Research on monitoring and prediction technology of distributed photovoltaic power generation based on grid clustering and meteorological group networking

In the context of global environmental protection and sustainable development, the installed capacity of distributed photovoltaic (PV) power plants has been rising. However, distributed PV output power is stochastic and fluctuating, posing a challenge to grid stability. Therefore, this paper proposes an innovative distributed PV power prediction technique that combines a Mark score evaluation model based on digital information maps with a power prediction model fused with LSTM and attention mechanism, so as to improve the accuracy of power prediction, Taking Heze City as an example, a regional prediction model is established and empirically analysed, and the results show that the overall average prediction accuracy is about 95.75%. The verified the effectiveness and accuracy of the proposed method. The research in this paper provides new ideas and technical means for the power prediction of distributed photovoltaic power plants, which has high theoretical value and practical application prospects, and helps to promote the sustainable and healthy development of China's photovoltaic industry.