Intelligent Information Process Method And Its Application In Fault Analysis For High-Voltage Transmission Lines

With the development of power system, safe operation and stability of power supply are given more and more attention. As pillar of power delivery system, high-volt transmission-line plays a so important role that its fault will not only threaten the safety of modern power system but cause huge economic loss. What's more, with the utility of different kinds of supervisory equipments in power system, when a fault in transmission-line occurs there will be so many alarm information entering control canter in a quite short period that it becomes very hard for operators to identify fault merely by their experience. In these circumstances, it is absolutely necessary to develop an intelligent fault diagnosis and analysis system for high-volt transmission-line because for one thing, this system can realize quick diagnosis for transmission line fault to minimize loss of power blackout, for another intelligent system also acts as a very useful assistant of operators when dealing with complicated fault. On the basis of generalization and comparison of current methods for transmission line fault diagnosis, this paper presents new way to identify, locate and diagnose high-volt transmission lines fault. And sufficient simulations have proven of the effectiveness and correctness of these new methods. Firstly, this paper utilizes a new algorithm-Support Vector Machines (SVM) to conduct Fault Type Identification. Compared with current method, which is mainly based on setting threshold value for specific fault type, SVM is more suitable to deal with nonlinear identification and transmission-line fault identification happens to be this kind of problem. The formulations deriving from SVM not only can realize quick and accurate Fault Type Identification but also need much less training time. Secondly, after analyzing existing methods for Fault Location and their weaknesses, this paper presents a new method using two-terminal electrical parameters to locate transmission-line fault. This method has more complete theory and does not need data available from two terminals seriously simultaneously. Tests have proven that this method can work perfectly no matter what fault type, location and transitional resistance are. Thirdly, this paper uses Radial Basis Function (RBF) neural network to diagnose high-volt transmission-line fault. RBF has a simple structure and can get trained easily and quickly. In order to determine center of basis function of RBF neural network, HCM algorithm is utilized. Compared with other algorithm, such as LBG algorithm, HCM avoids finding cluster center in every iterative process and therefore get a satisfying clustering speed and effectiveness. Additionally, this diagnosis system based on RBF neural network is more capable to process distorted information so that is more suitable for on-line fault diagnosis.