基于CPSO-LSSVM的陀螺仪故障趋势预测

Gyroscope Failure Trend Prediction Based on CPSO-LSSVM Algorithm

为了提高最小二乘支持向量机(LSSVM)的学习性能和泛化能力,提出了混沌粒子群优化(CPSO)算法和交叉验证(CV)算法相结合的LSSVM参数寻优方法。CPSO算法将混沌搜索引入到粒子群算法中产生初始混沌粒子,并在粒子运动中不断加入混沌扰动,实现LSSVM参数的自动选取。利用交叉验证误差构造粒子的适应度函数,为参数选择提供评价标准。陀螺仪随机漂移是影响陀螺仪性能可靠性的主要因素,将经过参数寻优的LSSVM用于建立陀螺仪随机漂移的时间序列预测模型,预测值与实际值相差较小,可为陀螺仪的故障趋势预测提供依据。实验结果表明CPSO算法是选取LSSVM参数的有效方法,所建的回归模型具有较高的预测精度。

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.