新能源配电网故障修复时间分析与预测研究

风能、太阳能、潮汐能等新能源作为可再生能源，具有节能、环保的优势性，以其为应用发展的新能源电源并网运行，可缓解煤炭、石油等发电的高能耗、高污染问题，促使电网趋向绿色生态发展。为提高电力服务质量，及时告知停电用户故障修复及停电恢复时间，提出了MCNNs模型，将停电原因、电路编号和天气事件等离散数据及连续数据进行二进制编码，代入深度神经网络进行训练，采用正则化和非线性激活优化训练过程，从而提高故障修复及停电恢复时间预测准确率。在仿真阶段，将所提方法与VGG16、ResNet和多层感知器模型进行比较，故障修复时间的预测模型优于停电恢复时间的预测模型，停电恢复时间MAE为118.20 min，比故障修复时间MAE高约90 min。

Research on fault repair time analysis and prediction of new energy distribution network

As renewable energy，new energy sources such as wind energy，solar energy and tidal energy have the advantages of energy conservation and environmental protection.The grid connected operation of new energy power sources developed for their application can alleviate the problems of high energy consumption and high pollution of power generation such as coal and oil，and promote the green ecological development of power grid.In order to improve the quality of power service and timely inform outage users of fault repair and outage recovery time，MCNNs model is proposed.Discrete and continuous data such as outage cause，circuit number and weather events are binary coded and replaced into deep neural network for training.Regularization and nonlinear activation optimization training processes are adopted，so as to improve the accuracy of fault repair and outage recovery time prediction.In the simulation stage，the proposed method is compared with VGG16，ResNet and multilayer perceptron model.The prediction model of fault repair time is better than the prediction model of outage recovery time.The outage recovery time MAE is 118.20 min，which is about 90 min higher than the fault repair time MAE.