High-accuracy road surface condition detection through multi-sensor information fusion based on WOA-BP neural network

High-accuracy road surface condition detection is crucial for enhancing traffic safety. However, existing methods struggle to comprehensively cover various road surface conditions, and the detection accuracy still requires improvement. In this study, we propose a novel road surface condition detection system integrating the whale optimization algorithm-enhanced backpropagation neural network (WOA-BP) with multi-sensor data, which significantly improves the detection accuracy. An experimental system was established, and the model's performance was validated through multiple experiments across a temperature range of -30 degrees C to 50 degrees C. The results demonstrate that, compared to traditional machine learning algorithms such as back propagation neural network (BPNN), support vector machine (SVM), and random forest (RF), the WOA-BP neural network model shows superior performance in terms of detection accuracy and model stability. Specifically, it achieves a maximum accuracy rate of 100 % and an average accuracy rate of 98.8 % for detecting dry, wet, icy, and snowy conditions. By employing the WOA to refine the initial weights and thresholds of the BPNN, the issues of susceptibility to local optima and slow convergence are addressed, thereby enhancing the model's classification performance and robustness. Furthermore, the collaboration between the capacitive measurement-based road condition sensor, temperature sensor, and microwave measurement-based water film thickness sensor further enhances detection accuracy. This research delves into high-accuracy road condition detection technology utilizing the WOA-BP neural network, offering valuable insights for improving traffic safety and developing intelligent transportation systems.