Research On Identification,Location And Evaluation Of Voltage Sag Source

With the development of industrial equipment,electrical automation and intelligent level,voltage sag has more and more significant influence on the production and operation of industrial and commercial users.Especially,the industry which uses power and electronic equipment in large quantities such as semiconductor manufacturing,precision instrument processing and automobile manufacturing is very sensitive to voltage fluctuation.When the effective voltage value is less than 90% for 1-2 weeks When the wave is above,it will trip and stop,resulting in economic loss or accident.Voltage sag is a common power quality problem.Motor start,transformer switching,short circuit fault and so on will cause voltage sag.The production interruption and delay caused by voltage sag interference are rising obviously.The direct and indirect economic loss is becoming more and more serious,which puts forward higher requirements for power supply quality.Therefore,it is a hot research content to analyze the mechanism of voltage sag event and to deal with the voltage sag problem to achieve the high quality power required by industrial production and power users.It is the hot research content at present: accurately identify the source of voltage sag,analyze,compensate and suppress the local voltage sag situation,timely and accurately locate the location of the fault sag source,which can be used as the power supply department and the user The basis of coordination disputes;accurate evaluation of voltage sag of each node in power grid is an important basis for taking measures to manage the sag,and is an essential step in the problem of voltage sag control.Therefore,this paper studies the identification,location and evaluation methods of voltage sag source.Firstly,in the aspect of voltage sag source identification,two kinds of voltage sag source identification methods are proposed: the voltage sag identification method based on deep belief network and the voltage sag source identification method based on convolution neural network.The voltage sag identification method based on deep belief network uses the feature extraction ability of DBN to extract the features from the measured waveform data,which solves the problem that the manual feature extraction relies too much on expert experience and is greatly affected by unknown features.The model integrates feature extractor and classifier,optimizes the structure of the model,selects the optimal parameters of the model,and improves the efficiency of sag source identification.A voltage sag source identification method based on convolution neural network is proposed.A self supervised CNN voltage sag source identification model is constructed based on convolution neural network and automatic encoder.The convolution layer and pooling layer of CNN are used to extract voltage sag features.The self supervision of network training process is realized based on AE principle.The self-monitoring training process does not need to input a large number of training sets and correct labels in advance,which overcomes the problem that the traditional method can not identify the unknown sag waveform correctly.The high accuracy and superiority of the two methods in the identification of sag source are verified by the measured data.Then in the aspect of voltage sag location,because most of the voltage sags are caused by short-circuit faults,and the severity and economic losses caused by short-circuit sags are more severe,this paper studies the source location method of short-circuit voltage sags.By using the full waveform information and feature matching technology,a fault branch detection method based on mutual approximation entropy is proposed.Based on the accurate identification of the fault line,the simplified R-L Model is used for accurate fault location.In the example,a 33 node radial distribution network is built by using Simulink,and the positioning error is tested to verify the effectiveness of the proposed method.The test results of fault branch detection method based on cross approximate entropy in 33 bus network simulation model can reach 100% accuracy,which provides a new idea for voltage sag location.The overall accuracy of the two terminal positioning method is 98.51%.Finally,in the aspect of voltage sag severity evaluation,this paper improves the evaluation index.The new index covers the information of system side and load side.The traditional evaluation method only considers the shortcomings of power companies or users,and the results are more comprehensive and in line with the reality.The subjective weight is constructed based on AHP method,and the objective weight is constructed based on critical method.Combining the two as the combined weight,the 33 node network is verified by Monte Carlo simulation fault event in Simulink system.The sag condition monitored by each node is scored,and the nodes are sorted according to the severity of voltage sag.The ranking results show that this method is in line with the theoretical results,more practical,and can reasonably screen out the vulnerable nodes sensitive to voltage sag.