Researches On Simulation And Identification Of Lightning Strikes On Transmission Lines

Due to the increasing voltage levels and transmission capacity which lead to the extension of line corridor and tower height, the probability of high voltage transmission lines being struck by lightning is growing. What’s more, lightning strikes on transmission lines would adversely destroy the electrical equipment and it is the basic premise of dealing with system failures and strengthening lightning protection measures to correctly analysis and identify lightning faults. Meanwhile, the identification of lightning strikes and short-circuit faults on transmission lines could guarantee the seeking direction of faults in the field. At the same time, it is also significant to quickly achieve the fault location and remove the fault interference.Firstly, based on the comparion of common modeling methods, this paper builds more reasonable simulation models. Due to the lack of uniform standards of lightning strikes on transmission lines and little comparative research on diverse methods, it divides the transmission line system into five parts: lightning strikes, overhead lines, towers, insulators and other electrical components and respectively compares and analyses the normal modeling methods for each part. In the meantime it builds a more realistic simulation model of the 220 kV transmission system and simulates the ordinary single-phase ground, two-phase fault,lightning interference and shielding and back striking lightning faults. The current data of traveling wave is obtained whice is the foundation of the following characteristic analysis.Secondly, the paper excavates the characteristics of transient current components from different faults in the time and frequency domain and proposes a lot of characteristics to achieve the identification. Most of existing recognition methods analysis the signal from the time domain or frequency domain dispersedly, the whole description is not enough, it is short of the integrated analysis of the lightning current from the time-frequency domain. While, it puts forwards four feature vectors against the waveform change, amplitude distribution and complexity of transient current in the time domain. It also defines two kinds of identification criteria in the frequency domain based on the different proportions and various distributions of high-frequency and low-frequency components in traveling waves. What’s more, it confirms two distinction standards upon the integration of characteristics in the time-frequency domain. And it respectively achieves the identification of normal short-circuit and lightning faults, lightning interference and faults, shielding and back striking lightning faults. At last, it verifies the effectiveness of criterias under different conditions and factors through extensive simulation.At last, relying on the study of the relevance vector machine, this paper presents and verifies the lightning recognition method based on the relevance vector machine. Every identification criterion has some limitations, what’s more, it may lead to the scatter of the characteristic values to achieve the identification with numerous criterions one time. In order to overcome the shortage above, this paper introduces the relevance vector machine algorithm.After the classification of the process and the assignment of vectors, the paper builds the recognition model and achieves the intelligent identification of short-circuit fault, lightning interference and lightning faults. What’s more, a large number of simulation tests confirm this method proposed in this paper can achieve better recognition results and higher accuracy.