Prediction of battery manufacturing capacity based on reinforcement learning network combination model

Abstract

Aiming at the problem of the battery manufacturing capacity prediction, this paper presents a prediction method based on reinforcement learning network combination model. First, the combined model expression for the battery manufacturing capacity prediction is designed. Then, reinforcement learning is used to construct the hidden layer learning environment of recurrent neural network and long-short-termmemory network model, to obtain the optimal number of hidden layers, and then to construct the weight learning environment of the battery manufacturing capacity combination prediction model and a combined forecasting model of battery manufacturing capacity after iterative training. Finally, a case simulation on actual battery workshop data shows the effectiveness and practicability of the proposed algorithm on solving the battery manufacturing capacity prediction problem.