【摘要】 针对变压器油击穿电压在线测量困难,提出核主元分析(KPCA)和模糊C均值聚类(FCM)的变压器油击穿电压预测模型。首先,通过KPCA提取输人数据的非线性主元;然后采用FCM将提取的主元集分成具有不同聚类中心的子集,同时,采用差分进化算法对KPCA核参数和FCM聚类数寻优,分别为每一子集建立最小二乘支持向量机(LSSVM)子模型;最后通过子模型切换策略得到模型的最终输出。实验结果表明,提出的预测模型具有较好的泛化能力和预测精度。更多还原

【Abstract】 To aim at the difficulty of measuring the breakdown voltage of transformer oil in real time,a modeling method combined kernel principal component analysis( KPCA) and fuzzy C-means clustering algorithm( FCM)was proposed based on the idea that the combination of multi-models can improve the accuracy and robustness of model. Firstly,the correlation of the input data was eliminated and the nonlinear principal components in input data were captured by using KPCA; Then,fuzzy C-means clustering algorithm( FCM) is used to separate the nonlinear principal components data set into several clusters with different centers,each sub-set is trained by least squares support vector machines( LSSVM) and sub-models are developed; Finally,the model output was obtained through the sub-models switch strategy. Experimental results show that the proposed model has higher accuracy and better generalization ability.更多还原