Capacitance Degradation Curve Prediction of Supercapacitors Based on Maximum Curvature Point Using Deep Neural Network

Predicting capacitance degradation of a supercapacitor is important for energy devices, supercapacitor manufacturers, and customers. To predict capacitance degradation of a supercapacitor, model based and data-driven methods have been used. These models require large amounts of cycle data and have a disadvantage in computational complexity. To overcome this problem, a supercapacitor capacitance degradation prediction method based on maximum curvature point using a deep neural network (DNN) is proposed. This model uses only first cycle data as input of the DNN to predict the entire degradation curve. This model successfully predicts the degradation curve using first cycle data with average root-mean-squared error (RMSE), mean absolute percentage error (MAPE), and R squared score of 0.0056, 0.5893, and 0.9328, respectively.