Degradation state mining and identification for railway point machines

Abstract

Point machines (PMs) are used for switching and locking railway turnouts, and are considered one of the most critical elements of a railway signal system. The failure of the point mechanism directly affects the operation of the railway and may cause serious safety accidents. Hence, there is a need for early detection of the anomalies in PMs. From normal operation to complete failure, the machine usually undergoes a series of degradation states. If the degradation states are detected in time, maintenance can be organized in advance to prevent the malfunction. This paper presents a degradation detection method that can effectively mine and identify the degradation state of the PM. First, power data is processed to obtain the feature set that can describe the PM characteristics effectively. Then, a clustering analysis of the feature set is carried out by self-organizing feature-mapping network, and various degradation states are mined. Finally, the optimized support vector machine is used to build the state classifier to identify the degradation state of the PM. The experimental results obtained with the Siemens S700K PM show that the proposed method could not only mine the effective degradation states, but also obtain high identification accuracy. 2The performance of a PM directly affects the safety of train operation. In the field investigation of the Chenzhou West Station, the related faults caused by the PM were found to account for 30% of all signal equipment failures, and the average time for repairing the PM is approximately 23.4 minutes, which has a serious impact on the operation of the high-speed railway. Therefore, it is necessary to detect the faults in the PM in time and make reasonable maintenance plans to improve the operational efficiency of the high-speed railway. A condition monitoring system (CMS) can measure, centralize, and analyze the data from sensors installed in the field equipment to detect equipment failure [1]. Many railway companies have implemented CMS for PMs. However, experience in this field has shown that these systems create an alarm only when a failure occurs or very close to the point of failure. In addition, most of those currently in use do not effectively detect faults, and the system simply creates an alert to indicate abnormal behavior, leading to a delay in repairing of the PM [2]. The fault detection technology of CMS needs to be improved in order to increase the maintenance efficiency of PMs. In recent years, research on fault detection for PMs has gradually transitioned from the threshold-based method to statistical analysis, model-based, and classification methods [3]. Compared to thresholding technology, these three methods can ensure early detection of machine failure. Among them, the classification method is widely used in CMS [4].The classification method can be divided into three main steps, viz. data feature processing, classifier establishment, and fault identification. The data feature processing step converts the measurement signal of the PM, e.g. power signal,...

Railway Axle Condition Monitoring Technique Based on Wavelet Packet Transform Features and Support Vector Machines

“…Once the model is trained, it is cross-validated, and then it can be used to predict the class of new data. SVMs are among the most widely used classification systems due to their high classification accuracy and good generalization performance, even with few samples [ 22 , 33 , 34 ].…”

Anomaly Detection for Controller Area Network in Braking Control System With Dynamic Ensemble Selection

“…The mean shift clustering algorithm is used to distinguish undefined abnormal patterns for anomaly detection in wireless sensor networks (WSNs) [21]. Some anomaly detection [22], [23] or fault detection [24] algorithms based on self-organizing maps (SOM) have been introduced.…”

Railway Point-Operating Machine Fault Detection Using Unlabeled Signaling Sensor Data

“…It also adopts a one-size-fits-all approach, and is not adaptable to the individual behavioral characteristics of mechanical devices [3]. Statistical approaches have fared better, [2,[6][7][8][9][10] and provide the added advantage of being able to function in real time, without the need for historically labelled data of "good" and "bad" examples. Garcia et al, used the Kalman filter approach and compared new data to a reference point to identify faults.…”