一部分，Python)

于 2024-09-02 11:56:13 发布
阅读量392 收藏 9
点赞数 4
分类专栏： 深度学习 信号处理 机器学习 文章标签： python 开发语言 人工智能 算法 深度学习 随机森林
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/weixin_39402231/article/details/141816337
版权
信号处理 同时被 3 个专栏收录
347 篇文章 29 订阅
订阅专栏
机器学习
191 篇文章 1 订阅
订阅专栏
深度学习
152 篇文章 5 订阅
订阅专栏

Importing the necessary Libraries and Dataset

## Importing necessary libraries
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import datetime
import warnings
warnings.filterwarnings('ignore')
sns.set(style = "whitegrid",font_scale = 1.5)
%matplotlib inline
plt.rcParams['figure.figsize']=[12,8]
#importing Dataset
sensor_df = pd.read_csv('sensor.csv')

Data Wrangling

print("The dataset has " , sensor_df.shape[0],"rows and", sensor_df.shape[1], "columns")
The dataset has  220320 rows and 55 columns
#First 10 rows
sensor_df.head(10)

# Last 10 rows
sensor_df.tail(10)

We have data for sensor readings from April to August, collected daily every minute.

sensor_df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 220320 entries, 0 to 220319
Data columns (total 55 columns):
 #   Column          Non-Null Count   Dtype  
---  ------          --------------   -----  
 0   Unnamed: 0      220320 non-null  int64  
 1   timestamp       220320 non-null  object 
 2   sensor_00       210112 non-null  float64
 3   sensor_01       219951 non-null  float64
 4   sensor_02       220301 non-null  float64
 5   sensor_03       220301 non-null  float64
 6   sensor_04       220301 non-null  float64
 7   sensor_05       220301 non-null  float64
 8   sensor_06       215522 non-null  float64
 9   sensor_07       214869 non-null  float64
 10  sensor_08       215213 non-null  float64
 11  sensor_09       215725 non-null  float64
 12  sensor_10       220301 non-null  float64
 13  sensor_11       220301 non-null  float64
 14  sensor_12       220301 non-null  float64
 15  sensor_13       220301 non-null  float64
 16  sensor_14       220299 non-null  float64
 17  sensor_15       0 non-null       float64
 18  sensor_16       220289 non-null  float64
 19  sensor_17       220274 non-null  float64
 20  sensor_18       220274 non-null  float64
 21  sensor_19       220304 non-null  float64
 22  sensor_20       220304 non-null  float64
 23  sensor_21       220304 non-null  float64
 24  sensor_22       220279 non-null  float64
 25  sensor_23       220304 non-null  float64
 26  sensor_24       220304 non-null  float64
 27  sensor_25       220284 non-null  float64
 28  sensor_26       220300 non-null  float64
 29  sensor_27       220304 non-null  float64
 30  sensor_28       220304 non-null  float64
 31  sensor_29       220248 non-null  float64
 32  sensor_30       220059 non-null  float64
 33  sensor_31       220304 non-null  float64
 34  sensor_32       220252 non-null  float64
 35  sensor_33       220304 non-null  float64
 36  sensor_34       220304 non-null  float64
 37  sensor_35       220304 non-null  float64
 38  sensor_36       220304 non-null  float64
 39  sensor_37       220304 non-null  float64
 40  sensor_38       220293 non-null  float64
 41  sensor_39       220293 non-null  float64
 42  sensor_40       220293 non-null  float64
 43  sensor_41       220293 non-null  float64
 44  sensor_42       220293 non-null  float64
 45  sensor_43       220293 non-null  float64
 46  sensor_44       220293 non-null  float64
 47  sensor_45       220293 non-null  float64
 48  sensor_46       220293 non-null  float64
 49  sensor_47       220293 non-null  float64
 50  sensor_48       220293 non-null  float64
 51  sensor_49       220293 non-null  float64
 52  sensor_50       143303 non-null  float64
 53  sensor_51       204937 non-null  float64
 54  machine_status  220320 non-null  object 
dtypes: float64(52), int64(1), object(2)
memory usage: 92.5+ MB
The data set consists of 51 numerical features ,timestamp and a categorical label.


