УДК 656.2.062.1

Development of a decision support system for railway track control

Замураев Алексей Владимирович – студент факультета Автоматики и информационных технологий Самарского государственного технического университета.

Аннотация: В статье обоснована актуальность разработки системы поддержки принятия решений при контроле железнодорожного пути. Определены основные задачи, которые должна решать система поддержки принятия решений. Приведены основные параметры анализа больших данных. Проведен анализ методов обработки больших объемов данных.

Abstract: The article shows the relevance of developing a decision support system for railway track control. The main tasks that the decision support system should solve are identified. The analysis of methods for processing large amounts of data is carried out. The main parameters of big data analysis are given.

Ключевые слова: железная дорога, система поддержки принятия решений, база знаний, большие данные.

Keywords: railway track, decision support system, knowledge base, big data.

Modern railway has a number of specific features that significantly influence the effectiveness of decision-making.

The basis of traffic safety is the planned and regular maintenance of the railway track, primarily based on monitoring the parameters of the track with inspections, measurements and measures aimed at eliminating identified track disorders and inconsistencies. This is the only way to manage risks. In the absence of a planned system for the current maintenance of the railway track, control over its parameters, in fact, there is nothing to manage. This leads to a significant decrease in the efficiency of the company and the quality of customer service. As a result, the search for optimal solutions for future periods of operation is a rather difficult problem, while to a large extent, the result obtained depends on the completeness and reliability of the necessary initial data provided by various information systems (IS).

The effective solution of the issue of inspection and assessment of the technical condition of the railway track is complicated by the presence of the problems listed below:

• heterogeneity of the utilized information support;
• large volumes of data;
• complexity of simultaneous use of the same data by information systems for decision-making tasks;
• incompleteness and errors in the input data, significantly reducing the reliability of the information used.

These circumstances indicate the relevance of the research. aimed at improving the information support of decision support systems (DSS) for railway track control. The latter was the basis for this study [1].

The development of information support for decision-making systems for railway track control is a complex engineering and economic problem, and a wide range of specialists are currently involved in its solution. In general, this task includes the following areas:

• creation of an integrated information system that ensures the completeness and reliability of stored data;
• development of an automation system for forecasting transportation processes;
• creation of an analytical system that ensures the development of optimal solutions for managing the transportation process.

DSS is an interactive automated system that uses decision-making rules based on models and databases, as well as an interactive computer process of their interaction.

Picture 1 shows the architectural and technological scheme of information and analytical decision support.

Picture 1. Architectural and technological scheme of information and analytical decision support.

Two people with the roles of "Expert" and "Knowledge Engineer" participate in the process of updating knowledge. The first provides knowledge often in an unstructured form, and the second transfers it to the DSS knowledge base in a formalized and fully structured form and in a format that is used in the system itself. After that, the expert verifies the knowledge already in the knowledge base, thereby confirming with his authority that the system can be used to support decision-making, and the recommendations it provides are based on correct inference methods and correct knowledge [2].

In that way, the list of the main functions of the DSS is as follows: [2].

1. Knowledge extraction;
2. Verification of knowledge;
3. Conclusion of recommendations;
4. Explanation of the recommendations.

All this allows you to draw the most generalized functional architecture of the DSS in the following form:

Picture 2. Generalized functional architecture of the DSS.

To be used as input data in modeling contextual probabilistic processes in the DSS, analysis and processing of big data coming from distributed sources is required [1].

Big data analysis can be characterized by the following parameters:

1. Volume - the amount of data generated. This indicator determines whether a certain data set can be considered big data or not.
2. Diversity - the category to which big data belongs. Knowing this affiliation allows analysts to work with information most effectively.
3. Speed - the speed of generating or processing data in order to achieve the set goals.
4. Variability - instability of data over time.
5. Reliability - the quality of the collected data, on which the accuracy of the analysis depends.
6. Complexity - the complexity of the process of correlation and building relationships between data.

In [3], the most well-known methods of collecting and analyzing big data were considered, on the basis of which their advantages and disadvantages were identified, which are presented in table 1:

Table 1. Methods used for analyzing and processing large amounts of data.

References

1. Prudnikov S.I. Big Data processing in systems decision support special appointments. – 2023. – 4 p.
2. Mikheeva T.I., Sidorov T.I., Mikhaiilov D.A. Neuroimaging models of decision support dislocations objects traffic management. – 2015. – 5 p.
3. Lander J. P. Big Data Analytics: Methods and Techniques for Effective Analysis. – 2014. – 307 p.
4. Schmöller F., Christ S. Optimization Models for Rail and Road Planning and Control. – 2019 – 284 p.