#10 2024

CONTENTS №10

Dozortsev V.M., Anosov A.A., Baulin E.S., Gorodnova M.V., Korostelev A.Ya., Slastenov I.V.  Artificial intelligence in process control applications

Artificial intelligence (AI) is a pervasive tendency of the 21st century. The interest to it is enormous as well as its influence on all aspects of our life. It concerns also the industry, including fuel and energy complex and chemistry. AI has a rich half-century history in chemical engineering. This article provides its concise overview, examines the features of the subject area in the context of AI approaches, and discusses the outlook for developing effective AI solutions in various areas of chemical process automation with the focus on the solutions from the Center of Digital Technologies (subsidiary of Rubytech Group).

Keywords: automation of chemical process control, hybrid modeling, top-down and bottom-up paradigms of artificial intelligence, first-principles modeling, artificial neural networks.

Zadiran K.S., Shcherbakov M.V.  Prediction of rear events in equipment’s instantaneous value assessment

Achievements in mechanical engineering enable the development of more effective equipment, but its complexity and ownership cost increase. There are a variety of maintenance planning strategies, among which the risk-oriented one based on failure prediction looks the most promising. A failure is actually a rare event in modern equipment lifecycle. Rare events hamper the application of standard solution patterns in prediction tasks, including the ones based on the recent neural network architectures. One more feature of the task is the large number of measurand and their measuring frequency. The existing techniques typically use the data aggregated per work cycles that causes no high dimension problem. However, popular prediction methods based on instantaneous values, whose amount is larger by orders of magnitude, are not used. The paper offers a residual life prediction method, which generalizes data aggregation in work cycles as a preprocessing stage with subsequent identification of degradation process for sample splitting. The method also comprises a new approach to learning sample formation in the conditions of rare events with the selection of optimal input parameter set for improving prediction accuracy. The application of the method proposed enabled the accuracy improvement by 12…17%.

Keywords: rare event, residual life of equipment, machine learning, prediction, turbine engines.

Kostenko A.V.  Unification of pneumatic valve types

The selection of pneumatic valves with specified discharge characteristics and speed is the most difficult task in the design of pneumatic systems. The variety of valves increases the complexity level of a pneumatic system and implies the need of them in stock, which however remains often unclaimed. To surmount this challenge, a versatile modular distributor with electric control was developed. Under parallel-serial connection, it enables all distributor types. The solution is delivered by a novel, nonstandard monostable (with a “pneumatic spring”) 3*/2 selector valve type, which presumes the availability of one intake and two offtake channels. The proposed distributor pattern is a versatile system covering the needs of pneumatic automation. The available circuitry enables flexible selection of the optimal automation level without extra modification and improvement costs. The application of a single valve type reduces lifecycle costs from design through modernization and enables the change of the valve’s standard functionality at any time.

Keywords: pneumatic valve, selector valve, pneumatic automation, modular valve, compressed air.

Chereshko A.A., Bakhtadze N.N., Kushnarev V.N., Elpashev D.V.  A method for proactive control of complex multi-connected plants based on digital associative models

Model-based predictive control technology is being extensively used in various processing industries. The paper offers an alternative method called “associative MPC”. The underlying associative search algorithm is based on the simultaneous calculation of predictive model’s coefficients and control impacts at every time step. The concept of structural identification matrix for multi-dimensional control objects is formulated.

Keywords: identification, associative search, model-based predictive control.

Dozortsev V.M., Maksimenkov V.N., Vlasov A.K.  Normative-descriptive technique for automatic evaluation of process operator actions

The paper offers and examines the hybrid normative-descriptive evaluation of procedure implementation by process operators. The normative part is necessary for determining the quality of artificially generated implementation variants. The obtained variant samples are visualized in the space with the dimensionality reduced by principal component analysis and separated subject to quality by means of enclosing tunnels. The membership of a path to a certain tunnel forms a descriptive estimate of operator’s actions either in the training session or at the real-life process plant. A pilot experiment was undertaken with a simulation model of a dual feed heater, which has confirmed the high reliability of the hybrid estimate.

Keywords: automatic evaluation of operator’s actions, normative and descriptive evaluation, normative and descriptive estimate of operator’s actions, procedure implementation path, principal component analysis, enclosing tunnels.

