Remaining Lifetime Prediction of Lithium Battery Based on Support Vector Machine Kalman Filter

This proposed lithium battery remaining useful lifetime prediction model, which combines support vector machine (SVM) and nonlinear Kalman filter, aims to address the limitations of existing models that cannot use both database data and real-time degradation data simultaneously. The SVM regression model uses the existing lifetime-cycle test data, and the nonlinear Kalman filter state update equation is established based on it. The time update equation is constructed considering the degradation characteristics of lithium batteries. The initial residual lifetime value and its variance are set, and the residual lifetime estimation value at each time along with the confidence interval of a certain confidence level are calculated through stepwise iteration. This calculation model effectively utilizes the full lifetime test data of existing and similar batteries, as well as the real-time state degradation data of the predicted batteries to achieve residual lifetime prediction. Using a certain type of rolling bearing as an example, the accuracy, stability, and engineering application value of the proposed residual lifetime prediction model have been demonstrated.