G1800105 自己組織化アルゴリズムによる電車線異常の分類・検知に関する基礎研究

抄録

This paper proposes a method for classifying and evaluating conditions on catenary systems by applying the self-organizing algorithm. The proposed method is separated into two steps: (1) development of classification map and (2) evaluation of catenary condition. This paper focus only on the 1st step. The objective to develop the classification map of catenary systems is to make a diagnosis of catenary systems and detect their abnormalities or presages of them. In order to develop a classification map, contact force data obtained under various catenary conditions, which includes abnormal condition, are utilized via several signal processing processes (wavelet transform, cross-correlation coefficient calculation, statistic value calculation and so on). Abnormality detections are done based on the classification map so that detectable abnormality modes depend highly on the stored abnormality modes in the classification map. From that perspective, the map should be developed using contact force data obtained under various catenary conditions includeing many kinds of abnormal conditions one wants to detect in the diagnosis step. The 1st step therefore is done based on the simulation because it is not easy for the actual catenary system to be abnormal conditions to acquire the contact force data. By applying the proposed method in this paper, the conditions on catenary systems are well classified on the self-organizing map.