摘要
电力系统的安全运行和高电力变压器的过热后会散发化学气体,气体浓度是检测设备过热和局部漏电的基础特征;将支持向量机和遗传算法应用于电力变压器的过热诊断中,首先,将变压器过热产生的五种特有气体的浓度作为过热特征参数,代入支持向量机和遗传算法的模型,得到备选的过热特征集,通过适应度函数对种群中的过热参数进行评价,然后,把优化后的训练样本输入到模糊支持向量机中进行相应的训练仿真;最后,通过基于遗传算法优化的支持向量机模型进行电力变压器的过热诊断,诊断结果表明该方法能够准确地诊断出电力变压器的故障类型,具有较高的故障诊断精度。
High power transformer overheating will send out chemical gas, gas concentration is the basis of the local leakage detection equipment overheating and characteristics. In this paper, support vector machine （SVM） and genetic algorithm was applied to power trans- former overheating in the diagnosis, first of all, will have a five special transformer overheating, the concentration of the gas as overheated characteristic parameters, into the model of support vector machine （SVM） and genetic algorithm, from overheating in the alternative feature set, through fitness function to evaluate overheating in the population parameters, and then, after the optimization of the training sample in- puts to the fuzzy support vector machine （SVM） in the corresponding training simulation. Finally, through support vector machine （SVM） based on genetic algorithm optimization model for power transformer overheating diagnosis, the diagnosis results show that the method can accurately diagnose the fault of power transformer type, has the high accuracy of fault diagnosis.
出处
《计算机测量与控制》
北大核心
2014年第2期342-344,共3页
Computer Measurement &Control
基金
河南省科技发展计划(102102210419)
关键词
支持向量机
遗传算法
电力变压器
故障诊断
support vector machine （SVM）
genetic algorithm
power transformer
fault diagnosis
