Power grid safety monitoring system based on machine learning algorithms

This article presents a power grid safety monitoring system based on embedded machine learning algorithms to improve the accuracy and real-time performance of power grid operations. A module for analysing and determining changes in optical channel performance was designed, and a long short-term memory (LSTM) model was used for analysis and prediction; a module for analysing and predicting the types of hidden danger degradation in optical channel performance was constructed, and a random forest model was used for identification and prediction. By integrating the outputs of the above modules using a cascaded model, the operational time of the optical channel was predicted. Thirty sets of comparative tests were conducted between traditional monitoring systems and embedded algorithms in the experiment. Experimental results showed that the embedded algorithm achieved anomaly detection accuracy of 89.1% to 99.2%, an error rate of 0.26% to 0.87%, and a response time of 0.71 seconds to 1.27 seconds, all of which were better than those of traditional monitoring systems.