Gyroscope Failure Trend Prediction Based on CPSO-LSSVM Algorithm

In order to improve the learning performance and generalization ability of the least squares support vector machine (LSSVM),Chaos particle swarm optimization algorithm (CPSO) combined with k -fold cross validation (CV) was proposed for selecting the optimal parameters of LSSVM.Chaotic search was introduced to PSO algorithm to generate the initial chaotic particles and chaotic interrupt was added to the particles in the motions for the selection of LSSVM parameters automatically.CV error was used to construct the fitness function of particles as the assessing criteria of the LSSVM parameters choice.Gyroscope random drift was the main factors affecting the reliability of the gyroscope performance.LSSVM regression model based on CPSO-CV algorithm was used to establish time series prediction model of gyroscopes random drift for gyroscope failure trend prediction.The results showed that the proposed method is an effective approach for LSSVM parameter selection and the regression model has a better prediction precision.