A Predict Method of Water Pump Operating State Based on Improved Particle Swarm Optimization of Support Vector Machine

Abstract

In order to improve prediction accuracy of water pump operating state, a chaotic prediction model of the pump vibration data based on improved particle swarm optimization of support vector machine is proposed in this paper. Firstly, a grouping optimization strategy particle swarm algorithm based on cosine function is proposed. Then, the training set is obtained on the time series of vibration data by phase space reconstruction. Secondly, The improved particle algorithm is used to optimize the penalty parameters, insensitive loss coefficient and width parameters of support vector machine. Then, a prediction model of vibration data is established by using support vector machine combined with training set and optimal parameters. Finally, the operating state of the pump is predicted according to pump vibration measurement and evaluation method. Compared with the method of linear decreasing weight strategy, the method proposed in this paper is more accurately.