Abstract: For disposal of the UHVDC blocking faults, this paper proposes a deep-learning-based fault feature modeling method and a post-fault grid dispatching strategy generation method. The proposed intelligent dispatch decision-making is based on the DC fault characteristics and operating environment information of power systems, and the post-fault dispatching strategy is generated through training with the big data driven model. Firstly, based on the fault environment information, the effective fault information is extracted to construct the fault feature model. And then, the principle of deep-learning neural network and the multi-layer perceptron model are introduced, and the idea is proposed to use deep network to extract the running characteristics before and after the training fault and automatically generate the dispatching strategy. Thirdly, the back-propagation algorithm is used to construct the deep learning framework, and the effective fault-disposal strategy is automatically generated by continuously calculating the loss function and the accuracy correction training model. Finally, the effectiveness of the proposed method is verified using the related power system of the Jinsu UHV DC transmission line.

Key words: deep learning, DC fault, data driven, dispatch decision-making, artificial intelligence