Health Degree Prediction of DCS Equipment Based on WOA-BP Neural Network

The health degree of the Data Communication Subsystem (DCS) in urban rail transit directly impacts train operation safety and efficiency. This paper proposes a Whale Optimization Algorithm (WOA)-optimized BP neural network (WOA-BP) prediction model for DCS health degree prediction to support intelligent operation and maintenance of DCS equipment. First, two health degree prediction modes are proposed: direct prediction of overall DCS health degree (Mode 1) and separate prediction of health degree for each sub- equipment category (e.g., access point, core network, communication security, antenna feeder) followed by weighted fusion (Mode 2). Second, addressing the sensitivity of BP neural networks to initial weights and thresholds, the WOA algorithm is introduced for optimization to enhance prediction accuracy. Simulation validation is conducted using DCS health degree from a specific line. Results demonstrate that the WOA-BP network model achieves optimal performance in direct prediction of overall DCS health degree (i.e., Mode 1), and its performance surpasses both the unoptimized traditional BP neural network and Mode 2. This study validates the applicability of the WOA-BP network model for DCS equipment health degree prediction, providing valuable insights for DCS equipment condition forecasting and preventive maintenance.