In order to reduce maintenance labor of overhead contact lines (OCL), a contactless measurement device for OCL was developed. This device is mounted on a roof of a vehicle of a train and measures static three-dimensional positions of wires of OCL and detects positions of OCL fittings without touching the OCL while the train is running. We proposed hybrid sensing method that combines stereo measurement by image processing with structure measurement by laser range scanners and it realized to measure OCL geometry with high-precision even in sections with complicated OCL structure. In addition, we developed position detection method of the OCL fittings that can cope with changes in height and stagger of OCL by using machine learning. Measurement data of OCL contactless measurement device is static position of OCL without influence of a probe such as a pantograph. With this device, the OCL static position can be measured continuously instead of at each support point or dropper point. For maintenance of OCL, the criterion of OCL is defined as a static position. The device is utilizable for OCL maintenance and it sophisticate maintenance of OCL. For example, this device realize the height difference measurement of the crossing section, which has been conventionally measured by maintenance workers. In addition, the device are utilizable for OCL fittings inspection. Maintenance workers can check the image of OCL fittings without on foot into the field. Furthermore the static position data of OCL can be used to create simulation model of OCL dynamic behavior. Using this model, it is possible to know the dynamic response of cases where various pantographs pass at various speeds. Running tests was conducted on commercial line, and the performance of the device was verified when running at a speed of 130 km/h. The results shown that the repeated measurement accuracy is within 10 mm, and the OCL fitting detection rate is 90 % or more.
Various unsupervised anomaly detection methods using deep learning have recently been proposed, and the accuracy of the anomaly detection technique for local anomalies has been improved. However, no anomaly detection dataset includes co-occurrence-related anomalies, which are combinationrelated. Thus, the accuracy of anomaly detection for co-occurrence-related anomalies has not progressed. Therefore, we propose SA-PatchCore, which introduces self-attention to the state-of-the-art local anomaly detection model, PatchCore. It detects anomalies in co-occurrence relationships and anomalies in local areas with the benefit of the self-attention module, which can consider contexts between separated words introduced first in the natural language processing field. As no anomaly detection dataset includes anomalies in co-occurrence relation, we prepared a new dataset called the Co-occurrence Anomaly Detection Screw Dataset (CAD-SD). Furthermore, we performed experiments on anomaly detection using the new dataset. SA-PatchCore achieves high anomaly detection performance compared with PatchCore in the CAD-SD. Moreover, our proposed model shows almost the same anomaly detection performance as PatchCore in an MVTec Anomaly Detection dataset, which is composed of anomalies in a local area. As a contribution to the anomaly detection task, we have released the CAD-SD to the public. This dataset can be downloaded from the following link: https://github.com/IshidaKengo/Co-occurrence-Anomaly-Detection-Screw-Dataset INDEX TERMS Anomaly detection, deep learning, self-attention
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.