Research On Fault Location Of Distribution Network Based On Multi-Feature Fusion And Optimized Wavelet Neural Network

There are two kinds of neutral point grounding way in our country’s power system, the neutral point directly grounding way and the neutral point non-effective grounding way.Especially in 6kV-10 kV distribution network,the neural point is generally not directly grounded. When the single-phase grounding fault occurs,the short circuit current is small, therefore calls the system for small current grounding system. The probability of single-phase earth fault is very high in distribution network, it can be accounted for about 80% of the total failure. When the single-phase grounding fault occurs, the three phase line voltage of the system is still balance, and the fault current is small, do not affect the load continuous power supply, the system can continue to run 1 to 2 hours.But long time running can easily make the failure expand for grounding short circuit between two or more points,even cause overvoltage in the system, and make damage to the equipment, affect the safe operation of the system. So it is necessary to find out the fault line and fault point and then the fault can be removed as soon as possible.In this paper, on the basis of fault line selection in small current, a method of fault location in distribution network based on optimized wavelet neural network was proposed. The IMF energy, fifth harmonic in fault point and zero-sequence active power are marked as the feature of fault location,combined with the nonlinear fitting ability of WNN to determine the fault point. Wavelet neural network has the disadvantage of slow convergence rate and is easy to fall into local optimum,which is improved by the use of genetic algorithm,and the partical fitnessfunction and weight calculation is worked out.The simulation results show that the convergence speed and positioning accuracy of the improved algorithm is significantly better than the wavelet neural network. Meanwhile, laboratory tests also verified the feasibility of the algorithm.In this paper, which aims at 6kV distribution networksystem on the power plant of Dexing copper mine, the original fault line selection device has been improved. Linux kernel of diagnostic machine has been cut, the root file system has been re-created, QT libraries and SQLITE database has been transplanted and also QT interface has been designed to display information about the status of the fault line.To realize the fault information of remote monitoring, this paper uses the socket network programming to realize the data transmission between diagnostic machine and the PSX600. Firstly, the fault information was transmitted to PSX600 via Ethernet based on IEC103 agreement, secondly,PSX600 would transmit the fault information to a remote control room via Ethernet based on IEC104 agreement, achieving the goal of remote monitoring.