Research and Early Warning Method for Power Quality Situation of Active Distribution Network Based on Clustering GA-BP

After large-scale distributed power generation (DG) is connected to the grid, due to the gap and randomness of DG's flexible access output, it is difficult for traditional prediction models to obtain more accurate situation predictions. Therefore, a power quality research and judgment method of active distribution network (ADN) based on density mathematics K-means (DMK-means) clustering and BP neural network is proposed to solve the problem of difficult situation prediction. By improving the DMK-means algorithm to divide the input variables into different sub-categories, reducing the dimensionality of the kernel matrix, and using the BP model for training and performance evaluation for each type of data set, and then using the genetic algorithm (GA) to optimize the BP nerve The network model is used to predict and judge the power quality index items of the active distribution network. The power quality status can reflect the power grid situation from the side, and then build an early warning mechanism to predict the future status of the distribution network. Finally, examples and application analysis verify the practicability and effectiveness of the designed power quality prediction and early warning mechanism.