A Terrain Adaptive Regenerative Braking and Motor Health Monitoring System for Electric Vehicles

This study details the engineering and validation of a control strategy designed to optimize Electric Vehicle (EV) regenerative braking across uneven terrains. Standard regenerative systems often react poorly to road anomalies, such as potholes, causing sudden torque cutoffs that result in mechanical shock and passenger discomfort. To mitigate this, a feed-forward intelligence layer was developed to adjust braking torque based on road conditions. The architecture employs sensor fusion, a combined data from a chassis accelerometer and a motor current sensor. In this work, a machine learning classifier processes these inputs to categorize terrain into smooth, rough, or pothole conditions. This study restricts its scope specifically to vertical mechanical roughness, excluding friction-based anomalies like ice. Upon classification, the system transmits a modulated control signal via the Controller Area Network (CAN) bus to the powertrain, enabling a proactive reduction in braking force. Additionally, a "Stress Index" metric was developed to act as a real-time proxy for electromechanical motor stress. Validation was conducted using a high-fidelity MATLAB/Simulink model and a hardware-in-the-loop (HIL) prototype, confirming the system’s ability to dampen braking transients.