Dovile Rama

A fault detection method for railway point systems

Failures of railway point systems (RPSs) often lead to service delays or hazardous situations. A condition monitoring system can be used by railway infrastructure operators to detect the early signs of the deteriorated condition of RPSs and thereby prevent failures. This paper presents a methodology for early detection of the changes in the measurement of the current drawn by the motor of the point operating equipment (POE) of an RPS, which can be used to warn about a possible failure in the system. The proposed methodology uses the one-class support vector machine classification method with the similarity measure of edit distance with real penalties. The technique has been developed taking into account specific features of the data of infield RPSs and therefore is able to detect the changes in the measurements of the current of the POE with greater accuracy compared with the commonly used threshold-based technique. The data from infield RPSs, which relate to incipient failures of RPSs, were used after the deficiencies in the data labelling were removed using expert knowledge. In addition, possible improvements in the proposed methodology were identified in order for it to be used as an automatic online condition monitoring system.

A Petri-Net-based modelling approach to railway bridge asset management

Management of a large portfolio of infrastructure assets is a complex and demanding task for transport agencies. Although extensive research has been conducted on probabilistic models for asset management, in particular bridges, focus has been almost exclusively on deterioration modelling. The model being presented in this study tries to reunite a disjointed system by combining deterioration, inspection and maintenance models. A Petri-Net modelling approach is employed and the resulting model consists of a number of different modules each with its own source of data, calibration methodology and functionality. The modules interconnect providing a robust framework. The interaction between the modules can be used to provide meaningful outputs useful to railway bridge portfolio managers.

A reliability analysis of railway switches

Switches and crossings (S&C) are critical elements on railway networks. Any failure of S&C usually leads to train delays and cancelations having a negative impact on the quality of service delivered, railway safety and also operating costs. S&C is a multi-component system and to enable proactive prevention of S&C unit faults, with their undesirable consequences, the ability to predict problems at component level is needed. This paper describes the derivation of lifetime distributions of individual S&C components based on field data collected. Maintenance was identified as a potential contributor towards increased frequency of reported S&C faults. The outcomes of the analysis and lifetime distributions for the elements of the switch units provide a means of predicting the expected number of maintenance activities, and associated costs for these units over any specified period of time. They provide an essential input to the optimisation of maintenance in developing an asset management strategy.

Accelerating Petri-Net simulations using NVIDIA Graphics Processing Units

Stochastic Petri-Nets (PNs) are combined with General-Purpose Graphics Processing Unit (GPGPUs) to develop a fast and low cost framework for PN modelling. GPGPUs are composed of many smaller, parallel compute units which has made them ideally suited to highly parallelised computing tasks. Monte Carlo (MC) simulation is used to evaluate the probabilistic performance of the system. The high computational cost of this approach is mitigated through parallelisation. The efficiency of different approaches to parallelisation of the problem is evaluated. The developed framework is then used on a PN model example which supports decision-making in the field of infrastructure asset management. The model incorporates deterioration, inspection and maintenance into a complete decision-support tool. The results obtained show that this method allows the combination of complex PN modelling with rapid computation in a desktop computer.

Railway bridge asset management using a Petri-Net modelling approach

Infrastructure assets can be difficult to manage due to the array of defects, the variety of environmental situations and the different operational scenarios. A number of studies have tried to model bridge asset management. The main focus of these models has been on the deterioration profiling as capturing this can be complex. The model presented tries to model railway bridge detrioration as well as the inspection and intervention processes to give a more rounded overview of railway bridge asset management. A Petri-Net (PN) modelling approach is used accompanied by historical data, used to calibrate the deterioration of the model. Industry policies are used to govern the inspection and intervention procedures. Various aspects of the model have been adjusted or enhanced by industry experts. The model is simulated to provide essential outputs for railway bridge portfolio mangers.