Fault Diagnosis of ABS Wheel Speed Sensor Based on Improved BP Neural Network

To achieve the diagnosis of different types of faults in the wheel speed sensors for anti-lock braking system (ABS), BP neural network was introduced to establish a BP neural network diagnostic model. In order to address issues such as slow convergence and overfitting, an improved BP neural network based on L-M algorithm was proposed. Additionally, the concept of sensor voltage was introduced to train the network using 400 samples and diagnose faults such as open circuit, large gap, and small gap in wheel speed sensors. The improved BP neural network achieved an accuracy rate of nearly 100% for diagnosing different faults in wheel speed sensors, thereby enhancing the pattern recognition and diagnostic capabilities when ABS wheel speed sensors encounter various malfunctions. This study validates the feasibility and accuracy of using BP neural networks to diagnose faults in wheel speed sensors.