The label contains string values that represent normal, broken and recovering operational conditions of the machine.

countmeanstdmin25%50%75%max
Unnamed: 0220320.0110159.50000063601.0499910.00000055079.750000110159.500000165239.250000220319.000000
sensor_00210112.02.3722210.4122270.0000002.4388312.4565392.4998262.549016
sensor_01219951.047.5916113.2966660.00000046.31076048.13367849.47916056.727430
sensor_02220301.050.8673923.66682033.15972050.39062051.64930052.77777056.032990
sensor_03220301.043.7524812.41888731.64062042.83853944.22742845.31250048.220490
sensor_04220301.0590.673936144.0239122.798032626.620400632.638916637.615723800.000000
sensor_05220301.073.39641417.2982470.00000069.97626075.57679080.91215099.999880
sensor_06215522.013.5015372.1637360.01446813.34635013.64294014.53993022.251160
sensor_07214869.015.8431522.2011550.00000015.90712016.16753016.42795023.596640
sensor_08215213.015.2007212.0373900.02893515.18374015.49479015.69734024.348960
sensor_09215725.014.7992102.0919630.00000015.05353015.08247015.11863025.000000
sensor_10220301.041.47033912.0935190.00000040.70526044.29134047.46376076.106860
sensor_11220301.041.91831913.0564250.00000038.85642045.36314049.65654060.000000
sensor_12220301.029.13697510.1139350.00000028.68681032.51583034.93973045.000000
sensor_13220301.07.0788586.9017550.0000001.5385162.92980912.85952031.187550
sensor_14220299.0376.860041113.20638232.409550418.103250420.106200420.997100500.000000
sensor_150.0NaNNaNNaNNaNNaNNaNNaN
sensor_16220289.0416.472892126.0726420.000000459.453400462.856100464.302700739.741500
sensor_17220274.0421.127517129.1561750.000000454.138825462.020250466.857075599.999939
sensor_18220274.02.3037850.7658830.0000002.4475422.5337042.5876824.873250
sensor_19220304.0590.829775199.3458200.000000662.768975665.672400667.146700878.917900
sensor_20220304.0360.805165101.9741180.000000398.021500399.367000400.088400448.907900
sensor_21220304.0796.225942226.67931795.527660875.464400879.697600882.1299001107.526000
sensor_22220279.0459.792815154.5283370.000000478.962600531.855900534.254850594.061100
sensor_23220304.0922.609264291.8352800.000000950.922400981.9250001090.8080001227.564000
sensor_24220304.0556.235397182.2979790.000000601.151050625.873500628.6077251000.000000
sensor_25220284.0649.144799220.8651660.000000693.957800740.203500750.357125839.575000
sensor_26220300.0786.411781246.66360843.154790790.489575861.869600919.1047751214.420000
sensor_27220304.0501.506589169.8231730.000000448.297950494.468450536.2745502000.000000
sensor_28220304.0851.690339313.0740324.319347782.682625967.2798501043.9765001841.146000
sensor_29220248.0576.195305225.7640910.636574518.947225564.872500744.0214751466.281000
sensor_30220059.0614.596442195.7268720.000000627.777800668.981400697.2222001600.000000
sensor_31220304.0863.323100283.54476023.958330839.062400917.708300981.2499001800.000000
sensor_32220252.0804.283915260.6023610.240716760.607475878.850750943.8776251839.211000
sensor_33220304.0486.405980150.7518366.460602489.761075512.271750555.1632251578.600000
sensor_34220304.0234.97177688.37606554.882370172.486300226.356050316.844950425.549800
sensor_35220304.0427.129817141.7725190.000000353.176625473.349350528.891025694.479126
sensor_36220304.0593.033876289.3855112.260970288.547575709.668050837.333025984.060700
sensor_37220304.060.78736037.6048830.00000028.79922064.29548590.821928174.901200
sensor_38220293.049.65594610.54039724.47916645.57291049.47916053.645830417.708300
sensor_39220293.036.61044415.61372319.27083032.55208035.41666039.062500547.916600
sensor_40220293.068.84453021.37113923.43750057.81250066.40625077.864580512.760400
sensor_41220293.035.3651267.89866520.83333032.55208034.89583237.760410420.312500
sensor_42220293.035.45345510.25952122.13541632.81250035.15625036.979164374.218800
sensor_43220293.043.87959111.04440424.47916639.58333042.96875046.614580408.593700
sensor_44220293.042.65687711.57635525.75231636.74768440.50926045.1388901000.000000
sensor_45220293.043.09498412.83752026.33101836.74768440.21991044.849540320.312500
sensor_46220293.048.01858515.64128426.33101840.50925844.84954051.215280370.370400
sensor_47220293.044.34090310.44243727.19907039.06250042.53472046.585650303.530100
sensor_48220293.0150.88904482.24495726.33101883.912030138.020800208.333300561.632000
sensor_49220293.057.11996819.14359826.62037047.74306052.66204060.763890464.409700
sensor_50143303.0183.04926065.25865027.488426167.534700193.865700219.9074001000.000000
sensor_51204937.0202.699667109.58860727.777779179.108800197.338000216.7245001000.000000

