Aiming at the prediction of transformer health index (HI),this paper proposed an improved gray wolf algo-rithm (IGWO)optimized multilayer perceptron (MLP)prediction model based on principal component analysis (PCA) feature selection.Using the Kaggle public dataset as an example,the 15-dimensional features are reduced to 10-dimension-al features with higher correlation by PCA.The optimal parameters are obtained by optimizing the MLP using IGWO and further inputted into the MLP model to obtain transformer health index for estimation.The experimental results show that the IGWO-MLP has higher prediction accuracy than the GWO-MLP,IGWO-SVR,PSO-MLP and MLP models and is more effective in predicting the transformer health index.
关键词
变压器/健康指数/改进灰狼算法/混沌映射/多层感知机
Key words
transformer/health index/improved gray wolf algorithm/chaotic mapping/multilayer perception machine
维普
万方数据