Fault Location And Identification Method Of Power Grid Based On Multi-source Information

With the development of power grid technology and artificial intelligence,China's power industry is moving towards a higher level of intelligence.At the same time,the requirement of high-quality economic development for power grid stability is increasing,and power grid faults will bring great losses to society and users.Therefore,the correct diagnosis of power grid faults is very important,including the location and diagnosis of fault elements.Fault type and cause identification are two main aspects.When complex faults occur,various factors may lead to malfunction or rejection of circuit breakers and other components.At the same time,in the process of information upload,because of interference,it is very difficult for operators to diagnose faults quickly in the face of a large number of power grid status and alarm information.Therefore,it is very difficult to improve the level of information mining and to determine faulty components and fault types quickly and accurately.And the reason is the key to further improve the level of operational decision-making.Accurate location of fault elements after fault occurs can help operators to quickly adopt corresponding operation strategies to ensure reliable and stable operation of the overall power grid;at the same time,identification of fault types and causes can help maintenance personnel to recognize the severity of the fault in advance,and take corresponding maintenance strategies and accuracy for different fault types and causes in advance.Preparatory work can greatly improve the efficiency of maintenance,shorten the maintenance time and improve the reliability of power grid operation.Because of the uncertainty and incompleteness of power grid information,accurate fault element location and identification requires more fault-tolerant methods.Artificial neural network has the ability to deal with complex non-linear problems and good fault-tolerant characteristics.Based on the methods of back propagation neural network and reverse sequential causal network,this paper proposes a multi-source fault diagnosis method for power grid.The method includes fault location and fault identification.Fault identification also includes fault type identification and fault cause identification.It realizes comprehensive fault diagnosis from multiple directions.Firstly,the overall framework and method of fault location and identification are described in detail,including multi-source information,fault element location and fault identification.The main flow chart of fault diagnosis is given.The method of determining the number of nodes in the optimal hidden layer of BP neural network and the method of determining the training function of comprehensive overall training error and iteration number are proposed.Finally,an example is given to illustrate the modification in detail.Secondly,a two-stage BP neural network fault location method is proposed based on the improved BP neural network modeling method.In the first stage,parameters such as the number of nodes in the hidden layer of the network are determined according to the improved modeling method,and the first diagnosis is made and the preliminary location results are output.In the second stage,the samples with inaccurate initial location are calculated by reverse reasoning logic combined with reverse sequential causal network.The additional matrix of the sample is calculated,the second diagnosis is carried out and the results are output,and the application steps are elaborated with an example.The results show that the two-stage BP neural network diagnosis method proposed in this paper not only improves the location accuracy of the fault components,but also makes the evolution of the fault mechanism transparent.Finally,the overall frame of fault type and cause identification is established.The frame identification model is established,and the characteristic analysis and data preprocessing methods of fault types and causes are discussed in detail.Seven electrical characteristic quantities and 38 fault cause identification characteristic quantities with multi-source information for fault type identification with current mutation rate are proposed.The model is transformed into the input vector which can be processed by BP network through pretreatment,and then the BP neural network identification model is explained in detail.Finally,an example is given to illustrate the accuracy and validity of the identification method.This paper presents a fault location and identification method for power grid based on multi-source information.The method of using neural network not only improves the accuracy of fault element location under double complex faults,but also has good fault tolerance.It can accurately identify the fault types and causes,which is helpful for operation and maintenance personnel to quickly determine fault elements and formulate maintenance strategies.It has important application value to improve the efficiency of fault treatment.