基于片段数据的储能电池SOH估计OA北大核心

To improve the accuracy and engineering adaptability of State of Health(SOH)for energy storage batteries,firstly,by analysing the operating data of a certain solar energy storage power station,design simulated operating conditions experiments.Based on the characteristics of the data and aging mechanism,the charging voltage difference within 30 minutes before 3.41 V(with time intervals of 1 minute,3 minutes,and 5 minutes respectively)was selected as the characteristic parameter for evaluating the SOH of energy storage batteries.Using the cyclic data of a 20 Ah lithium iron phosphate battery and combining it with a genetic algorithm(GA)improved the back propagation(BP)neural network for modelling,the accuracy of the model was verified using 1200 sessions of data that were not involved in model training.Among them,the model with a 1-minute interval voltage difference as the characteristic parameter had the highest accuracy,and the mean absolute percentage error(MAPE)was 0.37%,the root mean square error(RMSE)is 0.456 5.Secondly,model transfer reduced the maximum error in estimating SOH for 260 Ah lithium iron phosphate batteries from 5.52%to 1.89%.Finally,the model was used to perform batch SOH estimation on a cluster of energy storage battery cells in a photovoltaic power station,with good engineering adaptability.