print('all class labels:',sensor_df['machine_status'].unique())
all class labels: ['NORMAL' 'BROKEN' 'RECOVERING']

The label data has 3 machine status values:

BROKEN represents machine is failed.

RECOVERING represents machine trying to recover from failed status.

NORMAL represents the machine is working in normal status.

#Machine status distribution


sensor_df['machine_status'].value_counts()
NORMAL        205836
RECOVERING     14477
BROKEN             7
Name: machine_status, dtype: int64

Data Preprocessing

Remove redundant columns

Remove duplicates

Handle missing values

Convert Timestamp column which is of type object to datetime

#no values for sensor 15 and unnamed column is unnecessary,so I will drop these columns
sensor_df.drop(['sensor_15','Unnamed: 0'],inplace = True,axis=1)
sensor_df.head()

#check percentage of missing values for each column
(sensor_df.isnull().sum().sort_values(ascending=False)/len(sensor_df))*100
sensor_50         34.956881
sensor_51          6.982117
sensor_00          4.633261
sensor_07          2.474129
sensor_08          2.317992
sensor_06          2.177741
sensor_09          2.085603
sensor_01          0.167484
sensor_30          0.118464
sensor_29          0.032680
sensor_32          0.030864
sensor_17          0.020879
sensor_18          0.020879
sensor_22          0.018609
sensor_25          0.016340
sensor_16          0.014070
sensor_49          0.012255
sensor_48          0.012255
sensor_47          0.012255
sensor_46          0.012255
sensor_45          0.012255
sensor_44          0.012255
sensor_43          0.012255
sensor_42          0.012255
sensor_41          0.012255
sensor_40          0.012255
sensor_39          0.012255
sensor_38          0.012255
sensor_14          0.009532
sensor_26          0.009078
sensor_03          0.008624
sensor_10          0.008624
sensor_13          0.008624
sensor_12          0.008624
sensor_11          0.008624
sensor_05          0.008624
sensor_04          0.008624
sensor_02          0.008624
sensor_36          0.007262
sensor_37          0.007262
sensor_28          0.007262
sensor_27          0.007262
sensor_31          0.007262
sensor_35          0.007262
sensor_24          0.007262
sensor_23          0.007262
sensor_34          0.007262
sensor_21          0.007262
sensor_20          0.007262
sensor_19          0.007262
sensor_33          0.007262
timestamp          0.000000
machine_status     0.000000
dtype: float64
#too many missing values in sensor 50 , so dropping that
sensor_df.drop('sensor_50',inplace = True,axis=1)
sensor_df.head()

to impute some of the missing values with their mean

#imputing the remaining missing values with mean
sensor_df.fillna(sensor_df.mean(),inplace= True)
sensor_df.isnull().sum()
timestamp         0
sensor_00         0
sensor_01         0
sensor_02         0
sensor_03         0
sensor_04         0
sensor_05         0
sensor_06         0
sensor_07         0
sensor_08         0
sensor_09         0
sensor_10         0
sensor_11         0
sensor_12         0
sensor_13         0
sensor_14         0
sensor_16         0
sensor_17         0
sensor_18         0
sensor_19         0
sensor_20         0
sensor_21         0
sensor_22         0
sensor_23         0
sensor_24         0
sensor_25         0
sensor_26         0
sensor_27         0
sensor_28         0
sensor_29         0
sensor_30         0
sensor_31         0
sensor_32         0
sensor_33         0
sensor_34         0
sensor_35         0
sensor_36         0
sensor_37         0
sensor_38         0
sensor_39         0
sensor_40         0
sensor_41         0
sensor_42         0
sensor_43         0
sensor_44         0
sensor_45         0
sensor_46         0
sensor_47         0
sensor_48         0
sensor_49         0
sensor_51         0
machine_status    0
dtype: int64
#checking of duplicate rows
    
