Research on Fault Diagnosis and Prediction of Urban Intelligent Lighting System Based on Recurrent Neural Network
SU Huaiying
Abstract：In response to the problems of low efficiency and insufficient predictive ability in fault diagnosis of urban intelligent lighting systems,a fault diagnosis and prediction method based on recurrent neural network(RNN)is proposed to improve the intelligent level of system operation and maintenance.Firstly,construct a five layer system architecture based on smart lamp posts,integrating multi-source sensor data;Then analyze common types of faults and establish a fault feature library;Further design an RNN fault diagnosis and prediction model,utilizing its sequence modeling capability to process time series data;Train the network through backpropagation and gradient optimization algorithms,and introduce Dropout and early stop strategies to prevent overfitting.Through experimental comparisons with BP neural network models and LSTM methods,the results show that the RNN model has a fault diagnosis accuracy of up to 96.7%,a prediction lead time of up to 48 hours,and a mean square error(MSE)reduction of 32.5%compared to traditional BP networks.The advantages of RNN in capturing the temporal dependence of lighting system operation data,improving the accuracy and timeliness of fault diagnosis and prediction,and providing reliable technical support for intelligent operation and maintenance of urban lighting systems have been verified.
Keywords：intelligent lightingfault diagnosisfault predictionrecurrent neural networktime series analysis
Publication Date：2026-02-25
Online Publishing Date：2026-03-12(First online date of this platform, not the publication date of the document)
Pages：7( 42-47,58 )