CC BY
69INTELLIGENT CONTROL SYSTEMS
SOHRAB CAVADOV
Azerbaijan State Oil And Industry Universiti
Abstract
Control system intellectualization issues are noticed. There are valid reasons for the need to intellectualize a variety of systems and control techniques. Different artificial intelligence techniques are compared and examined as the hierarchy of intellectual control levels is observed.
Keywords: intellectualization; intellectual control levels; fuzzy logic.
Introduction
Beginning in the 1950s, the discipline of artificial intelligence (AI) research and development arose and progressed concurrently with the theory of automatic control, with the first significant applications in computers and information science and later in automatic control! The first commercial and industrial applications of artificial intelligence date back to the 1980s2. AI has progressed to a certain degree of maturity and stability during this time.
An critical issue which can lead to a rethinking of modern day achievements and make new u.s.of the principle and exercise of AI is the
pointy boom in possibilities of pc technology, together with hardware implementation of logical and different means of AI.
The term "intelligent operating system" refers to a combination of hardware and software that can operate autonomously or in a man-machine fashion and synthesize control objectives and find rational ways to achieve control objectives. (If you have the motivation and knowledge, including information about the environment and internal circumstances)1,3. Today, the ability to synthesize control objectives is achieved through human-machine interaction, and autonomous control systems that can only find rational ways to achieve control objectives are called "intelligent control systems".
Currently, there is great interest in management science and practice in combining classical automatic control methods with AI methods and AI applications in the field of managing complex and poorly formalized objects and processes. Especially when information, system health, audit criteria and audit objectives change over time, they are unclear and sometimes contradictory.
This report provides a hierarchical structure of levels of intelligent control and a comparative analysis of different forms of AI. Since the volume of theoretical and applied research in the field of fuzzy controllers has grown rapidly in the last decade, the main objective of the report is to discuss the main findings in this field. Unfortunately, this field alone is not sufficient to generate the full review the author desires.
1. General problems of control systems intellectualization
Successful solutions to the task of ensuring the technological independence of countries in the development and use of complex technological objects for civil and military purposes largely depend on the effectiveness of the developed management systems and technologies. Adequate theoretical and management skills are required, taking into account the potential lack of certain required resources (depending on the application) such as information, time, energy, finance, materials and manpower.
Accidents and known accidents in the industries of transport, industry, energy, etc. it is often due to so-called "human factors", including worker overtime. FF is often caused by quality issues in operating system design, especially for manageable emergencies. In the modern Russian situation, human errors and exhaustion of technical resources of facilities and control systems are common. Assured reliability and control quality, including upgrades to design, operation and retrofit management functions, are urgently needed.
One needs methods and technologies for evaluation of control systems and to ensure their
ОФ "Международный научно-исследовательский центр "Endless Light in Science"
optimality, functional and operational reliability, efficiency, fault tolerance and survivability are necessary under the following conditions:
• Lack of a priori information about the external environment of the control object and its functions (including under contradictory circumstances).
• Numerous non-stationary factors and their subjective nature that are difficult to take into account.
• Deterioration (due to failure, accident) or the need for targeted reconfiguration (remanufacture or development management).
As the function store expands, the control system becomes very complex. Among the many contributing factors to the complexity of modern advanced control systems are:
• Multi-level control, heterogeneous description of subsystems by quantitative and qualitative models, different scales of processes in space and time, multi-modality, multi-link, decentralization and branching, and modern control systems and their control objects General Structural Complexity of Systems,
• Coordinate Parametric, Structural, Regular and Idiosyncratic Effects, Including Active Countermeasures in Uncontrolled Existential Conflict Environments,
• Uncertainties in Information About System State Vectors and Parameters use of deterministic and probabilistic models to describe information uncertainties regarding volatility, properties of measurements and environmental errors,
• Nonlinearities, dispersion parameters, delays in control or object dynamics, and impulsive impacts, models such as higher dimensions of models and others.
Figure 1 shows the large-scale structure of control science and technology.
Fig. 1 Large-scale structure of control science and technology.
Adaptive, robust, predictive, and other control methods developed in control theory aim to account for imperfections in dynamics by acquiring missing information during the training phase or in real time. is. The use of AI reduces complexity by covering tasks that use unknown or quantitative models that are no longer valid from a certain moment of functioning, and tasks for which quantitative models are less efficient than the use of AI models. It means expanding the capabilities of the control
ОФ "Международный научно-исследовательский центр "Endless Light in Science"
system (similar to the action planning task). Or it can be used in combination with an AI model1.
Various artificial intelligence tools (neural, evolutionary, logic, etc.) can be used for action planning tasks and general control. Each of these classes has its strengths and weaknesses, and they warrant implementation of a higher level of heterogeneous control over complex systems, especially for real-time requirements (Figure 2).