sensor_df.duplicated().any()
False

No duplicate values, hence we don't need to remove any row.

# Now, lets make it a time series and set it as index
sensor_df['timestamp'] = pd.to_datetime(sensor_df['timestamp'])
sensor_df = sensor_df.set_index('timestamp')

The First 5 rows of the dataset looks as follows.

sensor_df.head()

Exploratory Data Analysis

## Machine status distribution
sensor_df.machine_status.value_counts()
NORMAL        205836
RECOVERING     14477
BROKEN             7
Name: machine_status, dtype: int64
#Plotting Machine Status Distribution
plt.figure(figsize=(5,3))
plt.title('machine_status')
sns.countplot(x='machine_status',data=sensor_df)

#machine status -  pie chart
plt.figure(figsize=(5,3))
stroke_labels = ["Normal","Recovering","Broken"]
sizes = sensor_df.machine_status.value_counts()


plt.pie(x=sizes,labels=stroke_labels)
plt.show()

plt.figure(figsize=(60,40))
sns.heatmap(sensor_df.corr(),annot=True,cmap='coolwarm');
corr = sensor_df.corr()

#The HeatMap shows Correlation between features greater than 0.8
corr80 = corr[abs(corr)> 0.8]
sns.heatmap(corr80,cmap='coolwarm')

#extract the readings from the Broken state of the pump
broken= sensor_df[sensor_df['machine_status']=='BROKEN']


#Extract the name of the numerical columns
sensor_df_2 = sensor_df.drop(['machine_status'],axis=1)
names= sensor_df_2.columns


#plot timeseries for each sensor with Broken state marked with X in red color


for name in names:
    _=plt.figure(figsize=(18,3))
    _=plt.plot(broken[name],linestyle='none',marker='X',color='red',markersize=12)
    _=plt.plot(sensor_df[name],color='blue')
    _=plt.title(name)
    plt.show()

#Converting the machine status column to numeric
MS_class_dict = {"BROKEN": 0, "NORMAL": 1, "RECOVERING": 2}
sensor_df['machine_status'] = sensor_df['machine_status'].map(MS_class_dict)
sensor_df.head()

Stationarity and Autocorrelation

# Resample the entire dataset by daily average
rollmean = sensor_df.resample(rule='D').mean()
rollstd = sensor_df.resample(rule='D').std()
# Plot time series for each sensor with its mean and standard deviation with the BROKEN state marked with X in red color
for name in names:
    _ = plt.figure(figsize=(18,3))
    _ = plt.plot(sensor_df[name], color='blue', label='Original')
    _ = plt.plot(rollmean[name], color='red', label='Rolling Mean')
    _ = plt.plot(rollstd[name], color='black', label='Rolling Std' )
    _ = plt.legend(loc='best')
    _ = plt.title(name)
    plt.show()

Pre-processing and Feature Engineering

1.Scale the data

2.Perform PCA and look at the most important principal components based on inertia

# Standardize/scale the dataset and apply PCA
from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA
from sklearn.pipeline import make_pipeline


# Extract the names of the numerical columns
df2 = sensor_df.drop(['machine_status'], axis=1)
names=df2.columns
x = sensor_df[names]
scaler = StandardScaler()
pca = PCA()
pipeline = make_pipeline(scaler, pca)
pipeline.fit(x)
Pipeline(steps=[('standardscaler', StandardScaler()), ('pca', PCA())])
features = range(pca.n_components_)
_ = plt.figure(figsize=(22, 5))
_ = plt.bar(features, pca.explained_variance_)
_ = plt.xlabel('PCA feature')
_ = plt.ylabel('Variance')
_ = plt.xticks(features)
_ = plt.title("Importance of the Principal Components based on inertia")
plt.show()