Mostakov N.A., Vinokurov A.O., Filchenkov A.A., Zakharova A.A.  A Robot system for automated painting of ship surfaces

Commercial solutions for automated preparation and painting of ship surfaces are overviewed. A robot system developed in the Institute of Control Sciences of the Russian Academy is presented; its structure, functionality, key modules, and hardware implementation are described. The results of field experiments proving the solution’s serviceability and efficiency are included.

Keywords: automated ship painting, robot system, robot manipulator, mobile platform, machine vision, 3D surface model.

Bredikhin A.V., Danilov A.D., Verkoshansky D.E., Sokolnikov V.V.  Intelligent system for real-time scheduling of multipart manufacturing

The paper discusses the development of the architecture for an intelligent system for production resources management based on the multi-agent approach. Key system components are described, their functional modules and implementation technologies are determined. Special attention is paid to the system interaction not only with software agents, but also with the equipment and production staff. Prototypes of HMI for the interaction with software agents during resource and order scheduling are presented. Algorithms implementing the process of equipment load rescheduling are offered. Implementation examples of real-time response to events allowing for the participation of software agents are included.

Keywords: scheduling, multi-agent system, software agent, pilot production.

Prokopiev A.P., Nabizhanov Zh.I., Bolshakov A.A.  Hard-/software complex of an intelligent system for monitoring the density of road concrete mix for vibrating rollers

The paper describes an innovative hard-/software solution for monitoring the density road concrete mix for vibrating rollers. It enables the calculation of standard intelligent compaction indices, such as CMV, CCV, Evib, kb, etc., based on the amplitude of vibratory drum acceleration and material rigidity parameters. Field test results are presented.

Keywords: hard-/software, monitoring of the density of road concrete mix, vibrating rollers, amplitude of vibratory drum acceleration.

Khamitov R.N., Kazymov I.M., Kompaneets B.S.  An application outlook for power profile of electricity users developed using integral, interval, and current meter readings

The paper discusses possible application of the power profile of electricity consumers for the automation of operation mode control in power supply systems. It presents the comparative analysis of various user power profiles based on meter readings obtained at different frequencies. The factors affecting the accuracy and information content of the data obtained. A case study demonstrates the possibility of simultaneous analysis of various capacity profiles for obtaining the most complete information about the processes in the power supply system under consideration.

Keywords: power profile, electricity user, analysis of electricity consumption parameters, power supply system.

Kugaevsky S.S., Starostin A.P.  Development of specialized structural-technological element in CAM environment

Feature Based Machining (FBM) is a most promising development line for CAM systems. This approach drastically speeds up and simplifies the development of CNC technology owing to well-developed processing algorithms for standard structural-technological elements (STE). However, it may be necessary to simplify the modeling of the machining of custom STE, the application of which is typical for a specific production site. The paper examines the process of automatic recognition of a custom STE based on its 3D model in NX CAM software environment with the example of “a nest for replaceable multifaceted plate” STE. Recognition results create a framework for developing a technological algorithm for manufacturing this STE. The algorithm allows reducing the duration of control program calculation for a CNC machine and optimizing the overall production process.

Keywords: CAM systems, structural-technological elements, standard element method, parts machining.

Voronov I.A.  Project management system at a coal enterprise

The application of Project Lad digital model for construction project management in Siberian Coal Energy Company is discussed. The paper describes project prerequisites, implementation stages and results. It shows that the application of Project Lad has improved forecasting accuracy and reduced the risks associated with project timeline and budget.

Keywords: project management system, artificial intelligence, automation, industrial enterprise, modeling

Glukhovsky V.A., Chaglakov E.L., Sysoev A.N., Malakhov A.I.  Application of digital radiography systems for flaw detection in welded ring joints of pipelines

The paper compares digital radiography against other radiographic inspection methods such as gamma radiography and imaging plate radiography. It presents digital radiography application experience for nondestructive testing of welded joints during pipeline inspection and maintenance at the gas production sites of Gazprom Dobycha Orenburg, LLC. The advantages of digital radiography systems are highlighted, application challenges and the ways to overcome them are outlined.

Keywords: nondestructive testing, digital radiography, welded joints, pipeline.