Short-term photovoltaic power prediction method for microgrid based on cda-bp

Abstract

The invention relates to a short-term photovoltaic power generation power prediction method for a microgrid based on CDA-BP, and belongs to the technical field of photovoltaic power generation. The randomness and volatility of photovoltaic power generation make it difficult to achieve ideal accuracy in photovoltaic power forecasting, and improving forecasting accuracy is a prerequisite for optimal load distribution in microgrids. Therefore, the present invention proposes a microgrid short-term photovoltaic power prediction method based on the chaotic dragonfly algorithm to optimize the BP neural network (CDA‑BP neural network). Based on the influence of solar irradiation change rate and temperature change rate on photovoltaic output power, the connection weight coefficient and threshold of BP neural network model were optimized by using chaotic dragonfly algorithm (CDA), and the optimal photovoltaic power generation prediction model was obtained. CDA uses chaotic search to prevent the algorithm from falling into local optimum, and uses adaptive inertial weighting factors to improve the convergence speed. The invention has good prediction accuracy for the short-term photovoltaic power generation of the micro-grid.

Description

technical field [0001] The invention belongs to the technical field of photovoltaic power generation, and relates to a short-term photovoltaic power generation power prediction method of a microgrid based on CDA-BP. Background technique [0002] Microgrid can promote the access of distributed clean energy, reduce environmental pollution and reduce power transmission loss, and improve the reliability of power supply for users by switching between isolated and grid-connected modes. Photovoltaic power generation power prediction is an important part of the microgrid energy management system, and it is the basis for the optimal scheduling and optimal load distribution of controllable micro-sources such as micro gas turbines, diesel engines, and energy storage. Electricity trading. Relevant studies have shown that higher prediction accuracy of photovoltaic power generation in microgrids can improve the overall economic profitability of microgrids. [0003] The methods of photov...

AI Technical Summary

Problems solved by technology

BP neural network is a multi-layer feed-forward neural network based on error backpropagation. It has strong robustness, can approach nonlinearity infinitely and has strong learning ability, but the convergence speed is slow, which is affected by the weight coefficient and threshold parameter. has a greater influence, and may also converge to a local optimum

Images