# Calculate PCA with 2 components
pca = PCA(n_components=2)
principalComponents = pca.fit_transform(x)
principalDf = pd.DataFrame(data = principalComponents, columns = ['pc1', 'pc2'])
principalDf.head()

pc1pc2
069.523134265.704571
169.523134265.704571
227.841569283.462197
324.548981290.213914
429.627090294.619639

sensor_df['pc1']=pd.Series(principalDf['pc1'].values, index=sensor_df.index)
sensor_df['pc2']=pd.Series(principalDf['pc2'].values, index=sensor_df.index)
sensor_df['pc1'],sensor_df['pc2']
(timestamp
 2018-04-01 00:00:00     69.523134
 2018-04-01 00:01:00     69.523134
 2018-04-01 00:02:00     27.841569
 2018-04-01 00:03:00     24.548981
 2018-04-01 00:04:00     29.627090
                           ...    
 2018-08-31 23:55:00   -308.531521
 2018-08-31 23:56:00   -294.603672
 2018-08-31 23:57:00   -300.207654
 2018-08-31 23:58:00   -285.141390
 2018-08-31 23:59:00   -298.182526
 Name: pc1, Length: 220320, dtype: float64,
 timestamp
 2018-04-01 00:00:00    265.704571
 2018-04-01 00:01:00    265.704571
 2018-04-01 00:02:00    283.462197
 2018-04-01 00:03:00    290.213914
 2018-04-01 00:04:00    294.619639
                           ...    
 2018-08-31 23:55:00   -274.597310
 2018-08-31 23:56:00   -256.328332
 2018-08-31 23:57:00   -256.932727
 2018-08-31 23:58:00   -263.099471
 2018-08-31 23:59:00   -264.545953
 Name: pc2, Length: 220320, dtype: float64)

Check stationarity with Dickey-Fuller Test

# Compute change in daily mean 
pca1 = principalDf['pc1'].pct_change()
# Compute autocorrelation
autocorrelation = pca1.dropna().autocorr()
print('Autocorrelation is: ', autocorrelation)
Autocorrelation is:  -6.773604518564134e-06
# Plot ACF
from statsmodels.graphics.tsaplots import plot_acf
plot_acf(pca1.dropna(), lags=20, alpha=0.05)

# Compute change in daily mean 
pca2 = principalDf['pc2'].pct_change()
# Compute autocorrelation
autocorrelation = pca2.autocorr()
print('Autocorrelation is: ', autocorrelation)
Autocorrelation is:  -1.4768185278992946e-06
# Plot ACF
from statsmodels.graphics.tsaplots import plot_acf
plot_acf(pca2.dropna(), lags=20, alpha=0.05)

知乎学术咨询：https://www.zhihu.com/consult/people/792359672131756032?isMe=1

担任《Mechanical System and Signal Processing》等审稿专家，擅长领域：信号滤波/降噪，机器学习/深度学习，时间序列预分析/预测，设备故障诊断/缺陷检测/异常检测。

分割线

基于自编码器的时间序列异常检测（Python，ipynb文件）

import pandas as pd import tensorflow as tf from keras.layers import Input, Dense from keras.models import Model from sklearn.metrics import precision_recall_fscore_support import matplotlib.pyplot as plt

