State prediction of hydro-turbine based on WOA-RF-Adaboost

Abstract

Aiming at the problem that the prediction performance of remote operation state of hydro-turbine needs to be improved, this paper uses whale optimization algorithm (WOA) to optimize random forest (RF) . It combines it with Adaboost algorithm to propose the prediction model in this paper. Firstly, the signal of hydro-turbine is analyzed by variational modal decomposition (VMD), and the penalty factor and the number of IMF components K of VMD are optimized by fruit fly optimization algorithm (FOA). The arrangement entropy of intrinsic mode function (IMF) components, kurtosis, mean value of original signal are calculated, and the input eigenvector of hydro-turbine state prediction model is constructed; Secondly, the number of split attribute sets and the optimal number of decision trees of RF are optimized by WOA, and multiple WOA-RF models are iteratively trained by Adaboost algorithm to construct WOA-RF-Adaboost state prediction model. The prediction effect of the proposed model and the traditional model is evaluated by correct rate and confusion matrix. The results show that the prediction accuracy of WOA-RF-Adaboost model proposed in this paper is 99.2%, it has good state prediction performance.