优化WOA-LSSVM的滚动轴承性能退化趋势预测

Prediction of Rolling Bearing Performance Degradation Based on Reverse Learning WOA-LSSVM

为了更加准确的预测滚动轴承性能退化趋势,针对最小二乘支持向量机模型中参数选择盲目的问题,利用鲸鱼优化算法对最小二乘支持向量机模型(WOA-LSSVM)参数进行寻优.引入反向学习算法,有效提高标准WOA算法中初始群体位置的质量,更易找到最优解,从而提高WOA算法的收敛精度和收敛速度.然后采用实测滚动轴承全寿命实验数据进行仿真,选择PCA第一主成分作为特征指标的最小二乘支持向量机模型预测滚动轴承退化趋势.结果表明基于反向学习的WOA-LSSVM模型与WOA-LSSVM模型、粒子群算法优化最小二乘支持向量机模型、交叉验证算法优化最小二乘支持向量机模型和固定参数的最小二乘支持向量机模型比较,具有更好的预测精度,可用于滚动轴承性能退化趋势预测.

In order to predict the trend of rolling bearing performance degradation more accurately, the parameters of least squares support vector machine(SVM) model were used to blindly select the optimal parameters of the least squares support vector machine(WOA-LSSVM). The introduction of the backward learning algorithm can effectively improve the quality of the initial group position in the standard WOA algorithm and make it easier to find the optimal solution, thereby improving the convergence accuracy and convergence speed of the WOA algorithm. Then the experimental data of rolling bearing life were used to carry out simulation experiments. The least squares support vector machine model with PCA first principal component as the characteristic index was used to predict the degradation trend of rolling bearing. Compared with WOA-LSSVM model, particle swarm optimization based on least squares support vector machine model, cross validation algorithm optimized least squares support vector machine model and fixed parameters least squares support vector machine, the results show that WOA-LSSVM model based on backward learning has better prediction accuracy and can be used to predict the trend of rolling bearing performance degradation.