Using soft computing methods in the design of intelligent controllers

Number of pages: 81 File Format: word File Code: 30923
Year: 2013 University Degree: Master's degree Category: Electrical Engineering
  • Part of the Content
  • Contents & Resources
  • Summary of Using soft computing methods in the design of intelligent controllers

    Obtain a master's degree

    in the field of electrical engineering, electronics orientation

    Abstract

    Mechanization of tools is one of the most important and extensive fields used in production processes. Due to the complexity and uncertainty of machining processes, recently soft computing techniques [1] based on physical models have been preferred to predict the performance of machining processes and optimize them over conventional methods. The main soft computing tools used for this purpose are neural networks[2], fuzzy set theory, genetic algorithms, annealing simulations[3], ant colony optimization and particle swarm optimization. In advanced systems that try to control a part or a device, using a robot is an important matter. The degree of control over the behavior and movements of this robot, depending on the purpose of its use, may include a wide range. For example, not much accuracy is needed in industrial applications, but very high accuracy is expected in the case of robots used in medical fields. This series examines the use of soft computing methods in intelligent control processes and examines the results of using these methods in the design of a control system of a robot with flexible behavior, which results show a significant improvement in the mechanization and design of control systems with the mentioned methods.

    Key words: soft computing, intelligent controllers, robot control

    [1] Soft Computing (SC)

    [2] Artificial Neural Networks (ANN)

    [3] Simulated Annealing

    The main idea of ??soft computing was introduced in 1981 by Professor Zadeh [1] in the first article published by him called "What is soft computing". In his definition, soft computing was a combination of several areas, including fuzzy logic, neural computing[2], evolutionary processes, genetic computing and statistical computing.

    This domain leads to the combination of methods that are used to model the behavior of complex systems in the real world (practical and non-theoretical applied systems) that are often impossible or very difficult to model using the computational laws of absolute mathematics and hard logic[3]. However, by using soft computing methods, practical and practical implementations and simulations can be provided for them.

    The advantage of soft computing is the behavior that it shows against non-deterministic, approximate and relative systems. This advantage makes it show a behavior similar to humans (which has a very high generalization ability).

    To show the development process of the boundaries of the field of soft computing, the following relationship can be considered:

    Table 1-1: Development process of soft computing

    Fuzzy logic

    +

    Neural networks

    +

    Computations Evolutionary

    =

    Soft Computing

    Zadeh

    1965

    McCulloch

    1943

    Rechenberg

    1960

    Zadeh

    1981

    Evolutionary computing which is one of the components of soft computing It includes several subsets as follows:

    Table 1-2: Development of soft computing subsets

    Genetic algorithm

    +

    Evolutionary programming

    +

    Evolution strategy

    +

    Genetic discussions

    =

    Computation Evolutionary

    Holland

    1970

    Fogel

    1962

    Rechenberg

    1965

    Koza

    1992

    Rechenberg

    1960

    Now we can express the definitions, goals and importance of soft computing. Then related areas such as fuzzy computing, neural computing, genetic algorithm and so on.

    1-1-Definition of soft computing (SC)

    The definition of soft computing by Professor Zadeh in 1992 is as follows: "Soft computing is an emerging method for performing calculations in parallel with the remarkable ability of the human mind, reasoning and learning in an environment full of ambiguity and inaccuracy."

    Soft computing includes several examples of computing fields. The following are among them: Fuzzy systems: systems based on knowledge and awareness by if-then propositions. These systems are the core of a soft calculation.These systems are sometimes used alone and sometimes combined and shared to model the systems of the surrounding world. The progress of soft computing methods is not limited to these systems and is still expanding. 1-2- Objectives of soft computing Soft computing method is considered a newer field compared to most multidisciplinary modeling and solution methods [4] due to its great variety. In structures that need to build a smart system based on artificial intelligence - which require smart calculations - soft computing methods can be used with high confidence. Problems that are often not possible to model with mathematical rules.

    In problems that are mixed with approximation[5], uncertainty[6], inaccuracy[7] and relative accuracy[8], the use of soft computing methods is the best choice to reach decisions similar to a human's decisions.

    Approximation: Here, the features of the model are very similar to the real sample, but not exactly the same.

    Uncertainty: The belief that there is about the features is that there is no 100% certainty of their correctness.

    Inaccuracy: The features of the model are not the same as the features of the real sample, but they are very close to them.

    1-3- The importance of soft computing

    As the name suggests as well as the explanations that went on it so far, these calculations are different from hard calculations [9]. Unlike hard computing methods, soft computing has good flexibility against inaccuracy, uncertainty, approximation and relative accuracy. The importance of using soft computing methods is determined when, with less cost and time, higher accuracy, and more flexibility, non-linear and complex physical systems can be modeled in such a way that they correspond with expert human decisions with a high percentage of accuracy.

