Research on Multi-Source Data Fusion Optimization Dispatch Model for Smart Grid

This paper proposes an innovative multi-source data fusion optimization dispatch model based on the combination of DNN and PSO, aiming to achieve more intelligent and accurate grid dispatch. The deep neural network is used to learn and predict complex patterns from different data sources (such as load forecast data, renewable energy generation data, meteorological data, etc.), and the particle swarm optimization algorithm is combined to optimize the global dispatch decision, thereby improving the economy and reliability of grid dispatch. Finally, this thesis is verified by actual operation data of an intelligent grid. This set of data covers grid load, wind power and PV power generation data as well as weather factors over the last three years. Experimental results show that compared with traditional optimization method, dispatch efficiency is increased 12.5% and dispatch cost is reduced 9.3%. At the same time, the accuracy of load forecast can be improved by 16.8%. More importantly, this model has successfully optimized renewable energy dispatching, increased the utilization ratio of wind energy and PV power by 20%, and reduced fossil energy consumption effectively. Moreover, power system stability is significantly improved. The recovery time of system fault is reduced 25% compared with traditional method. These experimental results show that PSO-DNN multi-source data fusion scheduling model can significantly improve scheduling precision and operation efficiency, and it is suitable for large power systems.