Prediction Of Battery Remaining Discharge Energy Oriented For Remaining Driving Range Estimation Of Electric Vehicles

Owing to the excellent performance in driving economy and environmental friendliness, electric vehicles(EVs) have attracted significant attention in automotive industry. However, the EV driving range is not competitive compared with conventional vehicles due to low energy density and high cost of vehicle battery. Additionally, the estimation error of EV remaining driving range(RDR) could not be neglected owing to inaccuracy in battery state estimation and leads to range anxiety of passengers. As a result, an accurate RDR depends on the precise prediction of battery remaining discharge energy(ERDE).This dissertation focuses on the prediction method of lithium-ion battery remaining discharge energy, which is oriented for EV remaining driving range estimation. As the basis of battery remaining energy prediction, battery modeling issue is firstly discussed. Based on electrochemical analysis, a real-time applicable battery model is developed with good accuracy in the whole SOC range. As battery ERDE is affected by future temperature variation due to heat generation, a real-time battery temperature prediction model is required. The battery ERDE prediction method is then studied based on predictive control theory, which is implemented to develop an EV remaining driving range estimation model and embedded in a type of pure electric vehicle for performance validation.Firstly, an accurate onboard battery model is required in the whole SOC range. Based on single-particle electrochemical model, an electrochemistry-based equivalent circuit model(EECM) is presented, in which the solid-phase diffusion process is represented by SOC difference within the electrode particle. The voltage estimation performance of traditional ECM and EECM is compared in the entire SOC area(0~100%) under different load profiles. Results imply that the EECM could effectively reduce the voltage error in low-SOC area. The EECM-based SOC estimation accuracy is then discussed employing extended Kalman filtering.Secondly, a real-time battery temperature prediction method is developed. The battery heat generation is comprehensively analyzed considering various factors, including temperature, SOC, battery aging, and current input. The future temperature variation is then prediction based on a real-time thermal model.Thirdly, the prediction method of battery remaining discharge energy(ERDE) is studied. Based on predictive control theory, a predictive-adaptive energy prediction method(PAEPM) is presented, in which a coupled prediction of future battery state variation, battery model parameter change, and voltage response, is implemented on the ERDE prediction horizon, and ERDE is subsequently accumulated and real-timely optimized. Three EPM approaches with different model parameter updating routes are evaluated, and the PAEPM combining real-time parameter identification and future parameter prediction offers the best potential. Under various dynamic profiles, the performance of different ERDE calculation methods is compared, and the correlation of SOC estimation and ERDE calculation is discussed to show the importance of accurate ERDE method in real-world applications.Lastly, a remaining driving range estimation model(RDREM) is developed for EVs based on battery ERDE prediction model. The RDREM is then integrated in a type of mass-production EV, while the estimation performance is validated under different driving conditions.