Study On Fault Area Location And Isolation Of Distribution Network

With the development of social economy, Distribution Automation is the inevitable trend of Distribution System's development. Fault area location and isolation is one of its core function.The paper analyses current main problems of distribution network fault area location and isolation home and abroad. It researches several traditional algorithms with typical significance: matrix algorithm with good generality, improved matrix algorithm with error-tolerated function and high efficient algorithm based on hierarchical model. It validates the validity and embraces everyone's predominance through practical calculation examples.As in the application of practical engineering, most of the information distribution network fault locating used is from outdoor FTU. It will inevitably suffer from interfering by different kind of environment. It's quite possible for fault information distorting or losing. To make location results correct and reliable, the algorithm must be with error-tolerated function. As one kind of artificial intelligent algorithm, Neural Network is typical representation.BP network is the most essential part of Neural Network, which has particular predominance when used in the field of distribution network fault location. Aiming at the shortages of BP network, the paper makes some relevant improvements, overcoming the disadvantages of traditional BP Algorithm such as slow convergence rate, easy to limit to local minimization and so on. It preferably solves the problem of Distribution Network fault area location and has certain error-tolerated function. But it's not high enough to meet project demand. To promote the network's error-tolerated function further, the paper researches a kind of new Neural Network- Neural Logic Network. Three-valued Neural Logic Network is its typical which has great logical reasoning ability. It can correct the error fault information effectively. Combining three-valued Neural Logic Network and improved BP Network, it can realize Distribution Network fault area location with high fault-tolerance performance. By practical calculation examples, they validate the combined algorithm's feasibility and validity, which has important practical worth. At last, it analyses the disadvantages of fault area location using Neural Network and puts forward the in-depth research direction in the future.