배터리 내구 성능을 고려한 전기차 배터리의 에너지 관리 최적화

Abstract

In this study, in preparation for the rapidly changing power demand and the saturation level of charging stations according to the demand for electric vehicles in the future, an optimal control simulation that can minimize battery degradation for various parking times was constructed. And its performance was compared with existing strategies. As a result, it was possible to mathematically prove the performance of the V2G strategy which to delay the charging time suggested in previous studies. And optimal strategy has presented charging strategy varying on SOC section and optimal charging starting point between fast charging and slow charging. In addition, the simulation had set to be able to adjust total parking time, so that various charging scenarios that change rapidly depending on the charging infrastructure and charging demand can be implemented. Through this, it was confirmed that the PMP theory commonly derives the optimal charging time without additional changes to the algorithm even at various parking times. In addition, it was confirmed that the optimality of PMP control was ensured by comparing the performance degradation for each parking time. Comparing results based on the longest parking time of 12 hours, it was concluded that up to 30% improvement in the optimal control strategy compared to the existing strategy can be expected. But those improvements were not for the overall vehicle lifespan but limited for charging situation while parking. Accordingly, as a result of creating an electric vehicle driving scenario based on statistics, it was possible to obtain a rather realistic figure of 4.6% improvement in overall lifespan. The reason that this paper could present the improvement as a specific numerical value is because the formula expressing the electric vehicle battery and its deterioration went through several verifications before being applied to the simulation. The battery model was constructed by referring to the Thevenin equivalent circuit model based on the chassis dynamometer test data of Hyundai Motor's 2019 IONIQ EV model, and its verification has been completed. Similarly, the battery degradation equation was also constructed by the experimental data and Arrhenius equation, and its verification has been completed through the paper. However, at the same time, it also means that the reliability of the model must be high for the accuracy of the performance index provided by the algorithm. Therefore, it means that information on the specifications of all electric vehicles connected to the power grid and the degradation model of their batteries is required for the wide range of versatility in the large-scale power grid unit, which is the goal of this study. In the case of battery specifications, it is not a big problem because the battery manufacturer can estimate the exact value in a much more direct way at the development stage without the need for an experiment through the chassis dynamometer. However, in the case of battery degradation, the time and cost required for the experiment are very large. And according to some research cases, there is some difference between operation in experiment through chamber and actual electric vehicle, making it difficult to predict the actual lifespan. Fortunately, the number of cell products that need to be measured for lifespan is smaller than the total number of electric vehicle models, thanks to the well-established standardization of EV-only platforms for vehicle manufacturers and cell products for battery manufacturers. However, in the end, the battery life model has a weakness that a large amount of data is required for its accuracy because it requires a prediction of more than 10 years. Therefore, in this paper, technologies such as the Internet of Things and remote software update, which are currently attracting attention, are proposed as improvements. Currently, in the automobile industry, based on the development of communication technology such as 5G, there is a movement to collect information generated from a vehicle and use it for analysis of driver's behavior pattern and software performance evaluation. Accordingly, it is expected that lifespan modeling data based on actual vehicle use can be obtained without experimental cost if data on battery usage patterns and performance reduction can be collected with the development of communication technology in the future|최근 몇 년간 차량의 연비 및 온실가스 배출에 대한 규제는 거듭하여 강화되어 왔으며, 이러한 경향은 앞으로도 유지되거나 더욱 가속화될 예정이다. 이에 따라 상대적으로 주행 중에 온실가스를 배출하지 않는 전기자동차의 수요가 급증하게 되었고, 세계의 여러 자동차 기업들은 앞다투어 기존 차량의 전동화 및 전기차 전용 플랫폼의 개발 등에 투자하고 있는 실정이다. 그에 맞게 전기자동차만의 개선 사항들이 새로운 연구 주제로 요구되어 왔으며, 가장 대표적인 것이 바로 주행거리와 충전 시간, 그리고 배터리의 수명과 교체 비용의 문제라 할 수 있다. 이 중에서 주행거리와 충전 시간 문제의 경우 사용자에게 즉각 다가오는 단기적인 해결 과제였고, 이에 따라 최우선적으로 열관리, 배터리 기술 발전 등의 다양한 방법으로 상당 부분 개선되어 왔다. 그리고 현재 전기자동차 업계는 장기적 해결 과제인 배터리의 수명 문제를 개선해야할 상황에 마주하였으며, 이는 본격적인 전기차 판매가 시작되고 시간이 지날 수록 운전자들이 성능 감소, 배터리 교체 비용 등의 형태로 체감하게 될 사안인 만큼 점점 그 중요성이 증가하고 있다. 이에 따라 본 논문은 기존의 실험에 기반하여 제시된, 리튬 이온 배터리의 전력 사용량과 외부 환경에 따른 열화 모델을 Simulink 및 MATLAB 함수로 구현하였으며, 이러한 모델링과 PMP 이론에 기반하여 배터리의 충전량과 열화 진행도를 등가화하는 해밀토니안을 계산함으로써 최적의 충전 전략을 도출할 수 있도록 하였다. 또한 대한민국 평균 주행거리에 기반한 전기차 주행 시나리오를 설정함으로써, 최적제어 이론을 통해 계산해낸 충전 전략이 실질적으로 전체 전기차 수명에 끼치는 영향이 얼마나 되는지를 분석해보았다. 이 때문에 결론적으로는 본 연구를 통해 추후 스마트 그리드, V2G 등의 전력망 관리 시스템이 도입되었을 때 전력 수요를 만족하는 동시에 전기차의 수명을 최적화하는 제어 시스템의 레퍼런스를 제공할 수 있게 될 것으로 기대된다.