Analysis And Forecast Of Open Capacity Of Distribution Transformer Based On Big Data Technology

In recent years,as power industry develops rapidly,the pressure and cost for operating distribution network equipment are increased.The reasonable analysis and accurate calculation and forecast for open capacity of distribution transformer are the basis of safe and economical operation of distribution transformer and steady development of power business expansion and installation,which also optimizes the operation of distribution system and improve the utilization rate of power lines.Therefore,based on big data technology,long short-term memory networ(LSTM)and Bayesian network,this thesis improves the traditional calculation method of open capacity of distribution transformer,and proposes a method of analysis,forecast and calculation of open capacity of distribution transformer based on big data technology.Traditional calculation for open capacity of distribution transformer merely makes use of experience coefficient,which neglects the rule of utility attributes in different regions and is inconsistent with distribution transformer changes.To settle this kind of problem,it is significant to understand the rule of power consumption characteristics.Therefore,this thesis takes the transformer district in Changzhou City of Jiangsu Province as an example to analyze the power consumption characteristics of the transformer district users by using big data technology.According to the different time span of day,week,month and year,the characteristics of power load and simultaneity rate are summarized and analyzed.The traditional calculation of open capacity of distribution transformer is only based on the data of previous years,which has a certain lag,so the calculation error occurs when the load of distribution transformer changes greatly.To solve this problem,it is necessary to predict and analyze the operation of distribution transformer.Therefore,based on the data of transformer district in Changzhou City,Jiangsu Province,this thesis constructs a model of forecasting the annual load peak of distribution transformer district based on Bayesian network and LSTM network(Bayes-LSTM load forecasting model).Firstly,the predictive analysis based on power big data is carried out,and the optimal partition and peak extraction ratio of model samples are obtained.Then,the LSTM network and Bayesian network are combined to construct Bayes-LSTM load forecasting model,and compared with the traditional LSTM sequence forecasting model in forecast accuracy and stability.To settle various problems in the traditional calculation for open capacity of distribution transformer,this thesis improves and proposes a method of measuring open capacity of distribution transformer using big data technology.First of all,after mastering the power consumption characteristics of the users in the transformer district,the load data of the transformer district is collected to build the dynamic simultaneity rate database.Then,the Bayes-LSTM load forecasting model is added to the calculation of open capacity,and its forecast data is used to instead of the historical data in the traditional formula.And the validity and feasibility of the method proposed in this thesis is verified.