Mechanical and electrical device condition trend prediction based on GA-SVR

摘要

This paper mainly discuss three kinds of optimization method to get the optimal penalty factor C and kernel parameter G of support vector regression. the mean square error MSE, correlation coefficient R, the number of support vector nsv was regarded as indexes to measure the merits of the various optimization prediction model, the experimental results shows that the prediction model based on genetic optimization is closer to the actual value in the prediction of vibration intensity, and prediction performance is better than other optimization methods. It also shows the prediction model has a good predictive ability on the condition trend of mechanical and electrical device.