Research On Short-term Prediction Mechanism Of Photovoltaic Power Based On BP Neural Networks

Due to the impact of solar radiation intensity, ambient temperature and other meteorological factors, photovoltaic power is volatile and intermittent. When large-scale photovoltaic power generation system connected to the power grid, the security and stable operation of the power system may be adversely affected. Therefore, short-term prediction of photovoltaic power can help carry out real-time scheduling and guide conventional energy generation planning to mitigate the adverse effects of photovoltaic power generation system, to ensure the power grid safe and stable operation.This paper introduces the background and significance of photovoltaic and the research status of photovoltaic power generation prediction. PV systems are described, the common photovoltaic power prediction short-term solutions are analyzed, the evaluation index to predict is summarized.Conventional prediction schemes only considered the intensity of solar radiation, ambient temperature, wind speed on the effect of the photovoltaic power generation. Single prediction model hardly leads to high prediction accuracy. This paper analyzes individual meteorological factors and weather type on the impact of the photovoltaic power. The analysis found that meteorological factors have a various impact on the photovoltaic power generation. Humidity, visibility and other meteorological factors on photovoltaic power cannot be ignored. Under different weather types, the influence of meteorological factors on the photovoltaic power is not the same. It is difficult to adapt to complex and changing weather conditions using a single model to predict the photovoltaic power generation.For traditional prediction models considering only the intensity of solar radiation, ambient temperature and wind speed on the impact of photovoltaic power generation, using a single prediction model with low accuracy, this paper proposes a short-term photovoltaic power prediction scheme based on BP neural network, using improved BP neural network algorithm, adopting multi-model for photovoltaic power in the short term, taking into account the meteorological factors of the photovoltaic power.Finally, the prediction evaluation of the short-term prediction of the photovoltaic power generation scheme based on BP neural network prediction and carried out. According to the instances of the photovoltaic power generation system, the photovoltaic power short-term prediction models were trained and tested respectively, and the error is analyzed. Prediction results indicate that full meteorological factors consideration to the model have greatly improved the prediction accuracy. Multi-model prediction scheme have been greatly improved in terms of the prediction accuracy, and also addresses high-frequency iterative, long-time convergence problem of the prediction model based on traditional BP neural network algorithm.