DEVELOPMENT OF AN ARTIFICIAL INTELLIGENCE-BASED SYSTEM FOR AUTOMATIC CONTROL OF TECHNOLOGICAL PROCESSES

Educational Research in Universal Sciences ISSN: 2181-3515 VOLUME 4 | ISSUE 16 | 2025 https://t.me/Erus_uz Multidisciplinary Scientific Journal December, 2025 194 DOI: https://10.5281/zenodo.18029023 DEVELOPMENT OF AN ARTIFICIAL INTELLIGENCE-BASED SYSTEM FOR AUTOMATIC CONTROL OF TECHNOLOGICAL PROCESSES PhD Ortikov E.E., Sobirov A.S., Keldiyorov S.T., Shokirov F.A. Tashkent State Technical University named after Islam Karimov ABSTRACT In the context of industry digitalization and the transition to the “Industry 4.0” concept, the development of intelligent automatic control systems for technological processes is becoming increasingly relevant. Traditional control methods based on classical regulators and rigidly defined algorithms prove insufficient when working with complex, nonlinear, multiparametric objects operating under conditions of uncertainty and external disturbances. This article examines the development of an automatic control system for technological processes based on artificial intelligence methods, including machine learning, fuzzy logic, and intelligent forecasting models. The architecture of an intelligent control system integrated with the PLC/SCADA level is proposed, methods for forming virtual analyzers and predictive models are described, and the results of modeling and comparative analysis of the effectiveness of the developed system in relation to traditional control algorithms are presented. Keywords: automatic control, artificial intelligence, intelligent systems, technological processes, digital twin, machine learning, predictive control. INTRODUCTION Modern technological processes in industry are characterized by high complexity, multidimensionality, and dynamism. In industries such as the chemical, oil and gas, energy, metallurgical, and food industries, process management is carried out under conditions of incomplete information, the presence of measurement noise, delays, nonlinear dependencies, and variable parameters of the controlled object. Under these conditions, traditional automatic control systems based on classical PID regulators and linear mathematical models often do not provide the required quality of regulation and optimal operating modes [1-3]. The development of computing technology, sensory technologies, the industrial internet of things (IIoT), and artificial intelligence methods has created prerequisites for transitioning from classical automated control systems to intelligent systems capable of adapting, learning, and making decisions in uncertain conditions. Artificial Educational Research in Universal Sciences ISSN: 2181-3515 VOLUME 4 | ISSUE 16 | 2025 https://t.me/Erus_uz Multidisciplinary Scientific Journal December, 2025 195 intelligence (AI) in control systems allows for the implementation of new approaches to data analysis, object behavior forecasting, optimization of control actions, and equipment state diagnostics. The application of intelligent methods in automatic technological process control systems (ATPCS) is of particular relevance, where it is necessary not only to maintain the specified parameters but also to ensure energy efficiency, improve product quality, reduce equipment wear, and prevent emergency situations. The use of neural networks, fuzzy regulators, genetic algorithms, and machine learning methods allows for the formation of adaptive and predictive control algorithms that are superior to traditional ones in a number of indicators. The purpose of this work is to develop and research an artificial intelligence-based automatic control system for technological processes, as well as to evaluate its effectiveness in comparison with classical control methods. To achieve the set goal, the following tasks are addressed in the article: analysis of existing approaches to automatic control of technological processes; development of an intelligent control system architecture; selection and justification of artificial intelligence methods for process analysis and control; construction of mathematical and intelligent models of the control object; modeling and analysis of the obtained results [4]. METHODS The developed automatic control system for technological processes based on artificial intelligence represents a multi-level architecture that combines traditional ATPCS tools and intelligent modules for analysis and decision-making. The main idea is to supplement the classical control loop with an intellectual level capable of adapting to changes in the object's parameters and operating conditions [5,6]. The system includes the following functional levels: − data collection level (sensors, measuring channels, IoT devices); − the level of preliminary data processing (filtration, normalization, elimination of emissions); − intelligent analytical level (AI models, forecasting, virtual analyzers); − the level of decision-making and management (formation of managerial influences); − executive level (PLC, actuators, drives, valves); − level of visualization and monitoring (SCADA, HMI). To formalize the management process, a generalized dynamic model of the technological object is used: x(t)=f(x(t), u(t), d(t)), Educational Research in Universal Sciences ISSN: 2181-3515 VOLUME 4 | ISSUE 16 | 2025 https://t.me/Erus_uz Multidisciplinary Scientific Journal December, 2025 196 y(t)=g(x(t)), (1) where: x(t) – is the process state vector, u(t) – is the control action vector, d(t) – is the perturbation vector, and y(t) – is the measured output