摘要
针对支持向量机(SVM)诊断变压器故障的效果不稳定的问题,利用Ada Boost集成算法对其强化,得到的AdaBoost-SVM模型诊断结果比较稳定,但准确度依然有待提高。因此,提出利用麻雀搜索算法(SSA)对Ada Boost-SVM模型进行优化,指定其弱分类器权重αt、SVM惩罚因子c和核参数g的寻优范围,使用SSA对三种参数在指定的寻优范围内寻优,提高模型的准确率。将提出的SSA-AdaBoost-SVM变压器故障诊断模型与PSO-SVM、SSA-SVM、AdaBoost-SVM、AdaBoost-SSA-SVM和PSO-AdaBoost-SVM五种模型对比,提出的模型具有更高的准确率和稳定性,平均准确率可达91.58%。实验结果表明,提出的SSA-AdaBoost-SVM变压器故障诊断模型具有更好的表现。
In view of the unstable effect of support vector machine(SVM) in transformer fault diagnosis, the AdaBoost integrated algorithm is used to strengthen it, and the results of the AdaBoost-SVM model are relatively stable, but the accuracy of the model still needs to be improved. Therefore, the sparrow search slgorithm(SSA) is proposed to optimize the AdaBoost-SVM model. The optimization range of the weak classifier weight αt, SVM penalty factor c and kernel parameter g is specified. The SSA is used to optimize the three parameters within the specified optimization range to improve the accuracy of the model. The proposed SSA-AdaBoost-SVM transformer fault diagnosis model is compared with PSO-SVM, SSA-SVM,AdaBoost-SVM, AdaBoost-SSA-SVM and PSO-AdaBoost-SVM. The proposed model has higher accuracy and stability, the average accuracy can reach 91.58%. The experimental results show that the proposed SSA-AdaBoost-SVM transformer fault diagnosis model has better performance.
作者
单亚峰
段金凤
付华
赵俊程
SHAN Ya-feng;DUAN Jin-feng;FU Hua;ZHAO Jun-cheng(Faculty of Electrical and Control Engineering,Liaoning Technical University,Huludao 125105,China)
出处
《控制工程》
CSCD
北大核心
2022年第2期280-286,共7页
Control Engineering of China
基金
国家自然科学基金资助项目(51974151,71771111)
辽宁省高等学校国(境)外培养项目(2019GJWZD002)
辽宁省高等学校创新团队项目(LT2019007)
辽宁省自然基金指导计划项目(20180550438)。
