输电线路故障三相与零序谐波数据，包含外力破坏、山火、异物等情况下的故障数据.rar

# -*- coding: utf-8 -*- """ Created on Wed Jun 14 15:17:36 2017 训练分类器 @author: baishuhua """ import warnings warnings.filterwarnings('ignore') import time import os import Extraction_DTFT as Ext import numpy as np import pandas as pd from sklearn import preprocessing from sklearn.model_selection import train_test_split,cross_val_score,GridSearchCV from sklearn.svm import SVC from sklearn.dummy import DummyClassifier from sklearn.pipeline import Pipeline random_state=np.random.seed() import pickle import sys;sys.path.append('E:\\大数据') import sampler from collections import Counter # ******************** 装载已构建特征集 ******************** # def DataSet(file): Data=pd.read_csv(file,encoding='gbk',na_values=['#NAME?','inf','INF','-inf','-INF']) Featurename=Data.columns.values[2:] Data=Data.dropna(how='all',axis=0,subset=Featurename) Data=Data.fillna(axis='index',method='pad') Feature=Data[Featurename].values Filename=Data['Filename'].values Label=Data['Label'].values return Filename,Featurename,Feature,Label # ******************** 剔除异常样本 ******************** # def RemoveAbnormal(BigFeatures,contamination = 0.05): print('******************** 剔除异常样本 ********************\n') from sklearn import ensemble clf = ensemble.IsolationForest(max_samples='auto', contamination=contamination, \ max_features=1.0, bootstrap=False, random_state=42) clf.fit(BigFeatures) y_detection=clf.predict(BigFeatures) mask = (y_detection==-1) return mask # 异常样本编号 def Encoding(Label): Encoder=preprocessing.LabelEncoder().fit(Label) return Encoder # ******************** 标准化处理特征集 ******************** # def Scaler(): scaler=preprocessing.MaxAbsScaler(copy=True) return scaler # ******************** 模型评估 ******************** # from sklearn.metrics import make_scorer from sklearn.metrics import accuracy_score,f1_score,fbeta_score,jaccard_similarity_score,precision_score,recall_score # 该类指标越大性能越好 from sklearn.metrics import hamming_loss,zero_one_loss # 该类指标绝对值越小性能越好 def Evaluate(score_func=accuracy_score): if score_func in [accuracy_score,f1_score,fbeta_score,jaccard_similarity_score,precision_score,recall_score]: greater_is_better=True elif score_func in [hamming_loss,zero_one_loss]: greater_is_better=False score=make_scorer(score_func=score_func,greater_is_better=greater_is_better,needs_proba=False,needs_threshold=False,average='micro')# return score # 混淆矩阵可视化预测结果 import itertools import matplotlib.pyplot as plt plt.rcParams['font.sans-serif']=['SimHei'] plt.rcParams['axes.unicode_minus']=False def plot_confusion_matrix(cm,classes,normalize=False,title='Confusion matrix',cmap=plt.cm.Blues): """ This function prints and plots the confusion matrix. Normalization can be applied by setting `normalize=True`. """ np.set_printoptions(precision=2) plt.figure(figsize=(12,6)) plt.imshow(cm,interpolation='nearest', cmap=cmap) plt.title(title) plt.colorbar() tick_marks=np.arange(len(classes)) plt.xticks(tick_marks,classes,rotation=45) plt.yticks(tick_marks,classes) if normalize: cm=cm.astype('float')/(cm.sum(axis=1)[:,np.newaxis]+1e-10) print("Normalized confusion matrix") else: print('Confusion matrix, without normalization') print(cm) thresh=cm.max()/ 2. for i,j in itertools.product(range(cm.shape[0]), range(cm.shape[1])): # itertools.product笛卡尔积 