Estimation Of Li-ion Battery State Of Health Based On Fuzzy Inference System

With the deterioration of the environment and the shortage of non-renewable resources, energy sources is transforming to the power from the traditional fuel gradually. Batteries as power application carrier are used more and more widely and the performance of the battery in the power system has been paid more and more attention. Capacity, charging time, reliability, production processes and cost are the essential factors to battery application. But in various of performance parameters of the battery, SOH is one of the most important parameters and it indicates the ability of battery storage capacity. Detecting of the battery state of health accurately and quickly, can avoid some unpredictable problems effectively. In the process of designing the battery SOH estimating system, this paper completes the following tasks:(1)This paper first introduces the research of li-ion battery SOH at home and abroad and the development trend of the research. Then we compare a variety of methods of estimating SOH, and finally propose a prediction method based on ANFIS. Fuzzy inference algorithm first through a lot of experimental data for training model, when the training results reached within the allowed error range, then estimates the output by the trained model.(2)We introduce the battery SOH estimation method using fuzzy clustering method ANFIS in detail. The process includes creating model, training model and using the trained model to estimate the battery SOH. This paper also study the deficiencies and shortcomings of basic ANFIS method, and improvements based on FletcherReeves method is proposed. The improvement not only can improve the accuracy of estimates, but also increases the estimated speed.(3)In order to verify the correctness of the algorithm, we design the SOH estimation system based on STM32 microcontroller. The system consists of two parts the data collection and the SOH estimation. The data acquisition process includes battery charging and discharging voltage, current and temperature monitoring, and the estimation is estimating the SOH value based on collected data using the trained model.(4)In the experiment, 2Ah and 100 Ah batteries are used for the test. Firstly we use one battery to charge and discharge, collecting experimental data for training model parameter, then we use the trained model to estimate the other battery SOH. After conducting constant current discharge test, in order to simulate the real battery used environment, we conduct the alternating current test. By the experiments of different models, different capacity and different test environment, the experimental result shows that the proposed scheme can predict the SOH of li-ion battery effectively.