Network Scheduling Optimization of Distributed Photovoltaic Power Generation Platform Based on Intelligent Algorithm

Abstract

The traditional distributed photovoltaic power generation platform scheduling faces the challenges brought by the volatility and intermittency of photovoltaic power generation. It is difficult to balance the supply and demand of the power grid in real-time and to cope with sudden changes in light. The existing scheduling methods are slow to respond, which can easily cause grid instability and waste of photovoltaic resources. Therefore, this paper applies a scheduling method based on the improved GWO (Grey Wolf Optimizer) to improve the adaptability and real-time performance of scheduling and overcome the shortcomings of traditional methods. First, a photovoltaic power generation prediction model is established by combining historical power generation data and regional meteorology, and RF (Random Forest) is used to predict future photovoltaic power generation. By correcting the predicted value, the impact of the volatility of photovoltaic power generation on scheduling is reduced, and the prediction error is reduced by using Kalman filtering. The GWO algorithm and chaos search are combined to enhance global exploration and reduce local optimal risks. According to the demand for photovoltaic power generation and power grid load, the coordinated scheduling of photovoltaic units is optimized to avoid overload. By adjusting photovoltaic output in real-time, load fluctuations can be coped with to ensure stable operation of the power grid. Studies have shown that the average utilization rate of photovoltaic resources under network scheduling optimization reaches 93.5%, and the power generation cost is reduced by about 9.24%, indicating that the applied method provides a new idea for optimizing photovoltaic power grid scheduling.