The Research Of Intelligent Fault Line Detection For Single-phase-to-ground In 10KV Distribution Network

Neutral ungrounded way and neutral through petersen-coil grounded way are widely used in 10 kV power distribution network in China. The two kinds of grounding mode can avoid protection tripping which caused power outages when power grid happen single-phase-ground fault, so as to effectively improve the reliability of power supply. However, the probability of single-phase-ground fault in distribution network is the highest, in order to prevent the failure of expanding, must find out the fault line and take measures as soon as possible. When the small current grounding fault happen, due to the influence of weak fault current signal, unstable arc and random factors, the problem of single-phase-ground fault line detection in small current grounding system of has not been solved effectively. Over the years, a lot of research experts and scholars at home and abroad, put forward a lot of line detection method based on different principles, but a single criterion line detection method generally have low accuracy, limited application scope, vulnerable to transition resistance and fault initial Angle the influence such as the problem.With the development of computing technology and artificial intelligence theory, methods of line detection based on intelligence fusion has become the development trend, especially the artificial intelligence algorithm began gradually to be applied to the fault line detection. With neural network for good parallel processing ability, adaptive learning and associative memory ability, it can effectively make up the lack of fault line detection method based on a single criterion, intelligently fusion the various line detection methods based on the single criterion, so as to detect the fault line. But it also has the long training time and solving easy to fall into local minimum values such as the shortcomings, which cause long time for line detection and low reliability. Ant colony algorithm has strong global optimization ability and distributed computer ability, combining ant colony algorithm with neural network is applied to the fault line detection, which can make up for the shortcomings of neural network, improve the reliability of the line detection method based on neural network.This article first analyzes the steady state and transient process of single-phase-ground fault in small current grounding system, combining with the existing principle of fault line detection method, extracts the steady state characteristics and transient characteristics. And then summarizes the existing model of ant colony algorithm combined with BP neural network and optimizes them, puts forward a fault line detection method that use the neural network optimized by ant colony algorithm to intelligently fusion the method based on energy, fifth harmonic and wavelet packet analysis.Finally, to build 10 kV distribution network model on the MATLAB/Simulink, and make simulation on the small current grounding system when single phase ground fault happen with the condition of different fault circuit, fault location, fault initial Angle and grounding resistance, to collect a large amount of data samples, respectively, as the training sample and testing samples of neural network, and to validate the proposed line detection method. Through the comparison with the line detection method based on BP neural network, prove the validity and reliability of the proposed method.