完整代码：

mbd.pub/o/bread/mbd-Zpmcl5xq

图片

哥廷根数学学派
关注
  • 4
    点赞
  • 踩
  • 9
    收藏
    觉得还不错? 一键收藏
  • 打赏
    打赏
  • 0
    评论
专栏目录
使用 Python 从故障轴承振动信号序列判断故障出现时间
cocofisher的博客
08-05 6459
从故障轴承振动信号序列判断故障出现时间
时间序列预测的多Agent系统方法
光剑书架上的书
07-01 432
时间序列预测的多-Agent系统方法 作者：禅与计算机程序设计艺术 / Zen and the Art of Computer Programming / TextGenWebUILLM 时间序列预测的多-Agent系统方法
时间序列模式识别_空气质量传感器数据的时间序列模式识别
weixin_26705191的博客
09-19 1787
时间序列模式识别· 1. Introduction· 2. Exploratory Data Analysis ∘ 2.1 Pattern Changes ∘ 2.2 Correlation Between Features· 3. Anomaly Detection and Pattern Recognition ∘ 3.1 Point Anomaly Detection (System Fau...
python 时间序列异常值_时间序列异常检测（一）—— 算法综述
weixin_39646706的博客
12-10 2567
更新：2020/09/02：新增一篇写的很好的关于视频监控异常检测的学报http://jst.tsinghuajournals.com/CN/rhhtml/20200609.htm​jst.tsinghuajournals.com一. 介绍异常检测(Anomaly detection)是目前时序数据分析最成熟的应用之一，定义是从正常的时间序列中识别不正常的事件或行为的过程。有效的异常检测被广泛用于...
基于SARIMA、XGBoost和CNN-LSTM的时间序列预测对比
数据派THU
12-19 945
来源：DeepHub IMBA本文约6800字，建议阅读10+分钟本文将讨论通过使用假设测试、特征工程、时间序列建模方法等从数据集中获得有形价值的技术。利用统计测试和机器学习分析和预测太阳能发电的性能测试和对比。本文将讨论通过使用假设测试、特征工程、时间序列建模方法等从数据集中获得有形价值的技术。我还将解决不同时间序列模型的数据泄漏和数据准备等问题，并且对常见的三种时间序列预测进行对比测试。介绍时...
时间序列 工具库学习（1） tsfresh特征提取、特征选择
酒心★
01-07 1万+
1. 更新清单： 2022.01.07：初次更新文章 2. 了解、安装tsfresh tsfresh 可以自动计算大量的时间序列特性，包含许多特征提取方法和强大的特征选择算法。 有一个名为hctsa的 matlab 包，可用于从时间序列中自动提取特征。也可以通过pyopy 包在 Python 中使用 hctsa 。其他可用的打包程序是featuretools、FATS和cesium。 pip install tsfresh 3.快速使用案例【官网案例】 得到了一个包含[1] 中讨论的机器人故障的数据集。该
基于深度学习的时间序列分类[含代码]
格拉迪沃的博客
03-12 2万+
基于深度学习的时间序列分类引言数据集实验环境搭建实验设计实验代码实验结果 引言 目前，深度学习在计算机视觉和语音识别上有了非常广泛的应用，但是在工业应用方面还没有完善的体系，一方面缺乏数据集另一方缺乏优秀的顶级论文。在工业上的故障诊断领域，大多数据都来自于传感器的采集，如是西储大学轴承数据，TE化工数据集等，都是典型的时间序列，因而绝大多数问题可以抽象成时间序列分类（TSC）问题。因此本人准备从时...
用python将时间序列信号或一维数组 转化成 图像的几种方法
热门推荐
qq_41281244的博客
09-24 3万+
系列文章目录 提示：这里可以添加系列文章的所有文章的目录，目录需要自己手动添加 例如：第一章 Python 机器学习入门之pandas的使用 提示：写完文章后，目录可以自动生成，如何生成可参考右边的帮助文档 文章目录系列文章目录前言一、pandas是什么？二、使用步骤1.引入库2.读入数据总结前言一、pandas是什么？二、使用步骤1.引入库2.读入数据总结 前言 提示：这里可以添加本文要记录的大概内容： 例如：随着人工智能的不断发展，机器学习这门技术也越来越重要，很多人都开启了学习机器学习，本文就介
故障诊断：美国西储大学（CRWU）轴承故障数据集
02-18
每个文件可能包含多个变量，如时间序列数据、采样率、传感器位置等信息，这些都是进行故障诊断分析的关键元素。 总的来说，CWRU轴承故障数据集为研究者提供了一个宝贵的资源，用于开发和评估故障诊断算法，特别是在...
物联网预测分析：使用传感器数据预测设备故障的方法。 安装在IoT设备等设备上的传感器，机器人手臂，过程监视和控制设备等自动化制造设备可连续收集和传输带有时间戳的数据