parameter vector. In real conditions, the functions f (t) and g (t) are often unknown or have pronounced nonlinearity, which makes it difficult to use classical control methods and requires the application of intelligent models. Artificial neural networks (ANN) are used to approximate nonlinear relationships and predict the behavior of the technological process in the work. The neural network model is trained based on historical data of the technological process and is used to predict output parameters under various control actions [7]. Fuzzy regulators allow for the formalization of technologists and operators knowledge, as well as ensure smooth control under conditions of uncertainty. Intelligent algorithms are implemented at the top level of the system and interact with PLC through standard protocols (OPC UA, Modbus TCP/IP). The SCADA system provides visualization of current parameters, forecast values, and management recommendations. RESULTS To assess the effectiveness of the developed system, simulation modeling of the technological process was carried out using a classical PID regulator and an intelligent AI-based control system. During the modeling process, the following indicators were analyzed: transient process time; over-regulation; resistance to disturbances; accuracy of maintaining the specified parameters. Fig.1. Structural diagram of the intelligent automatic control system for the technological process. Educational Research in Universal Sciences ISSN: 2181-3515 VOLUME 4 | ISSUE 16 | 2025 https://t.me/Erus_uz Multidisciplinary Scientific Journal December, 2025 197 The figure shows the architecture of an intelligent control system, including levels of primary sensors, a programmable logic controller (PLC), an intelligent analytical module, and a SCADA system. Data from the technological process sensors (flow, temperature, pressure, concentration) enters the PLC, from where they are transmitted to the intelligent level. Here, data analysis, forecasting of the future states of the process, and the formation of optimal control actions are carried out, which are then transferred to the executing mechanisms. The presence of a digital twin module allows for conducting virtual experiments without interfering with the real process. Fig.2. Comparison of transition processes in classical and intellectual management. Fig.3. Dynamics of regulation error in various control algorithms. Educational Research in Universal Sciences ISSN: 2181-3515 VOLUME 4 | ISSUE 16 | 2025 https://t.me/Erus_uz Multidisciplinary Scientific Journal December, 2025 198 Fig.4. Comparison of actual and projected values of the technological parameter. The modeling results showed that the intelligent control system provides: a reduction in transient process time by 20-35%; a reduction in over-regulation by 1525%; an increase in system stability during sharp load changes; and a reduction in the standard error of regulation. The use of neural network forecasting models has made it possible to assess changes in technological parameters in advance and form preemptive control actions, which has significantly increased management efficiency. DISCUSSION The obtained results confirm the expediency of applying artificial intelligence methods in automatic control systems for technological processes. Intelligent algorithms demonstrate high adaptability to changing operating conditions and are capable of compensating for the uncertainty of the control object model. The possibility of forming a digital twin of the technological process, which allows for conducting virtual experiments, optimizing operating modes, and reducing risks when implementing new management strategies, is of particular importance. It should be noted that the implementation of intelligent control systems requires solving cybersecurity problems, increasing the reliability of computing modules, and training qualified personnel. CONCLUSION As a result of the conducted research, an artificial intelligence-based automatic control system for technological processes integrated with traditional PLC/SCADA Educational Research in Universal Sciences ISSN: 2181-3515 VOLUME 4 | ISSUE 16 | 2025 https://t.me/Erus_uz Multidisciplinary Scientific Journal December, 2025 199 systems was developed. It has been shown that the use of neural networks, fuzzy logic, and predictive models allows for a significant improvement in the quality of control, system stability, and energy efficiency of technological processes. The obtained results confirm the prospects for further research in the field of intellectualization of ATPCS and the implementation of digital twins in industry. REFERENCES: 1. Zadeh L.A. Fuzzy logic, neural networks, and soft computing. Communications of the ACM, 1994, Vol. 37(3), pp. 77–84. 2. Qin S.J., Badgwell T.A. A survey of industrial model predictive control technology. Control Engineering Practice, 2003, Vol. 11, pp. 733–764. 3. Lee J., Bagheri B., Kao H.A. A cyber-physical systems architecture for Industry 4.0-based manufacturing systems. Manufacturing Letters, 2015, Vol. 3, pp. 18–23. 4. Wang L., Gao R.X., Váncza J. Symbiotic human–machine systems. CIRP Annals, 2019, Vol. 68, pp. 701–726. 5. García C.E., Prett D.M., Morari M. Model predictive control: Theory and practice. Automatica, 1989, Vol. 25, pp. 335–348. 6. Zhang Y., Jiang J. Bibliographical review on reconfigurable fault-tolerant control systems. Annual Reviews in Control, 2008, Vol. 32, pp. 229–252. 7. Venkatasubramanian V. AI in process engineering. Computers & Chemical Engineering, 2019, Vol. 125, pp. 347–356.