Multi-Sensor Fusion Techniques for Monitoring Emissions and Health of New Energy Vehicle Powertrains

Abstract

The rapid development of new energy vehicles (NEVs) has intensified the demand for accurate and reliable monitoring of both emission-related performance and powertrain health. In contrast to traditional vehicles, NEV powertrains are characterized by a high degree of interconnection between electrical, thermal, mechanical, and chemical subsystems that results in complicated dynamic behavior and degradation behavior. The conventional single-sensor or isolated-model monitoring systems are becoming less effective in dealing with the challenges during real-world driving conditions. Multi-sensor fusion in this regard has become one of the key enabling technologies to achieve higher levels of observability, robustness, and diagnostic capability in NEV powertrain monitoring. The paper gives an in-depth review of multi-sensor fusion that is being used to monitor emissions and powertrain conditions in NEVs. The review initially examines the monitoring requirements that are linked to various NEV architectures and sensing technologies, and the heterogeneous and uncertain character of multi-sensor data. Significant sensor fusion strategies, such as model, data, and hybrid strategies, are also presented systematically about their theoretical basis, implementation architecture, and trade-offs. The applications in emission monitoring and health management in powertrains are then considered, including real-time estimation of emission-related factors, fault detection, state estimation, and prognostics. Other critical issues in terms of data quality, computational requirements, strength, and explicability are also underscored. The synthesis of existing research advances and the uncovering of gaps are what make this review give a single view on the importance of multi-sensor fusion to develop intelligent, reliable, and sustainable NEV powertrain monitoring systems.