which is transmitted during numerous interviews of a knowledge engineer and an expert in the subject area. The stage of knowledge acquisition is one of the main bottlenecks in the technology of creating expert systems due to the low rate of filling the system’s knowledge base. It should be added to this that there are subject areas for which it is often difficult to find an experienced expert person, and sometimes there simply does not exist one. In addition, it has long been noticed that not all experts are ready and able to share their knowledge [2,8.10].
An important quality of technical systems that allows them to be classified as intelligent is the presence of such properties as:
■ learnability – the ability to generate new knowledge and data (models, decision rules) based on inductive inference mechanisms, generalization of statistical data, etc.;
■ classification ability – the ability of the system to independently differentiate control objects, environmental influences, control signals, automatically structure data;
■ adaptation – the ability of the system to adapt to the changing conditions of the operating environment, correctly take into account the non-stationarity of control data, etc
One of the promising approaches to the creation of intelligent systems may be to attract the ideas of situational management as a system – wide approach based on formal methods of theoretical artificial intelligence – logical-linguistic models, models of learning technical systems in the construction of management procedures for current situations, deductive systems for building multistep solutions, etc. In this important area of research, as well as in the development of general methodology, theoretical foundations and specific applications, priority undoubtedly belongs to Russian scientists.
The problem of industrial implementation of intelligent information systems capable of processing data with their inherent a priori uncertainty in railway transport is becoming more and more urgent. In many cases, the data is not only inaccurate and uncertain, but also incomplete, and sometimes unreliable. The development of methods that allow obtaining reliable conclusions based on such data is another direction for fundamental research.
4 AUTOMATED DISPATCH CENTERS AS INTEGRATED INTELLIGENT TRANSPORTATION PROCESS MANAGEMENT SYSTEMS
Currently, the development of various automated control systems in railway transport is increasingly taking place in the direction of their intellectualization. As a rule, intelligent railway systems are created to control individual processes.
World experience shows that the greatest effect is achieved when developing and implementing an integrated interconnected complex of intelligent systems. In this case, a unified information support is created, the mutual influence of managed processes is taken into account.
General integration principles
An illustrative example of the need to create an integrated complex of intelligent systems is the existing network (TCC) and regional (RTCC) automated dispatch control centers. There are dozens of automated workstations (AWSs) in various areas of organization of the transportation process, maintenance and repair of infrastructure and rolling stock devices, as well as security. Each AWS as a human-machine system performs a specific target function. However, a full-fledged interconnection of these functions can be carried out only with the integrated construction of a complex of intelligent dispatch systems. In principle, we can talk about a unified intelligent system in automated dispatch control centers. Let’s consider this provision in relation to regional (road) control centers – RTCC.
In each RTCC, a hierarchical dispatching structure solves tasks of three main types:
1) ensuring loading in accordance with the daily and current loading plans;
2) ensuring the passage of trains (including those performing local work) in accordance with the traffic schedule, the formation plan and the plan for the transfer of wagons along internal and external joints with unconditional compliance with traffic safety;
3) performing various kinds of special transportation and tasks.
There are obvious direct relationships in the work of various dispatchers when implementing these types of tasks. Close relationships also occur when solving tasks of various types, so delays in the passage of trains (task type 2) may entail non-fulfillment of tasks for tasks of types 1 and 3. Untimely completion of a special task (task type 3, for example, the promotion of a train with oversized cargo) may cause disruption of the transfer of trains and wagons at the joints (task type 3) and loading plan (task type 1), etc.
Therefore, synchronous integrated intellectualization of the AWSs of the entire control unit of the RTCC is advisable.
The main provisions are defined, the implementation of which is a necessary condition for the intellectualization of management processes in regional dispatch centers. These include:
■ the use of principles for the development of automated process control systems (TP ACS);
■ ensuring efficiency in solving various types of tasks and resolving emerging conflict situations;
taking into account market conditions in the work of control centers;
■ saving all kinds of resources.
When building management processes, it is necessary that the developed algorithms for solving specific tasks (for RTCC dispatchers these are operational tasks) make it possible to obtain rational, and if possible, optimal solutions. For this condition, it is necessary to have a sufficient amount of information about the processes, take into account the influence of various factors, including disturbing influences, as well as constantly monitor the situation on the basis of special feedback subsystems.
It is these requirements that are taken into account when building TP ACS as closed control systems with feedback.
Each dispatcher constantly accumulates experience, which is used when making decisions. Therefore, when developing intelligent systems, it is important to use the principle of their self-learning.
At the present stage of development of intelligent RTCC systems, the control solutions developed should be used in the «adviser» mode. With the accumulation of experience in the operation of such systems, the refinement of the complex of factors and algorithms taken into account, the transition to the automatic mode of their operation will be carried out.
The dispatcher’s work proceeds in the constant adoption of operational decisions. The degree of efficiency depends on the needs and capabilities of forecasting specific situations.
The need for an operational forecast can extend over a very long period. Let’s imagine the situation in a RTCC, the scope of which includes a large seaport, and the cargo comes from loading stations located at distances of several thousand kilometers. Linking the approach of wagons with the approach of ships, especially taking into account weather conditions, requires a forecast of the operational situation for 10—15 days ahead.
A multi-day forecast is also required to solve the problem of organizing the turnover of locomotives and locomotive crews. At the same time, a forecast for 20—30 minutes may be sufficient for the train dispatcher to solve a specific conflict situation of train traffic on the section.
Therefore, for each task performed in the RTCC, the developer of an intelligent management system determines
