Real-Time lifetime prediction method based on wavelet LS-SVR Optimized by GA

Some products have nonlinear performance degradation paths. As for as the comparability of degradation paths is concerned, a real-time lifetime prediction method is proposed based on wavelet least square support vector regression (LS-SVR) which is optimized by genetic algorithm (GA). The Euclid distances of the specific individual and the same kind of products are used to determine the degree of membership. The specific individual degradation path model is built by the weighting models of same kind products. It is updated with real-time measurement. The proposed method is applied to fatigue crack growth data. Experimental results verify its validity.