Prediction of Remaining Life of Lithium-ion Battery Based on Improved Sparrow Algorithm

Lithium-ion battery has been widely applied during daily life, and unstable state may cause system operation failure or accident. The accurate prediction of its remaining life is beneficial for maintenance and management, making the use of lithium-ion battery safer. The ssa-lstm (sparrow search algorithm based on long short-term memory) model is able to reduce the difficulty in adjusting the hyperparameters, offering a good prediction ability. However, it has strong local optimization capabilities, but weak global search capabilities, and is easy to fall into the local optimum. To predict the remaining life of lithium-ion battery more precisely, an improved sparrow algorithm is proposed in this paper. Aiming to solve the problems coming from the basic sparrow algorithm such as uneven distribution and individual repetition, the good node set method, crossover operator of genetic algorithm and cultural evolution framework are utilized to facilitate algorithm stability, convergence speed and calculation accuracy. The simulation experiments demonstrate that the proposed algorithm is superior in global search ability and local optimization accuracy.