The Aging Life Prediction Of Supercapacitors Based On Heuristic Kalman Filter Optimization Extreme Learning Machine

The aging of supercapacitors(SCs)would have a negative impact on the power system.To ensure the reliability of the energy storage system(ESS)and the power system,the steps that estimates the aging life of SCs in time and updates SCs before they are in the end of life(EOL)are important.In this thesis,a new kind of extreme learning machine(ELM)model is constructed to predict the aging state of SCs,in order to realize the life prediction of SCs.The aging models of SCs have many characteristics,such as multivariable,nonlinear,and high complexity.In order to reduce the research complexity for aging model,two parameters are introduced in the thesis,which are the equivalent capacitance and the equivalent series resistance(ESR).The equivalent capacitance represents the collection of healthy factors in ESS of SCs,and the ESR represents the collection of aging factors in ESS of SCs.By introducing two parameters,the characteristics of aging model of SCs are expressed as the decrease of equivalent capacitance and the increase of ESR,which simplifies the aging model and lays the foundation for the aging life prediction in later.In order to predict the aging life of SCs,a kind of data-driven model is used to predict the aging life of it.Based on the advantages of small iteration error,a smaller number of parameter,and the high accuracy of prediction of the optimal solution,the heuristic Kalman filter(HKF)algorithm could solve the singularity of the square matrix when the ELM randomly generates input weights and offsets,and improve the prediction accuracy of the ELM.Therefore,the ELM optimized by the HKF(HKF-ELM)could accurately predict the aging life of SCs.In order to verify the prediction accuracy of the model of HKF-ELM,the traditional ELM and the ELM optimized by the particle swarm optimization(PSO-ELM)are established to predict the same group of SCs,and the prediction results of three models are compared.The comparison results show that the HKF-ELM model has the higher prediction accuracy.Based on this situation,the thesis decided to predict the state of health(SOH)and the remaining useful life(RUL)of SCs under different working conditions by using HKF-ELM model.The results show that the higher temperature,the higher external voltage,and the greater charge-discharge current cause the faster speed of the capacitor aging,the greater SOH decay rate of the SCs and the shorter RUL of SCs.