The intensive development of technological systems and processes (networking, sensors, miniaturization of control devices and computers, increasing their performance, etc.) imposes new demands on modern control systems, increasing the demand for embedded control systems at both levels. opens up new possibilities. Distributed multi-agent systems at various scales and levels of group interaction. Current is the research and development of the transition from robots operating in hazardous environments but with an operator interface (surveillance of UAVs) to intelligent robots. This requires inexpensive robots based on modular principles of construction and miniaturization that solve the problems of sensitization, environment modeling, robot control team goal attainment, and application range expansion. Rapid changes in standards also require robots with high-precision navigation and intelligent control in the fields of agriculture and road construction.
An example of very important technical processes and intelligent control objects is the large-scale infrastructure systems of the power industry. in this case:
• Inefficient structure of electrical networks and generating capacity,
• Lack of energy savings in electricity consumption,
• Technical and commercial losses in electrical networks,
• Technical backlog and high wear and tear on equipment,
Energy Market Dominance,
• Power System Vulnerability to Terrorist and Cyber Threats
It is necessary to develop models of complex infrastructure dynamic systems such as and create efficient and reliable intelligent control systems for smart grids)4-6.
Fig. 2. complex system with heterogeneous management
Controls based on logic-reactive (production) knowledge models in so-called expert, recommendation, or decision support systems that need to be extended to include new functionality:
• Coordination with other means of intelligentization of control systems (artificial neural networks, genetic algorithms) and adaptive, robust and predictive control algorithms,
• Logic control by combining symbolic and multimedia methods Representation and Knowledge Processing to Reduce the Complexity of Interfacing Systems with the External Physical World,
• Handling Partially Formalized Natural Language Texts,
• Abstract and Inductive Updating of Knowledge,
• Problems Integration of quantitative and qualitative models with ontologies from various subject areas that characterize the situation.
Table 1 lists the strengths and weaknesses of AI tools.
There are several ways to combine different AI tools. For example, neural-reactive and logical-reactive (productive) AI means can be integrated with primary logical methods of intelligent control. While the latter method can handle a wider layer of knowledge, the first two measures are "reasonable" by providing the simplest heuristic response of the control system to changes in the environment or controlled object. support the action. At the logical reaction level (sometimes with a large number of "if-then" rules), knowledge presentation testing is especially necessary. For boolean production rules with compositional semantics, knowledge-based validation can be reduced to dynamic analysis of automaton networks. This analysis is monotonically further simplified with the class of automata. State 8 by applying the method of transfer of the mathematical model of properties.
An essential problem of AI is the problem of automatic evaluation of meaninglessness of knowledge. Not only the lack of information, but also the oversupply of information leads to deterioration of the intelligent control system.
Recent advances in the field of intelligent control include automating the search for ways to achieve externally mandated control goals, but automating target setting and validating control quality criteria are still lacking. It is also recognized that improving the "machine components" of mature man-machine systems alone is not sufficient to significantly improve their utilization efficiency. This goal in creating human-centric systems can be achieved through the efforts of engineers and scientists to improve the intelligent component of the human-centric system's "system kernel." 8,9 as "onboard intelligence".
Table 1. Comparison of intelligent control measures
А1 шеаш Typical advantages Typical disadvantage-»
1 Neuro lietvvoiked 1 Applicable in muliivaiiable pioblems with 1. Necessity of Uainiug information. i.e.
(nruro rrarthr) poorly formalized regularities representative set of input-output examples < "rather
2. High degree ot panllizabihty nod eye. I ban the brain" 1.
performance 2. Slowness ol learning.
3. Capacity' to learn
Ц. EvotuUuuoJ High degree of parallizability am! perfoiuiance 1. A ptiori uncertainty of efficiency ш applications.
(вгоиЮ 2. "Rather sell-organization in nature, than the
creative process"
til Logical- reactive 1 Saturnine» of rule» ("if-then"). 1 Complexity of execution of a large set of
Iprmlurlioa npf) 2. Possibility of representation of declarative production rule; Poor structuring of knowledge
and procedural knowledge base
2. Complexity of providing the correctness of
knowledge processing
< Incompleteness of the languages w.r t the first-
ordet description.
