混合核函数相关向量机在污垢预测中的应用

Application of Hybrid Kernel Function Relevance Vector Machine in Fouling Prediction

污垢的热阻反映了污垢结垢的程度,准确预测污垢热阻,为污垢监测以及污垢对策提供了重要参考依据。近年来,混合核函数相关向量机开始用于预测变压器顶层油温。将基于混合核函数的相关向量机应用到污垢预测研究中,介绍混合核函数相关向量机的基本原理,通过在线监测装置,采集样本数据,进行仿真实验。结果表明混合核函数相关向量机能准确预测污垢热阻值,相对于单一核函数相关向量机,混合核函数相关向量机的预测精度更高。

Fouling resistance reflects the scale of the fouling. The precision of fouling prediction provides some important references for fouling monitoring and fouling countermeasure. Recently,the hybrid kernel function relevance vector machine(RVM)has been applied to oil temperature prediction for transformers.In this paper, the hybrid kernel function RVM is applied to the research of fouling prediction.The basic principles of the hybrid kernel function RVM are described . According to the online monitoring devices,the sample data is collected and the simulation is carried out. The simulation result shows that the hybrid kernel function RVM could accurately predict the fouling resistance.Comparing to the single kernel function,the precision of prediction is increased.