Applications of machine learning and data mining in damage provision and maintenance of operation for Greek railway rolling stock

Abstract

Recently, important advancements in the international financial development and the introduction of new technologies in the business sector have arisen new challenges in the management of data and information, through computer science, having accumulated huge number of various kinds of data. Thus, it is commonly accepted, that any kind of organization or modern business it is currently highly dependent on technology. The use of new technologies provides new businesses with an important competitive advantage, whereas the lack of new technologies could be fatal for a business and even lead to its closure. Every business making use of advanced technologies, collection of stored and processed and explicit data is capable to predict malfunctions and avoid failures of the equipment used. The use of data from the production and the equipping of businesses can improve the effectiveness of the production and the arrangement of the maintenance. The adoption of explicit data and digital technologies in the maintenance or the function of the equipment can increase the production and decrease the cost of maintenance. The future and the success of each business are defined by its capacity to evaluate and make use of its ultimate source, which is business knowledge. In every business the production of knowledge from the process of efficient information is identified as strategic qualification and source of competitive advantage. In the current changing environment, the businesses must improve and innovate in fast pace throughout the whole spectrum of their activities. One of the most important responsibilities of the executives of a business is the sound decision-making. It is of great importance the development of information, processed by the administration of the business, from selected data with the use of appropriate tools in order to make fast, valid and precise decisions. The transportation companies are very critical for the economy and the prosperity of every country since they significantly contribute to all the services offered that derive from the state. The services offered to the citizen must constantly be improved and upgraded with primary aim the safe transportation of passengers and products. In the Railway sector a huge balk of data is produced that needs to be evaluated, exploited and used as a mechanism which will lead to the most sensible decision-making, aiming at saving resources and maintaining the fundamental principle of Railways which is the safety of passengers and products. The safety is the comparative advantage of Railways that the administration of every business must enforce, making the optimum decisions. The transport companies are based on the effective and ongoing development of data along with the systems which provide them that is information systems. Railway companies use applications of machine learning and mining data for the optimum development of data and mining useful conclusions, in order for the business to collect and use the best information for the programming of the strategic movements in order to accomplish its major aim the safety of passengers and trains. The maintenance of the Rolling stock is a sector of importance of transport companies whose purpose is the achievement of safe and reliable function of the railway system. It is internationally the most expensive sector of the railways, at the same time though it is the most vulnerable one, since its negligent performance leads to consequences from reducing credibility to causing fatal destruction as huge material damage even casualties. In the present dissertation is presented the technology of machine learning and the mining data as well as their contribution to transport. Innovative applications are mentioned that are used in railways networks and specifically in railway companies. In the referring dissertation the problems of the Greek Railways are arisen and the modern requirements for the maintenance and the restoration of malfunctions of the Rolling stock are recorded. An innovative method of improving the management of the circulation of rails is developed with the aid of the rapidly improved sector of machine learning and mining data aiming at the prediction, diagnosis and dealing with malfunctions that immobilize a train on the main passenger’s rail. A holistic approach and utilization of stored – inactive data of the Greek Railways is carried out in order to make strategically based decisions and the establishment of a new plan of rails maintenance aiming at the realization of predictive maintenance, getting away from the defined kilometres of the preventive maintenance. The method suggested can be applied as an innovative mechanism for the extraction of specialized information, without the addition of new recording machines – monitoring the trains. At the same time, it is suggested the addition of a new application creating a complete system of business intelligence into the present system of information with modular architecture concerning the administrations of Rolling Stock, in collaboration with the administrations of Information of the Greek Railways and all the procedures are described that are demanded for the planning and its implementation. In the meanwhile, the benefits of business intelligence are presented with the collaboration of the techniques of machine learning and mining data in the Greek Railway companies that use obsolete procedures of maintenance. To sum up, they are mentioned the conclusions, the motives and the future perspectives that the new technologies offer for the creation of innovative applications on railway companies and can be used as a tool for the improvement of the procedures of making decisions from the executives in order to achieve the goal of major importance of the railways which is the safety of passengers and trains.