基于机器学习的电池剩余使用寿命RUL预测

​代码较为简单。

import numpy as np # linear algebraimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)import os

import matplotlib.pyplot as pltimport seaborn as sns%matplotlib inline​import warningswarnings.filterwarnings('ignore')

df=pd.read_csv("Battery_RUL.csv")df.head()

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 15064 entries, 0 to 15063
Data columns (total 9 columns):
 #   Column                     Non-Null Count  Dtype  
---  ------                     --------------  -----  
 0   Cycle_Index                15064 non-null  float64
 1   Discharge Time (s)         15064 non-null  float64
 2   Decrement 3.6-3.4V (s)     15064 non-null  float64
 3   Max. Voltage Dischar. (V)  15064 non-null  float64
 4   Min. Voltage Charg. (V)    15064 non-null  float64
 5   Time at 4.15V (s)          15064 non-null  float64
 6   Time constant current (s)  15064 non-null  float64
 7   Charging time (s)          15064 non-null  float64
 8   RUL                        15064 non-null  int64  
dtypes: float64(8), int64(1)
memory usage: 1.0 MB

df.describe()

numlist=df.columns fig = plt.figure(figsize=(20,20))for i in range(len(numlist)):    plt.subplot(3,3,i+1)    plt.title(numlist[i],fontsize=12)    plt.hist(df[numlist[i]])​plt.tight_layout()

sns.pairplot(df)

​

plt.figure(figsize = (15,8))sns.heatmap(df.corr(),annot=True, cbar=False, cmap='Blues', fmt='.1f')

df1=df.drop(['Cycle_Index','Discharge Time (s)', 'Decrement 3.6-3.4V (s)', 'Time constant current (s)','Charging time (s)'],axis=1)from pycaret.regression import *exp_reg = setup(df1, target = 'RUL',remove_outliers=True,session_id=1)

compare_models()

ExtraTreesRegressor(n_jobs=-1, random_state=1)

reg_model = create_model('et')

plot_model(reg_model,plot='error')

plot_model(reg_model,plot='residuals')

plot_model(reg_model,plot='feature')

xgb = create_model('xgboost',verbose=False)rf = create_model('rf', verbose = False)blend_specific = blend_models(estimator_list = [rf,xgb,reg_model])

工学博士，担任《Mechanical System and Signal Processing》《中国电机工程学报》《控制与决策》等期刊审稿专家，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。