代表两层（嵌套循环） plt.text(j,i,round(cm[i,j],2),horizontalalignment="center",color="white" if cm[i,j]>thresh else "black") plt.tight_layout() plt.ylabel('真实标签') # True label plt.xlabel('预测标签') # Predicted label # 可视化二分类器的ROC曲线，并计算曲线积分面积 def Plot_ROC_binary(ys_true,ys_scores,view): y_true=ys_true;y_score=ys_scores fpr,tpr,thresholds=roc_curve(y_true,y_score,pos_label=None,sample_weight=None,drop_intermediate=True) area=auc(fpr,tpr,reorder=False) if view: plt.figure() plt.plot(fpr,tpr,label='ROC and AUC is %.2f' %area) plt.legend(loc='lower right') plt.xlim([0.0,1.0]);plt.ylim([0.0,1.1]) plt.xlabel('假正类');plt.ylabel('真正类') plt.title('ROC曲线') plt.show() return area # 可视化多分类器的ROC曲线，并加权计算曲线积分面积 def Plot_ROC_multi(ys_orig,ys_true,ys_scores,classnames,view): fpr=dict();tpr=dict();roc_auc=dict() for i,classname in enumerate(classnames): y_true=ys_true[:,i];y_score=ys_scores[:,i] fpr[i],tpr[i],thresholds=roc_curve(y_true,y_score,pos_label=None,sample_weight=None,drop_intermediate=True) area=auc(fpr[i],tpr[i],reorder=False) roc_auc[i]=area # Compute weighted-average ROC curve and ROC area,对于weighted模式，各类的加权系数为各类在总类中的占比 all_fpr=np.unique(np.concatenate([fpr[i] for i in range(len(fpr))])) # Then interpolate all ROC curves at this points mean_tpr=np.zeros_like(all_fpr) from scipy import interp for i in range(len(classnames)): mean_tpr += interp(all_fpr,fpr[i],tpr[i])*(sum(ys_orig==i)/len(ys_orig)) # Finally compute AUC fpr['weighted']=all_fpr;tpr['weighted']=mean_tpr roc_auc['weighted']=auc(fpr['weighted'],tpr['weighted']) if view: plt.figure(figsize=(8,6)) for i,classname in enumerate(classnames): plt.plot(fpr[i],tpr[i],label=classname+' ROC and AUC is %.2f' %(roc_auc[i])) plt.plot(fpr['weighted'],tpr['weighted'],label='综合 ROC and AUC is %.2f' %(roc_auc['weighted'])) plt.legend(loc='lower right') plt.xlim([0.0,1.0]);plt.ylim([0.0,1.1]) plt.show() return roc_auc['weighted'] if __name__=='__main__': IsExt = 0 if IsExt: print('******************** Start extracting features ********************\n') FilePath = r'E:\大数据\线路故障诊断\解压后数据\广东中调\*.etr-*' SavePath = r'E:\大数据\线路故障诊断' SaveFile = '故障整次谐波_三相_gd.csv' BigSamplenames,BigFeaturenames,BigFeatures,BigLabels = \ Ext.ExtractingAllFile(FilePath,SavePath,SaveFile,SelectCol=[4,5,6]) else: print('******************** Load extracted features ********************\n') SavePath = r'E:\大数据\线路故障诊断' SaveFile = '故障整次谐波_零序_gd.csv' BigSamplenames,BigFeaturenames,BigFeatures,BigLabels = DataSet(os.path.join(SavePath,SaveFile)) # 装载matlab提取的小波能量均值、标准差和能量熵特征 if 1: from scipy.io import loadmat BigFeatures = loadmat(r'E:\大数据\线路故障诊断\XY.mat')['Features'] # 3指标的对数 BigLabels = loadmat(r'E:\大数据\线路故障诊断\XY.mat')['Labels'][0] # 注意数据格式 BigLabels = np.array([element[0] for element in BigLabels]) BigSamplenames = loadmat(r'E:\大数据\线路故障诊断\XY.mat')['Files'][0] # 注意数据格式 BigSamplenames = np.array([element[0] for element in BigSamplenames]) IsRemove = 1 if IsRemove: mask = RemoveAbnormal(BigFeatures,contamination = 0.05) print('异常样本类别:\n',BigSamplenames[mask]) Samplenames,Labels,Features = BigSamplenames[~mask],BigLabels[~mask],BigFeatures[~mask,:] else: Samplenames,Labels,Features = BigSamplenames,BigLabels,BigFeatures # 筛选 select = (Labels=='山火') | (Labels=='外力破坏') Labels = Labels[select] Features = Features[select] # 编码 Encoder =