#7 2026

CONTENTS №7

Strelkov V.V.  Validation and verification of the specification model of the onboard situational awareness system for the crew during landing

Using the example of an expert review of the design of an onboard situational awareness system for the crew during landing, the paper examines the methods for validating the development process and verifying the specification model of the system in accordance with modern requirements and approaches to such processes. Such a review is a necessary element in the evidence base that is formed by the system developer and presented to the certification authorities.

Keywords: landing, rollout, situational awareness, on-board system, specification model, validation and verification.

Golubev P.A.  Constructing a validation scenario for the application software of the A-plant process control system in a distributed structure

The application software for the upper-level systems (ULS) of A-plant process controls is a complex distributed structure. The specificity of the process controlled by these systems necessitate the placement of the same control objects on different video frames (VF) within a single workstation and on the VF of different workstations. Thus, the development of a validation scenario for the ULS software that allows for these specifics becomes relevant. Against this background, the paper discusses the application of clustering theory for developing scenarios of “parallel” automated validation of the VF from the application software of the A-plant’s control ULS. A comparative analysis of various scenario development algorithms was carried out, taking into account not only time intervals for validating subobjects, but also the need to change the video frames themselves. This enables the reduction of testing time of ULS and process control systems in general.

Keywords: upper-level systems, process control systems, A-plant, validation, application software, databases, video frames.

Samovolik A.N.  Software implementation of formalized step-by-step diagnostics of industrial electrical automation panels

The paper describes the software implementation of the automated troubleshooting system (ATS), which formalizes the step-by-step diagnostics of industrial electrical automation panels. The need to develop such a software tool ensues from the traditional strong dependence of panel troubleshooting on the performer's qualifications, experience and ability to consistently interpret measurement results. The purpose of the ATS, its main functions, the structure of user interaction with the system, the logic of passing diagnostic stages, and the organization of screen forms are outlined. The paper shows how the system guides the user through the diagnostic process, records the results of actions, and generates a diagnostic report and troubleshooting recommendations. The results of comparative testing on a test board are included. They confirm a reduction in diagnostic time and an increase in the completeness of fault detection as against the traditional approach.

Keywords: automated diagnostics, troubleshooting, electrical control panels, software product, step-by-step diagnostics, ATS.

Shikhin V.A., Petriev S.K.  Neural network model for detecting anomalies in the load profile of an industrial enterprise

The paper examines the energy consumption profile, characterized by the presence of possible abnormal readings from the energy metering system of a petrochemical enterprise. The goal is to develop models for automated recognition of critical events resulting in abnormal changes in energy consumption profiles using deep learning of artificial neural networks. The proposed use of the attention mechanism in the architecture of the ANN model makes it possible to surpass the capabilities of popular ANN architectures based on convolutional and recurrent networks in detecting abnormal deviations in the analyzed data. Testing conducted on real energy consumption data of a petrochemical plant demonstrated the high efficiency of the proposed neural models.

Keywords: load profile, neural network, energy consumption, process, automation, critical event, anomaly.

Riham S.  Real-time detection and tracking of laser spots in industrial robotic systems using artificial intelligence technologies

The paper discusses the problem of real-time detection and tracking of laser spots for the semi-autonomous control of industrial robotic manipulator systems. The operator specifies a target point with a laser pointer, and the machine vision system must detect the laser mark, distinguish it from false light objects, and ensure its tracking over time. The complexity of the task is due to the small spot size, changing illumination, surface properties, glare, and optical interference. It is shown that YOLOv8 architecture provides the best balance of accuracy and speed when solving the problem, compared to other neural networks. Applying the proposed approach to solving the problem with a classical algorithm demonstrated that classical methods can be competitive in certain conditions, but require manual parameter tuning. The results obtained are planned for use in the development of laser targeting systems for industrial robotic manipulators.

Keywords: laser spots, small objects, object detection and tracking, optical interference, industrial robotics, real time.

Obaid S.S., Pogonin V.A.  Sensor failure response algorithm in the Smart Garden information and control system

The paper examines an algorithm for responding to sensor failures in the Smart Garden information and control system. A recurrent neural network is used to predict soil moisture levels in the event of sensor failure. The approach ensures timely failure detection and a high level of control, thus increasing system reliability. Experimental results demonstrate high forecasting accuracy and irrigation control efficiency. This improves the sustainability and automation level of climate control in industrial horticulture.

