As the new generation of power load management systems is gradually being implemented, there is an increasing demand for accuracy in power load forecasting. To meet this demand, a new load forecasting method of the power system based on multi-index clustering optimization is proposed in this paper. Firstly, the sensor technology is fully utilized to realize the data acquisition of the power system and the real-time monitoring of key parameters, providing more accurate and timely data for the load forecasting model. Secondly, the real-time data collected by the sensor is combined with the K-means clustering algorithm, which is improved, and the optimal number of clusters is selected by evaluating the clustering effect of multiple indicators. Then, an improved Particle Swarm Optimization-Back Propagation (PSO-BP) neural network prediction model is built by combining the ability of the BP neural network to solve complex nonlinear function approximation and the global optimization ability of the PSO algorithm. The model’s parameters are adjusted flexibly to improve the prediction accuracy and response speed. Finally, the model is compared with the BP neural network to verify the optimization effect. The experimental results of the simulation test show that the proposed method can effectively ensure the reliability of data transmission, realize the effective classification and aggregation of heterogeneous data, provide accurate load prediction results quickly, and improve robustness.