A neural network-based intelligent system for substation surveillance video analysis with edge and IoT integration

Substations are vital components of power infrastructure, and their security is crucial to the stable operation of the power grid. With the rapid development of Internet of Things (IoT) technologies and edge computing, substation monitoring systems are evolving toward greater intelligence and autonomy. Therefore, to improve the efficiency of anomaly recognition and differentiation in the substation monitoring system, this study proposes a video anomaly recognition algorithm that combines multi-instance learning optimized wavelet transform algorithm with long short-term memory network. Meanwhile, an intelligent analysis system for substation monitoring videos has been established. The results showed that the system exhibited high accuracy, with an overall accuracy value maintained between 90 and 100%, and no values below 90% were observed. The classification accuracy of the system for images was above 90%. This means that the method can more accurately classify surveillance video content into the correct categories. In summary, the intelligent analysis system for monitoring videos of the constructed substation has improved the safety management level of the substation and provided strong technical support for the automation and intelligent management of the substation. By incorporating emerging mobile computing technologies, the system ensures scalable deployment and enhanced adaptability in complex and dynamic substation environments.