Parameters Aging Trend Prediction Of Supercapacitor

The aging state of the supercapacitor directly affects the life of the energy storage system.By predicting the aging trend of supercapacitor performance parameters,it provides predictive maintenance information for the control and management of the system,which will improve the reliability and stability of the system.Therefore,this paper analyzes the aging characteristics of the parameters by the supercapacitor aging test experiment,and constructs the prediction model based on PSO-SVM to predict the aging trend of the supercapacitor parameters.Firstly,according to the principle of energy storage of supercapacitor,the aging mechanism of capacitance and equivalent series resistance(ESR)is analyzed.The aging test platform is designed to investigate the effects of temperature,voltage and depth of discharge on the aging rate.Then,the AC impedance spectroscopy is used to quantify the aging characteristics of capacitance and impedance.The effect of "capacity regeneration phenomenon" on the life of supercapacitor is analyzed by testing the cycle characteristics of the supercapacitor after "regeneration phenomenon".Secondly,the implementation process of SVM on regression estimation is studied,and the influence of the parameters of the support vector machine on its learning ability and the promotion abilitythe is analyzed and further set up the process of using the particle swarm optimization algorithm to optimize the parameters of the support vector machine.The evaluation index of model prediction performance is also established.Finally,the training set data is substituted into the support vector machine,and the prediction model based on PSO-SVM is established.The aging trend of the super capacitor capacitance value and the equivalent series resistance value is predicted by the above model,and the root mean square error,square correlation coefficient and average absolute percentage error are used to evaluate the accuracy of the prediction results.The prediction results show that the average absolute error of the total sample capacitance and ESR prediction results is0.131% and 0.339%,respectively,and the prediction model has strong learning ability and generalization ability.In addition,the cube interpolation algorithm is used to extrapolate the aging trend of the capacitance value.The aging trend of the capacitance value with the number of cycles is shown in the form of the three-dimensional graph and this can provided a reference for predicting the aging trend of the capacitance value under t more working conditions.