Research On The Interval Algorithm Of Fault Location On Continuous Power Transmission Lines

Continuous railway power transmission lines supply power for the signal lamps and the loads of station along the lines, etc. So the high reliability of power supply and quick elimination of fault is required. The probability of single-phase-to-ground fault is highest in the neutral non-grounded system for 10kV continuous power transmission lines. Correctly diagnosing and processing the single-phase-to-ground fault is difficult problem which has troubled the railway departments for many years. Just for the problem, the dissertation presents a new fault diagnosing method - interval location method which is based on the multi-segmented and multi-branched structure characteristic of continuous power transmission lines and the requirement of quick processing fault.Firstly, the error factor influencing the precision of conventional location methods is analyzed and their adaptabilities are also compared in the dissertation. In order to process effectively the influence of the branch load along the lines, the correction formulas which convert the branch load to the line parameters are deduced based on the two models of lumped parameter and distributed parameter.The branch load resistance and the error of current which are difficult to be measured accurately are represented quantitatively by interval number. Corresponding three interval location algorithms are constructed based on the line models of lumped parameter and distributed parameter, which are resistance interval algorithm, fault analyzing interval algorithm and distributed parameter interval algorithm. The interval mathematics and corresponding interval searching algorithm are applied to deduce the algorithms.In order to improve the recognising reliability of the faulty segmented interval, the results of every interval algorithm are fusioned by the D-S evidence theory. According to the characteristic of interval number, it presents the constructional methods of the deviation matrix of deviation degree and basic probability assignment function. At the same time, the composition rule of evidence theory is established by the local conflict assignment rule.When fault happen on the boundary interval. It is difficult to build accurately the model of location algorithm. Because of the problem, it presents the method of pattern recognition to recognize fault. The classified model is built by the classified method of support vector machines. In order to further improve the accuracy of classified algorithm, the algorithm of fuzzy support vector machines is studied. Constructional method of fuzzy membership, data reduction strategy and the construction of the classified model are also discussed detailedly.Finally, the hardware and software are designed and the field simulated test was done to verify the algorithms' feasibility and suitability.