IV. Object-oiiemed I Good structuring of know ledge presentation. I. Complexity of ptogtainnimg «avoiding AI ideals)
1 trames I 2- High performance of mechanisms of 2. Insufficient expressiveness
inheritance of properties etc
V Logical 1. High expressiveness 1 Insufficient performance, traditional are the off-
2 Corrector» line applications
A High complexity of off-line tasks 2 L'osolvability of rich logics
3. Insufficiency of a single logic
VI Object- Imergratinn of advantages of the object-oriented 1. Lack] of logical models
logical and logical models 2. Complexity of programming.
vn Multi-agent Accounting for reflection and self-ognnmtioa Correctness requires hither investigation
First and foremost, in combat situations typical of aviation, especially fighter aircraft: the most aggressive external environments and tight crew time limits, in-flight intelligence is required. Inflight intelligence is a functionally integrated complex intended to carry out all aircraft tasks9. Scientific and technological advances in this area will also benefit other applications of AI in multi-criteria, uncertain, and risky situations, where operators are overloaded with information, and time is scarce. You can improve the quality of control in situations where you are busy or stress is prevalent. The development of practical expert systems for in-vehicle decision making, including systems based on fuzzy logic and similar situational reasoning, has reached the practical stage of building models and prototypes. They have advanced rapidly to create 4++ and 5th generation manned combat aircraft and combat drones. Their shards appeared on upgraded fighters of generation 4++. The overseas development is primarily planned for use in the new US F-22 and F-35 fighter jets, upgraded F-16, F-15, F/A-18 aircraft and helicopters that provide multiple intelligent functions on board. . Tactical decision-making system9. Research has shown that improvements to on-board computers, cockpit displays and controls, and other avionics systems will enable the next generation of aircraft/helicopter designers to design and implement new types of in-flight information systems. Such systems can support tactical decision-making (targeting current flights in time and choosing rational methods to reach them). In previous generations of aircraft, solving these tasks was done only by the efforts of the crew.
We also discuss in detail some of the questions about the intelligentization of automatic control systems.
Combine with fuzzy control and other AI assets. The first controllers developed in Greece in the 3rd century BC. they can be considered linguistically described fuzzy controllers with logical operations. Today, in Japan, China, USA, Germany, France, UK, Russia, we can see many practical applications of purge control systems in industry, transportation, energy, oil and gas, metallurgy, medicine and other fields and appliances domestic . and other countries. other countries. We consider four main types of controllers: fuzzy logic language, analysis, learning, and proportional integral differentiation (PID) controllersl, 7, 11-17. Since information about them is unstructured and scattered in many publications, our analysis will help guide experts in this field.
REFERENCES
i. Russell SJ, Norvig P. Artificial Intelligence: A Modern Approach. Prentice-Hall, Inc., 1995.
2. Vassilyev SN, Yadykin IB, Morgin YuI., Voropai NI, Efimov DN. Intelligent operating and emergency control technologies in Russia CD Proceedings of the 2th IEEE PES International conference and Exhibition on «Innovative Smart Grid Technologies - Europe» (ISGT 2011 Europe, Manchester). Manchester: University of Manchester, 2011. p. 1-7.
3. Vassilyev SN, Voropai NI, Rabinovich MA, Yadykin IB. Unified energy system of Russia: intellectual solutions. Energy of development. 2012; 2: 48-57.
4. Fedunov BE. On-board operatively advising expert systems of tactical level for manned aircraft. International aerospace magazine "Aviapanorama" 2016; 1: 9-20.
5. Aliev RA, Abdikeev NM, Shakhnazarov MM, Production system with artificial intelligence. M.: Radio and communication; 1990. 264 p.
6. Kuzmin VB, Travkin SI. Theory of fuzzy sets in control problems and principles of the device of fuzzy processors: a Review of foreign literature. Automatics and telemechanics. 1992; 11: 3-36.
7. Zakharov VN, Ulyanov SV. Fuzzy models of intelligent industrial controllers and control systems. II. Evolution and principles of construction, Izv. Russian Academy of Sciences. Technical Cybernetics. 1993; 4: 189-205.
8. Zakharov VN, Ulyanov SV. Fuzzy models of intelligent industrial controllers and control systems. IV. Simulation modeling, Izv. Russian Academy of Sciences. Technical Cybernetics. 1994; 5: 169-211.
9. Vassilyev SN, Zherlov AK, Fedosov EA, Fedunov BE. Intelligent control of dynamical systems. M.: Fizmatlit; 2000. 352 p.
10.Pupkov KA. Some results of development of scientific-technical program "Intellectual systems" of FKP "Universities of Russia" //
Proceedings of the 2 International. Symp. "Intellectual systems" (INTELS'96). 1996; 1:5-9.
11. Berdnikov RN, Bushuev VV, Vassilyev SN, etc. The concept of intellectual power system of Russia with actively adaptive network. Ed. by V.
E. Fortov, A. A. Makarov. M.: JSC "FGC UES"; 2012. 235 p.
12. Vassilyev SN. Formalization of knowledge and control on the basis of positively-formed languages. Information technologies and computing systems 2008; 1: 3-17.
13. Zakharov VN, Ulyanov SV. Fuzzy models of intelligent industrial controllers and control systems. III. Design methodology, Izv. Russian Academy of Sciences. Technical Cybernetics. 1993; 5: 197-220.
