Deep neural network and genetic algorithm synergistic optimization of new energy generation power combination prediction technology research

Due to the influence of atmospheric chaos, new energy power generation has inherent strong stochastic volatility, and prediction errors are inevitable. New morphological changes such as source-load interaction and complex coupling of the power system increase the complexity of the prediction problem, and there are multiple uncertainties in the prediction, which bring great risks and challenges to the safe and economic operation of the new energy power system. This paper explores the advantages and disadvantages of the combined prediction model, outlines the practical basis of the model combination, and lists specific programs for the implementation of new energy power prediction. The entire deep confidence network model is pre-trained by stacking Boltzmann machines. The K-means classifier is set at the end of the model to extract data features to complete the cluster analysis. The Adam algorithm is used to complete the optimization of the deep confidence network. At the same time, the initial weight matrix obtained from GA is used to design the GA-DBN model and establish a new energy power generation prediction model based on GA-DBN. The model is used to analyze the power generation of new energy farms under different weather conditions, and the power generation under sunny weather conditions changes synchronously with the change of solar radiation intensity, and the power generation is higher at 10:00-15:00, and the change curve is relatively smooth, and the power generation is in the range of 300-375kW. Compared with the real value of power generation, the overall maximum error does not exceed ±0.2, and the accuracy is higher, and the whole is closer to the trend of the real curve, so it can be seen that the model proposed in this paper is real and effective in the prediction of new energy power generation.