基于遗传算法与支持向量回归的发电机运行参数趋势预测

摘要

In the trend prediction of equipment operating parameters, the prediction accuracy will descend with the empirical parameters selection in support vector regression (SVR). A method that genetic algorithm (GA) uses to optimize the parameters of SVR (GA-SVR) is proposed in this paper. This method was used for the real time trend prediction of stator coil outlet temperature of the generator, results show that the prediction accuracy of GA-SVR was higher than SVR, and it can meet the requirement of generator operating parameters trend prediction accuracy in power plant.%针对支持向量回归(SVR)模型在设备运行参数趋势预测中.根据人为经验选取模型参数导致预测精度不高的问题,提出了一种使用遗传算法(GA)优化 SVR 模型参数的方法(GA-SVR).将该方法应用于发电机定子线圈出水温度的实时趋势预测中.结果表明,相较于SVR模型,GA-SVR具有更高的预测精度,能够满足电厂对发电机运行参数变化的趋势预测精度要求.