A Genetic-Firefly Algorithm Based CNN-LSTM for Lithium-Ion Battery Fault Diagnosis

Lithium-ion batteries are a crucial component of new energy. Studying the fault diagnosis method of the batteries can ensure the safety of the system during operation. In this paper, a diagnosis method is studied based on the efficient channel attention (ECA) mechanism. The ECA is combined with convolutional neural network (CNN) and long short-term memory (LSTM) to design the diagnosis model. To determine the values of hyperparameters, a hyperparameter selection method based on genetic-firefly algorithm is implemented. The hyperparameter optimization methods are illustrated by the simulated battery fault data. The results show that the proposed method can make the classification accuracy reach 100%.