Keywords: algorithm, sensors, recurrent neural network, information and control system, smart garden.

Zhuchkov F.G., Antamoshkin O.A., Kolosova O.V.  An algorithm for controling the quality of changes in the organizational and production process using an AI agent

The paper discusses the problem of managing the quality of changes in an enterprise's organizational and production process. An AI agent acts as a software executor, making changes to digital documents, records, and process data. When solving such problems, technically correct execution of the assigned task by the AI agent may not be sufficient. It is important whether change management rules were followed: was the change authorized, did the agent exceed the assigned task, were the necessary checks carried out, is the agent's history of actions traceable, is the latest approved version of the documents preserved, and does the agent’s report comply with actual changes. To solve this problem, a change quality control algorithm is proposed that detects loss of control events, limits the autonomous continuation of actions by the AI agent, audits changes made after the last approved state, and returns the process to a controlled state.

Keywords: quality management, organizational and production process, AI agent, change management, change audit, replay from event log.

Kruchinin A.E., Gareeva S.N., Anisimov A.V., Balashov S.M.  Reconciliation of material balances under incomplete process flow measurements

The paper reviews the recommendations of MI 3689-2025 “State system for ensuring the uniformity of measurements. Mathematical methods of data reconciliation and gross error detection”. The methods described in the document make it possible not only to calculate the material balance of process flows under incomplete coverage by measuring instruments, but also to identify with sufficient accuracy the values of unmeasured flows, localize losses and leaks, and identify faulty measuring instruments. The use of MI 3689-2025 methodology results in a significant slackening of the requirements for instrumentation of process units, optimization of the calibration schedule, and reduction of downtime for diagnostics. The methodology provides a ready-made tool for daily calculation of material balances in conditions of the shortage of measuring instruments. The weighted least squares method supplemented by MILP-search for gross errors and validation with window functions enables the correct processing of instrument readings with different accuracy classes, the evaluation of unmeasured flows, and the unambiguous classification of the causes of imbalances.

Keywords: mass balance, incomplete measurement, process flows, verification, downtime, diagnostics, measuring instruments.

Vasin D.O., Efimova G.V.  Automated system for recording inconsistent products and calculating damages in the integrated 1C:ERP environment

The paper presents the developed automated system for recording inconsistent products and calculating damages incurred at manufacturing enterprises in the integrated 1C: ERP environment with the 1C: Astrea.RPD claims framework using the example of the enterprises of Transmashholding JSC. The system’s key merit is that it provides a direct calculation of the actual costs incurred for defective products, with the subsequent transfer of information to the supplier in the form of a complaint for the presentation of claims in monetary terms. A role model of the system is presented, a formula for forming the actual cost of a defective unit of production is described, as well as a data transfer mechanism for substantiating a claim to the supplier of purchased components. The criteria for assessing the effectiveness of the system implementation at the enterprise are included.

Keywords: quality management, inconsistent products, defective products, defective product report, labor costs, complaint handling, automation, defective product accounting system.

Berner L.I., Roshchin A.V., Zeldin Yu.M., Ishkov M.M., Kovalyov A.A.  Gas consumption forecasting as a function of the “digital twin” of the gas transportation system

The software module SPURT-R.Prognoz developed by ATGS JSC is presented. The module is aimed at short-term forecasting of gas consumption by settlements. The developed module was tested on real data from two gas transportation enterprises of Gazprom PJSC and was used in conjunction with the Volna non-stationary modeling system for gas transportation systems.

Keywords: gas transportation system, modeling, gas consumption forecasting.

Dozhdev V.S., Daraselia L.Sh., Khramov A.E.  Integration of expert knowledge into hybrid machine learning models for industrial automation using large language models.

The problem of low interpretability of artificial intelligence models in process control systems is discussed. A method is proposed for automatic extraction of diagnostic rules from technical documents using large language models and subsequent integration of the obtained knowledge into hybrid machine learning models. The results of testing the proposed approach in predicting the state of rolling bearings are presented.

Keywords: industrial artificial intelligence, interpretable machine learning, large language models, hybrid models, data quality, automation.