Research on precise injection control algorithm of lithium battery based on improved Smith predictor

Abstract

Aiming at the problems of large overshoot and poor robustness of valve control in the battery during liquid injection, this paper proposes an improved Smith prognosticator model based on BP network. This paper firstly analyzes the influence of lithium battery electrolyte permeability on the liquid injection accuracy, secondly studies the key factors affecting the liquid injection valve time delay problem, and then constructs the battery liquid injection system model. The fuzzy PID in this system model is to ensure the accuracy of valve flow by accumulating the output variables of fuzzy control to make them infinitely close to the desired PID parameters. In the paper, the feedback link of the Smith prognosticator is improved by using BP neural network to fit the valve, and the fitted model is applied to the Smith prognosticator to solve the problems of delay error and robustness of the traditional Smith model. The test results show that compared with the traditional Smith predictor, the Smith predictor with BP network has the advantages of short adjustment time and small tracking error, which is suitable for complex and changing working conditions and has strong robustness, and can be well applied in the lithium battery injection system.