The dissertation is devoted to solving the current scientific problem of developing the theoretical foundations of locomotive control systems by using methods of the theory of artificial intelligence, which made it possible to increase the level of traffic safety during the operation of traction rolling stock due to the improvement of the train control process.
The impact of the human factor on the quality of train management has been established. Of the total number of traffic accidents, up to 80% are related to the influence of the human factor. This gives grounds for asserting the need for further work on eliminating the harmful influence of the human factor on traffic safety and improving locomotive control systems.
The scientific novelty of the dissertation consists in solving the scientific problem of developing the theoretical foundations of locomotive control systems by using the methods of the theory of artificial intelligence.
For the first time:
- an additive criterion for evaluating control actions when driving a train was developed in the form of a ratio of formalized performance indicators of the "train-driver" system, such as traffic safety, consumption of energy resources for traction and execution of the traffic schedule;
- the "driver's work intensity" parameter in the process of driving a locomotive is formalized, which, unlike existing approaches to the driver's work assessment, takes into account the types of current train situations and their mutual influence on the work of the locomotive crew.
Refined:
- the method of determining the value of the weighting factors for train situations by using the Saati method for different modes of train movement. The use of such an approach made it possible to conduct an objective assessment of train situations and classify them according to traffic modes as dangerous, requiring additional attention, those requiring immediate action, and safe.
The practical significance of the obtained work results is that, on the basis of the obtained results of theoretical studies, improved algorithms of the decision support system for locomotive drivers are proposed. New algorithms were developed and expansion of simulation modes was proposed.
In the first chapter, as a result of the study of the influence of the human factor on the quality of train management, it was established that the main causes of transport events in the locomotive industry are the human factor. Thus, it can be argued that reducing the influence of the human factor on traffic safety is a significant reserve for its improvement. Currently, not enough attention is paid to the intellectualization of the decision-making processes of the locomotive control system. At the same time, with the development of more and more modern software complexes, attempts to build systems are allowed, the nature of functioning of which is more and more close to the intellectual activity of a person.
In the second chapter, the main causes of the locomotive driver's mistakes are determined, namely, unsatisfactory training or low level of qualification; non-compliance with the prescribed operational work procedures; unsatisfactory working conditions associated with such negative phenomena as excessive noise, vibration, temperature fluctuations in the driver's cabin; lack of attractive stimulating factors to achieve the optimal level of work quality. A criterion of the quality of the work of the "engine driver-locomotive" energy system during operation is proposed, which is presented in the form of a ratio of various quality indicators that reflect various properties of the system. Based on the main tasks performed by the locomotive industry, the following management strategies are defined in the work: compliance with the traffic schedule, maximum traffic safety, minimum power consumption for traction, maximum level of rolling stock reliability.
In the third section, Euler-Venn diagrams are used to improve the quality of management activity analysis and visual representation of relationships between subsets of the universal set "train situation". Based on the Saati method, an approach to determining the weighting factors of train situations has been developed. To determine the circle of the most informative signs of train situations, the method of random search with adaptation was used.
In the fourth chapter, an assessment of the prospects of decision-making support systems for locomotive crews was carried out. Improvement of adaptive control algorithms leads to their significant complication and difficulty of implementation directly on board the locomotive. The structure of costs for the implementation of decision support systems for drivers consists of costs for the purchase of an on-board computer, an interface part, a number of sensors, software development and installation of the system on the locomotive.
