Study On SOC Estimation Of Lithium Iron Phosphate Battery In Energy Storage Power Station Based On KF-ESN Algorithm

In the new energy power generation, the battery energy storage technology can make the time to develop electricity and power transmission, so as to ensure that the power output can be smooth, stable output to the power grid. In the battery energy storage system, battery management is essential. The battery’s SOC estimation is part of the battery management system, The SOC of the battery can directly indicate the degree of charge and discharge of the battery, so as to prevent the battery from over charging and discharging, and accurate SOC can also be used as a battery charge and discharge control and other management of the important basis for battery equalization, for electric vehicles, the battery SOC can accurately reflect the continued driving range. Because the battery SOC cannot be obtained by direct measurement, and the battery internal chemical changes in the work process is very complex, very difficult to estimate. So the battery’s SOC estimation is the key and difficult point of the battery management system.The research on SOC estimation algorithm of battery has always been a hot topic, The neural network algorithm has received more and more attention because it does not rely on accurate battery model, high precision, fast speed and so on. But for the traditional neural network algorithm, the existence of complex computing, long learning time and poor generalization ability for small samples, data into a local minimum even can not be trained for large samples. Propose a SOC estimation method which with Kalman Filtering algorithm (KF) to optimize the reserve pool network Echo State Network (ESN) output weights to solve the problem above-mentioned. Take the large capacity lithium iron phosphate battery (3.2V/180Ah) in the laboratory as the research object, Constant current discharge of the battery by the charging and discharging experiment platform. The characteristics of the battery are analyzed by using the data obtained. Then determine the input and output quantity and establish the SOC prediction model of the battery. Respectively, the SOC of the battery with different current constant current discharge is estimated, Compared with the traditional BP neural network prediction model, it is found that the accuracy of KF-ESN prediction model is higher, and the operation speed is faster, Can be used as a new SOC estimation algorithm.Research on battery management system of scale energy storage power station. In the battery management system, the function of the monitor system of the host computer monitor system comes with the Battery Cluster Management System (BCMS) is monotonous, simple and can not be remote monitoring, so designed and developed a monitoring system based on MCGS touch screen. The monitoring system features a rich, smooth appearance, and through the Modbus protocol and serial communication, remote real-time monitoring the running state of storage battery power station.