Intelligent Analysis Of Abnormal Line Loss In Distribution Network Based On Big Data

With the improvement of the social and economic level,the social power supply has increased significantly,and the line loss problem has become more and more serious.The line loss of the 10 k V distribution network is the main part of the line loss of the power grid.In terms of power grid planning,design,production and management,power grid companies are actively promoting the lean management of line loss to improve operational efficiency while reducing line loss rates.However,in actual production work,the diagnosis and cause identification of line loss abnormalities are mostly relied on a lot of manpower,resulting in low efficiency and poor effect,and it is impossible to analyze and diagnose a large number of line loss abnormalities.Power grid companies still lack the means of cause identification and analysis,and need to be improved in terms of accuracy and timeliness.In addition,there are problems such as incomplete line loss indicators,unreasonable indicator weights and strong subjective preferences in the process of line loss management of power grid companies.It is urgent to find a more comprehensive,scientific and reasonable line loss management optimization method.Firstly,machine learning was combined with the big data of line loss to conduct research on the methods of line loss abnormal diagnosis,abnormal cause analysis,and line loss management optimization.Secondly,in order to improve the quality of data,the data of line loss in actual engineering projects were used for preprocessing to eliminate outliers and make up for missing values.Then,according to the difference of line loss laws in different regions,the Canopy method was used to coarsely cluster the line loss data,and combined with k-means to finely cluster the line loss data,so as to divide the line loss data with the same law into the same cluster data to improve the accuracy of model fitting calculations.In order to improve the convergence speed of the BP neural network and avoid falling into local extreme values,the BP neural network model optimized by genetic algorithm was constructed in this thesis.After the line loss data was passed to the model for training and learning,the model was used to predict the line loss rate.When the difference between the predicted value and the actual line loss value was too large,the line loss was abnormal at this time.For the abnormal points of line loss,the relevant data indicators were deeply excavated,and the specific cause for the abnormal line loss were analyzed based on the principal component regression method.According to the different causes of abnormal line loss,take corresponding line loss control measures.In this thesis,a line loss management indicator system was constructed,the subjective weight and objective weight were combined with the idea of game theory.Combination weights combined with approximation to ideal solution ranking method were used to study the line loss management of power grid companies.Among them,the G1 method was used to determine subjective weights.The method did not require consistency checks,and was concise and efficient.The CRITIC method was used to determine the objective weight,which not only considered the contrast between indicators,but also took into account the conflict of each indicator.To sum up,a complete system from abnormal identification,cause analysis to management and optimization of line loss was proposed.According to the rule of line loss data,the power grid line loss was optimized,which provided reference for power grid companies to save energy and reduce loss.Furthermore,it would help power companies better adapt to the rapid development of distribution network under the new situation,to provide a reference for power grid companies to save energy and reduce losses.