Deep Learning Based Automatic Fast Adjustment Method For Cross-section Power Of Bulk Power Grid

The key section power adjustment in power system is an important part of operation mode analysis.At present,the power adjustment of complex large grid section requires a large workload and high repeatability,and the calculation speed is difficult to meet the requirements of online auxiliary decision-making.Therefore,based on the deep learning theory,this paper proposes a method of feature self-learning of sectional power adjustment data to realize rapid automatic adjustment of sectional power in large power grids.Firstly,a section search method based on AP clustering and graph theory cut set is proposed to determine the key transmission sections in large power grids.Considering the number of lines and the weight of branch power flow,Dijkstra algorithm was used to construct the similarity matrix of the system,and AP clustering algorithm was applied to partition the system.Combined with the graph theory cut set,the grid cross-section was quickly searched,and the fault pass rate was taken as the index to determine the key cross-section.The rationality of the proposed method is verified in the IEEE39-node power grid,and the proposed method is applied to the actual large power grid in a certain region,and the key section and the adjustable contact line are determined.Secondly,the power limit calculation method of large power grid based on reduced dimension particle swarm is proposed,and the power adjustment range of key section based on the limit coefficient index is determined,which solves the problem of crossing the limit in section power adjustment.The breadth-first algorithm was used to search the units,and the adjustable units were grouped according to the aggregation coefficient.The optimal adjustment quantity of the units satisfying both the static security and stability constraints of the power grid and the transient stability constraints were calculated by the particle swarm optimization algorithm.The transmission power limit of the transmission section was obtained.Based on the limit coefficient,the adjustment range of transmission power of actual grid transmission section is determined.Thirdly,the relative sensitivity index of the adjustable unit is constructed,the consistency criterion of transmission cross section is designed,and the sample method of power adjustment of large grid cross section based on reverse equal matching method is proposed.The generation plate of the adjustable components of transmission section,the generation plate of power flow sample and the statistical plate of power flow result are designed,and the generation rules of the sample set are formulated.The adjustment process and adjustment results of the massive transmission section are obtained by adjusting the sectional power of the grid data under the initial samples,so as to construct the data set of the power adjustment of the transmission section.The reasonableness and accuracy of the sample set adjustment results are verified in the actual large power grid.Finally,the key unit index is constructed,the optimal regression model based on the feature self-learning of transmission section adjustment data is established,and the automatic adjustment method of transmission power of large grid section based on deep learning is proposed.The proposed method is applied to the actual large power grid system.The calculation results show that,compared with the traditional manual adjustment method,the proposed method avoids the iterative problem in the adjustment process,greatly improves the calculation speed,and the calculation efficiency will not be affected by the system operation mode and the difference between the actual power and the target power of the section.