Traveling Wave Fault And Interference Recording Data Identification Based On Information Aggregation Technology

Traveling wave data contains a wealth of information, if can effectively mining the characteristics of the traveling wave data, will enhance the traveling wave ranging nature of discriminant accuracy, fault type and fault identification ability, for the subsequent wave record data in the fault location and construction of the main research work lay the foundation. However perennial of the cumulative amount of travelling wave wave record data file, it has a large amount of data and the value of low density, difficult to extract the useful information, but the characteristics of high application value. Now used a lot of these wave record data files in addition to the historical record to occupy a large amount of disk space, no form of traveling wave data feature extraction of effective, the lack of recorded wave data description of characters, and the traveling wave ranging device adopts mutation startup mode, start the threshold low, lead to a lot of clutter is recorded, the traveling wave interference wave record and unbalanced ratio of fault wave record, how to go from a lot of traveling wave data of the fault identification and interference wave record data is particularly important.Therefore, in this paper, the use of information aggregation technology line was proposed and recorded wave data information aggregation scheme, according to information gathered in each layer is put forward, different functions and algorithm realization of large amounts of data from the feature extraction, information fusion, data dimension reduction, character portrayal to decision to guide the whole process. When facing different wave record data, based on the characteristics of wave record data to select different feature extraction methods, according to the information aggregation scheme, finally realizes the decision-making guidance.Based on information aggregation technology, traveling wave fault and disturbance record wave data to identify, for example, in data level aggregation layer uses the fault and disturbance recording data signal at a frequency different wavelet scales consisting of different changes occurred, resulting in the distribution also changed the distribution of wavelet entropy law has different variations of feature extraction data; in the aggregation layer made using wavelet energy entropy, mean and variance under three measures of information fusion; polymeric layer to the feature level proposed by the main component analysis dimension reduction methods; the decision-making level in the polymeric layer made using conventional and mainstream computing Mahalanobis distance Gaussian mixture model clustering algorithm proposed criterion for effective identification of faults and disturbances traveling wave record wave data, and proved its feasibility aggregation in general recorded wave file analysis.