Fault Prediction and Fault-Tolerant of Lithium-ion Batteries Temperature Failure for Electric Vehicle

Design and implementation of dual-redundancy was developed to predict Lithium-ion battery failure for electric vehicle. Data fusion unit, prediction unit and determination unit were designed. Outputs from original and redundant sensors were integrated based on adaptive weighed fusion algorithm in the data fusion unit. Then, next prediction value was predicted with outputs from original and redundant sensors and their fusion data based on radial basis function neural network theory in the prediction unit. Finally, an optimal value was determined among outputs from original and redundant sensors and their fusion data and prediction values in the determination unit. Experiment and simulation test results showed that the prediction unit was able to predict next value from temperature sensors and the biggest error was less than 2.37%.