基于遗传优化神经网络的电力潜在敏感用户画像聚类算法

Image Clustering Algorithm for Power Potential Sensitive Users Based on Genetic Optimization Neural Network

为了解决电力潜在敏感用户画像聚类和识别结果准确度较低的问题,提出一种基于遗传优化神经网络的电力潜在敏感用户画像聚类算法。构建电力用户画像,精准刻画电力用户行为;选取电力用户画像的数值、时间、统计及聚类四种特征作为卷积神经网络模型的输入,识别电力潜在敏感用户画像;采用改进遗传算法优化卷积神经网络,使得识别结果更为精准。实验结果表明,该方法能够聚类、识别电力潜在敏感用户画像,且聚类和识别的性能及准确度较好。

In order to solve the problem of low accuracy of power potential sensitive user portrait clustering and recognition results,a power potential sensitive user portrait clustering algorithm based on genetic optimization neural network is proposed.Construct a portrait of power users and accurately depict the behavior of power users.The numerical,temporal,statistical and clustering features of power user portraits are selected as the input of convolutional neural network model to identify the potentially sensitive power user portraits.The improved genetic algorithm is used to optimize the convolution neural network to make the recognition result more accurate.The experimental results show that this method can cluster and identify the portraits of potentially sensitive power users,and the performance and accuracy of clustering and recognization are good.