Intelligent communication quality prediction model for Internet of Things in power systems based on multisource optical-electrical perception fusion

The Internet of Things in Power Systems (IOTIPS) is built on a high-speed bidirectional communication network, achieving reliable, safe, economical, and efficient operation of the power grid through advanced sensing technology, control methods, and decision support. With the growth of power communication networks, online status prediction of communication devices in the network is of great significance for improving operational reliability. This article combines optical signal processing technology to design an IOTIPS intelligent communication quality prediction model based on multisource perception fusion. Firstly, the study adopts the low rank autoregressive tensor completion (LATC) algorithm to integrate multi-source heterogeneous data of power flow and information flow transmitted back to the terminal, combining optical signal processing technology to extract key degradation features. Furthermore, the extreme learning machine (ELM) algorithm is used to construct the connections between data as data features, and output a feature set to complete data feature level fusion. On this basis, a link bandwidth occupancy prediction model (LBOP-GCN) constructed by graph convolutional neural network (GCN) was studied and designed to predict the communication quality of IOTIPS. Simulation experiments show that the model can improve the prediction precision of IOTIPS communication quality and provide strong guarantees for the stable operation of power communication.