The Research Of VRLA Battery SOC Estimation

Battery remaining capacity is one of the important performance parameters of thebattery, customarily is characterized by the state of charged (SOC). It is advantageous tothe battery diagnosis, maintenance, scientific and rational using by online monitoring SOCinformation, to find out capacity loss or failures. Timely getting rid of bad battery thatensure the safety of the normal power supply and prevent production accident. Thecomplex dynamic and static characteristics makes accurate estimate SOC become aresearch hotspot in the field of battery application.The paper, firstly, simply introduces the electrochemical principle of VRLA battery,summarizes currently physical models and SOC estimation models. On this basis, for theapplication in storage power station according to frequent charging and discharging workstate, the structure of the model is established based on the Thevenin circuit model; SOCmathematical model based on E-SOC model, also have be revised in two cases of chargeand discharge. For UPS, according to the batteries in the long-term floating charge, themainly concern discharge, it adopts simple circuit model, and V-R mathematic model ofSOC. Secondly, the paper introduces the single frequency AC impedance method toidentify a simple circuit model and dual frequency AC excitation method to identifyThevenin circuit model parameters. It also refers to the principle of waveletdenoising, dealing with internal resistance testing signal noise problems in AC impedancemethod, through the Matlab simulation experiment, gained good denoising effect.Using the Yuasa company VRLA battery as research object, the paper has carried outa series about battery fully charging and discharging characteristics experiments at25℃and collected a lot of online data, Using these data, obtained the specific SOC V-Rmathematical model by the least squares identification method, also established thespecific equation of E-SOC mathematical model. Finally combined with Ah measurement,it has established kalman filter algorithm Matlab program and made simulation.Thesimulation results show that the algorithm can effectively solve error accumulationproblem caused by the initial SOC. And SOC optimal estimation error is controlled withinthe range of8%, proved that the algorithm can be used for on-line estimation of VRLAbattery SOC. Accordingly, the algorithm has been embedded into the battery monitoringequipment PITE3920software system, and got practical application.