机器学习下随机森林算法在电网故障分析指挥系统中的应用  

Application of Stochastic Forest Algorithm in Power Grid Fault Analysis Command System under Machine Learning

作　　者：汤卫东 肖大军 谈林涛 于文娟 TANG Wei-dong;XIAO Da-jun;TAN Lin-tao;YU Wen-juan(Central China Branch of State Grid Corporation of China,Wuhan,Hubei 430077,China)

机构地区：[1]国家电网有限公司华中分部,湖北武汉430077

出　　处：《计算技术与自动化》2022年第3期59-63,共5页Computing Technology and Automation

摘　　要：针对电网系统的故障问题,在Weka语言软件上对随机森林算法为核心的电网故障分析系统模型进行实例分析。同时将随机森林算法与决策树(decision tree)算法、神经网络算法(Neural Network Algorithm, NNA)以及支持向量机(Support Vector Machines, SVM)的预测准确率进行对比,验证随机森林算法的优越性。结果表明,随机森林算法非常适合应用在电网故障分析系统中,在预测准确率方面,故障等级越高预测难度越大,准确率较低,而故障等级越低其故障预测的准确率越高。Aiming at the fault problem of the power grid system, the Weka language software is used to analyze the power grid fault analysis model with the random forest algorithm as the core. Additionally, the random forest algorithm is compared with the decision tree algorithm and the neural network algorithm(NNA), and support vector machine(SVM) regarding the prediction accuracy to verify the superiority of the random forest algorithm. The results show that the random forest algorithm is very suitable for application in the power grid fault analysis system. In terms of prediction accuracy, the higher the fault level, the more difficult the prediction and the lower the accuracy rate. The lower the fault level, the higher the accuracy of the fault prediction.

关 键 词：机器学习 随机森林 电网故障 故障分析 数据挖掘 

分 类 号：TM93[电气工程—电力电子与电力传动]