02-04
**数据预处理**是预测分析的第一步。这涉及到数据清洗，去除异常值，填补缺失值，以及将时间序列数据转换为适合机器学习算法的格式。例如，可以使用Python编程语言中的Pandas库进行这些操作，同时利用Jupyter ...
python12 中，No module named‘distutils‘错误
qq_43557600的博客
08-27 1万+
python12 “ No module named'distutils' ”,​ 安装 pip install setuptools ，解决 ​
centos7 安装 superset4.0.2 教程
m0_37759590的博客
08-27 1万+
centos7 安装 superset4.0.2 教程
【python】关于判断空值None、判断len()=0，和 if not 的区别
Meggie35的博客
08-28 6453
python的“空”与“无”之辩，关于None，和空数值。在程序中，判断里面没有值和判断东西是不是为空，也是不一样的东西，我们可以往下看看具体的区别。
【Python-办公自动化】批量修改EXCEL指定内容
2202_76035290的博客
08-29 3034
欢迎来到"花花 Show Python"，一名热爱编程和分享知识的技术博主。在这里，我将与您一同探索Python的奥秘，分享编程技巧、项目实践和学习心得。无论您是编程新手还是资深开发者，都能在这里找到有价值的信息和灵感。我热衷于将复杂的技术概念以简单易懂的方式呈现给大家，让每个人都能享受到编程的乐趣。我相信，通过不断的学习和实践，我们都能够成为更好的开发者。如果您喜欢我的内容，别忘了点击关注哦！这样您就可以第一时间获取我的更新和分享。您的支持是我不断前进的动力。
宝塔面板网站SSL证书到期如何手动和自动续签Let’s Encrypt
weixin_45734165的博客
08-27 1480
宝塔面板提供2种免费的SSL证书：宝塔SSL，Let’s Encrypt。前面第1，2步骤主要是手动续签，比较稳定可靠，唯一的缺点是需要进入后台进行操作，并且有时候可能忘记续签。实际上宝塔面板还提供了很多计划任务的脚本，可以自动续签Let’s Encrypt证书。不管是给网站部署SSL还是给网站SSL续签，利用宝塔后台来进行网站SSL证书的管理还是比较方便高效的。2.手动续签：网站设置后台点击【SSL】》【Let’s Encrypt】》【续签】1.手动续签：登录宝塔面板后台，点击【网站】》【设置】
【Python报错已解决】`TypeError: ‘numpy.ndarray‘ object is not callable`
最新发布
鸽芷咕的博客
09-01 196
在编程中，你是否遇到过尝试调用一个Numpy数组对象时出现了`TypeError: 'numpy.ndarray' object is not callable`的报错？这个问题通常是由于误将Numpy数组当成了一个函数来调用。接下来，我们将分析这个问题并提供解决方案。
np.ndarray和np.array区别；MXNet的 mx.array 类型是什么；NDArray优化了什么：并行计算优化
ZJQ的博客
09-01 119
如果使用传统的密集格式（如NumPy的ndarray）来存储这些梯度，将会占用大量的内存，并导致计算效率低下。MXNet是一个开源的深度学习框架，它提供了多种数据结构和操作来支持高效的数值计算和深度学习模型的训练。总之，MXNet的NDArray通过异步计算、内存优化、高效的数学和线性代数运算、稀疏数据支持以及自动化求导等优化措施，为深度学习模型的训练和数值计算提供了强大的支持。在表面上看似不同，但实际上它们之间有着紧密的联系，并且在大多数情况下可以互换使用，但它们的本质和用途略有不同。
yield：生成器
mohen_777的博客
08-28 1051
作为生成器，因为每次迭代就会返回一个值，所以不能显示的在生成器函数中return 某个值，包括None值也不行，否则会抛出“SyntaxError”的异常，但是在函数中可以出现单独的return，表示结束该语句。使用yield，可以让函数生成一个序列，该函数返回的对象类型是"generator"，通过该对象连续调用next()方法返回序列值。另外一种说法：生成器就是一个返回迭代器的函数，与普通函数的区别是生成器包含yield语句，更简单点理解生成器就是一个迭代器。处为止，next()的返回值就是生成值。
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

哥廷根数学学派

码字不易，且行且珍惜

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付：¥1
获取中
扫码支付

您的余额不足，请更换扫码支付或充值

打赏作者

实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值