基于深度学习的斜拉桥速度传感器信号异常检测方法研究OACHSSCD

The rapid development of sensor technology has made it possible to conduct comprehensive and continuous health monitoring of cable-stayed bridges.However,in practical engineering applica-tions,sensors are inevitably affected by various factors,leading to potential abnormal fluctuations or systematic biases in the collected data,which may result in abnormal fluctuations or systematic devia-tions in the collected data.Therefore,developing reliable sensor signal anomaly detection method has important theoretical significance and engineering value for the safe operation of bridges.This paper generates five common types of sensor faults based on normal speed signal data collected by DH5907N to simulate different types of fault signals.Specifically,these types of faults include offset,drift,accu-racy degradation,gain variation,and complete failure.When creating the dataset,80％of the collected speed sensor signal data(including normal and faulty)will be used for network training,and the remai-ning 20％will be extracted for network testing.Comparing the model proposed in this paper with Long Short-Term Memory(LSTM)and Gated Recurrent Unit(GRU)model,and considering various fault fu-sion scenarios.The experimental results show that the deep learning method constructed in this paper has high accuracy in detecting anomalies in speed sensor signals and good noise resistance.