Research On Estimation Of Residual Capacity Of Lead-acid Battery Based On Neuro-Fuzzy Network

Because of mature manufacturing technology and high security of lead-acid battery, it has been widely used in the fields of transport, energy, communications etc.The residual capacity of lead-acid battery state-of-charge (SOC), have an important effect on the safe and stable operation of the battery.The premise condition of battery use is to obtain SOC of battery timely, accurately. How to obtain the SOC of battery accurately and effectively is one of the tasks of battery management system, and it has became a focus of current research. Compared the advantages and disadvantages of several common measuring method of SOC of battery in accuracy and universality, an universal estimation strategy was selected.The neuro-fuzzy network which combined the merits of neural network and fuzzy reasoning can improve the processing ability of complicated system effectively, which meet the demand of the thesis. The estimation strategy of lead-acid battery SOC was based on neuro-fuzzy network, and has performed estimated research.This thesis chose the estimation strategy of SOC of lead-acid battery based on neural-fuzzy network, and established the corresponding prediction mode to make the estimation research of SOC of lead-acid battery.Lead-acid battery SOC is affected by many factors. In order to improve the precision of forecasting model, the choice of input variable of the prediction model is very important. On the base of the working principle and performance of battery, some parameters ware chosen as the input variables of the model which are easily measured and connected with the SOC of battery and independent of each other. In order to obtain the actual sample data to establish model, physical discharge experiments for lead-acid battery was done. The real-time value of battery voltage, discharge current, battery internal resistance as well as the SOC of battery were collected in different load cases, and the data were pretreated by affine propagation clustering algorithm, which was used for model establishment.In the simulation condition, the SOC prediction model based on neural-fuzzy network of lead-acid battery was establish. Battery terminal voltage, discharge current and battery internal resistance was considered as the network's input variables which were used to predict SOC of battery. After the simulation analysis, application of neuro-fuzzy network method can effectively improve the estimation precision of SOC of batteiy, and the model has strong generalization ability and wide applicability. In the estimation process of the SOC of battery, the introduction of battery internal resistance and combining with external battery parameters provided a new idea for the estimation of the SOC of battery.