Research On Operation Optimization Of Intelligent Equipment In Active Distribution Network Based On Deep Learning

The intensifying electric power automation and the integration of distributed energy into power grid drive significant complexity in distribution network.Faced with various problems produced by load fluctuations,active distribution network emerges in order to support the safe and stable operation of the power grid and the introduction and lean control of smart devices is one of the main development directions,This paper presents deep learning algorithms to analyze the intelligent operation and lean control for smart devices emphasizing on-load capacity transformers and further achieve the optimal operation of intelligent equipment.The capacity regulation criterion and strategy determines the working efficiency of the on-load capacity transformer.However,the prevailing ramp-up timer control method has great drawbacks,including limitations of the adjustment boundary,omissions of optimal moment and improper operation following the adjustment.In this paper,the operating load data is decomposed and fused into sequence current load data,and then the capacity adjustment criterion of on-load capacity regulation transformer established based on sequence current load can solve the problems of unity and simplification of traditional capacity regulation boundary.This criterion not only retains the tolerance of the capacity switching of the on-load capacity regulation transformer,but also reflects the rationality of capacity switching,effectively reduces the number of capacity switching,reduces switching losses,and ensures the operational safety of the on-load capacity regulation transformer.In terms of capacity adjustment strategy formulation,long-short term memory network with attention mechanism is introduced to predict the sequence current load of the on-load capacity transformer.Then,the occurrences and durations of the on-load capacity are optimized reasonably with the previous prediction results,which benefit equipment intelligent control.Due to the shortcomings of long-short term memory network in dealing with long-term dependence problems,the lagging forecasting exerts a negative impact on the lean management of the equipment.Faced with the above problem,this paper make full use of unique memory units in Neural Turing Machine to further ensure the operating rationality of on-load capacity transformers.On the basis of ensuring the optimization of the operation mode,the lean control is further realized.Based on the real data in "The replacement of coal with electricity" project,the superiority of the proposed scheme is verified in view of economic analysis.The aim in energy saving and loss reduction is achieved,and the lean control of on-load capacity transformer is enhanced notably in complex cases.