基于改进粒子群算法的避雷器状态预测

Prediction of Lightning Arrester Condition Based on Improved Particle Swarm Algorithm

避雷器状态监测在变电站中起着重要作用。基于深度信念网络对避雷器输入样本集进行降维处理,使输入样本维度得到优化;在降维后的数据基础上,利用支持向量机建立避雷器健康状态预测模型,并通过改进粒子群算法对该组合模型参数进行迭代寻优,进而得到避雷器监测状态的变化趋势。实验表明,该改进的组合模型相较于直接对数据进行预测有了很大的提升,可提高避雷器预测精确度,为研究避雷器健康状态提供了可能,为运行人员提供可靠地指导。

The condition monitoring of arrester plays an important role in substation.In this paper,the input sample set of lightning arrester is reduced its dimension based on deep belief network(DBN).The dimension of input sample is optimized.Based on the reduced dimension data,a health state prediction model of arrester is established by using support vector machine(SVM).The improved particle swarm optimization algorithm is used to optimize the parameters of the model iteratively,and the variation trend of the arrester monitoring state is obtained.Experiments show that the improved combined model can improve the prediction compared with the direct prediction.The prediction accuracy of the arrester is improved,and it is possible to study the health state of the arrester,and provide reliable guidance for operators.