Research On Distribution Network Feeder Loop Current Prediction And Risk Assessment Based On CNN-LST

As customers’ requirements for power supply reliability continue to rise,load transfer is generally performed through loop-closing operation to improve certain power supply reliability.However,when there is a significant change in the voltage and phase between the buses,a large loop-closing current may be generated,which may lead to the protection action of the medium voltage feeder and may cause a short-circuit fault and even threaten the personal safety of the operator,seriously affecting the normal operation of the power supply system.Furthermore,loop closing operation preassessment without considering feeder load level,grid-connected renewable energy output,and other boundary conditions may leave distribution network dispatchers uncertain about which days and time periods carry lower risks for medium voltage feeder loop closure.Therefore,in the context of smart grid and digital grid construction,aiming to reduce the average customer outage time and improve power user satisfaction with the goal of "zero planned outages and minimizing fault outages," this study proposes a deep learning network method for predicting the loop closing current of distribution network feeders and assessing the risk of loop closing operation is proposed.This paper begins with a brief review of the research on loop closing operation in distribution network feeders,introduces the development of renewable energy and its impact on loop closing operation when connected to the distribution network,and outlines the current state of research on loop closing operation risk assessment using deep learning.A summary and classification of distribution network feeder loop closing operation methods are presented,with mathematical equivalent modeling analysis performed for a loop closure feeder method in a city in Guizhou as a case study.The closed-loop current calculation formula is derived,and the causes of closed-loop current in distribution networks are analyzed.The principles of convolutional neural networks(CNN)and long short-term memory(LSTM)are introduced,along with DIg SILENT/Power Factory simulation technology.The 10 kV loop feeders of a city’s distribution network in Guizhou is modeled and simulated using DIg SILENT software,and the feasibility of the modeling and simulation technology is verified by comparing the simulation results with the actual test results.Next,a method for medium voltage distribution network closed-loop current prediction is proposed based on CNN-LSTM.Firstly,historical load data,grid structure parameters,and operation modes are obtained using supervisory control and data acquisition(SCADA)and simulation technology,and preprocessing is performed on the data.Then,the preprocessed massive data is structured into continuous feature matrices based on time sliding windows and used as input.Finally,a CNN-LSTM hybrid model is established to map input features to closed-loop current,generating a medium voltage feeder closed-loop current prediction model based on CNN-LSTM to achieve regression prediction.The proposed model demonstrates high accuracy and feasibility,and the conclusions and discussions provide valuable reference for the research and development of smart grid technology.Finally,combining with risk assessment theory,we calculate the closed-loop current overrun severity function,introduce the closed-loop current prediction model proposed in the previous paper into the quantile regression model,and make probability prediction of the feeder closed-loop current,first obtain the closed-loop current prediction value under different quantile conditions,then use the nonparametric kernel density estimation,and use the closed-loop current under different quantile conditions as input to obtain the closed-loop current probability density function at different times.The probability density function is then integrated and calculated to obtain the probability distribution function of the looped current,and then the severity function of the looped current crossing is calculated and combined with the risk assessment theory to evaluate the risk of the medium voltage feeder looped operation.