    The important point is that soft computing is not exactly a combination[10], mixture[11] or integration[12], while soft computing is considered a partnership that each member They move towards their intended goal in their own unique way. Basically, the main component in soft computing is complementarity, not competition. Therefore, soft computing is considered an emerging foundation in cognitive intelligence. Chapter 2: Fuzzy computing, neural computing and algorithms based on genetics and particle swarm algorithm 2-1 Fuzzy logic in the real world of fuzzy knowledge[13] have a vague, imprecise, inaccurate, uncertain and incomprehensible behavior. Of course, analyzing and making decisions about these issues is not a problem for humans due to having the level of reasoning, generalizability and flexibility, reasoning, inference and perception of work. Also, due to his frequent involvement with that problem or similar problems, he often provides a suitable analysis for it from his pre-learned knowledge in order to model and solve the problem. Now the question is whether a car also has this ability. The answer is definitely negative. Machines and robots with automatic operation and without human intervention have not been able to make decisions like a human until today. But with the emergence of artificial intelligence systems, adding the ability to learn and benefit from soft computing methods, it is possible to provide a system that makes decisions close to a human's decision in a certain direction.

    Computational systems that are based on classical permutation theory or work from binary logic are not able to answer all the questions that humans answer, and if they do have an answer, in most cases the error will be very high.

    Although the assumption of a machine functioning like a human is considered an ideal, but the expectation that the presented system can find the meaningful relationships of a problem (with an acceptable percentage of error) is an acceptable expectation. It is obvious that when dealing with a problem that has uncertainty, one should have a flexible behavior. 2-1-1- The difference between fuzzy sets and classical sets. In classical sets, a member of the reference set is a member of the A set or is not a member of the A set. For example, consider the reference set of real numbers. The number 2.5 is not a member of the set of integers, while the number 2 is a member of this set. In other words, belonging to the number 2.

  • Contents & References of Using soft computing methods in the design of intelligent controllers

    List:

    Title

    Page number

    Presentation A

    Acknowledgment B

    Abstract T

    List of contents C

    List of figures H

    List of tables H

    1: Introduction of basics and main concepts 1

    1-1- Introduction. 2

    1-2- Definition of soft computing (SC) 3

    1-3- Objectives of soft computing. 4

    1-4- The importance of soft computing. 5

    2: Fuzzy computing, neural computing and algorithms based on genetics and particle swarm algorithm 6

    2-1- Fuzzy logic. 7

    2-1-1- The difference between fuzzy sets and classical sets. 8

    2-1-2- Dry and non-dry sets. 9

    2-1-3- description of fuzzy sets. 10

    2-1-4- The process of using fuzzy logic. 11

    2-1-5- Fuzzy logic and its connection with artificial intelligence. 13

    2-2- neural networks. 14

    2-2-1- An introduction to artificial neural networks. 14

    2-2-2- similarity with the brain. 14

    2-2-3- Artificial neural networks. 17

    2-2-4- artificial nerve cell. 18

    2-2-5- Structure of artificial neural networks and their function 19

    2-2-6- Division of neural networks based on structure. 21

    2-2-7- Division of neural networks based on learning algorithm. 22

    2-2-8- A general view on network education. 23

    2-3- evolutionary optimization algorithms. 25

    2-4- Genetic algorithm. 26

    2-4-1- Introduction. 26

    2-4-2- Chromosome display. 29

    2-4-3- Encoding maps. 31

    2-4-4- Population initialization. 32

    2-4-5- Proportion function. 33

    2-4-6- genetic operators. 34

    2-4-7- selection methods. 38

    2-5- Particle Swarm Algorithm (PSO) 40

    3: Application of fuzzy logic in mobile robots 44

    3-1- History. 45

    3-2- Introduction. 45

    3-3- Reasons for using fuzzy controllers. 46

    3-4- The structure of a fuzzy controller. 47

    3-5- Fuzzy methods used in robots 49

    3-5-1- Position control in moving robots 50

    4: Controller design based on soft computing 56

    4-1- Soft computing techniques. 57

    4-2- Feedback control proportional to derivative and acceleration. 60

    4-3- Multivariable fuzzy logic controllers. 62

    4-4- Fuzzy neural control systems (HFNC) 63

    4-4-1-Radial basis function neural model training (RBFNN). 65

    References: 69

     

     

     

    List of figures

    Title

    Page number

    Figure 2-1: The difference of sets of values ??in (a) fuzzy and (b) non-fuzzy logic. 10

    Figure 2-2: Continuous mapping of membership function values. 11

    Figure 2-3: The amount of temperature dependence determined by an expert to regulate the temperature of the environment. 12

    Figure 2-4: The degree of influence of a value of the corresponding belonging functions. 13

    Figure 2-5: The main characteristics of a biological neuron. 16

    Figure 2-6: The structure of the biological synapse. 17

    Figure 2-7: Network with one neuron. 19

    Figure 2-8: Artificial neural network and different layers in it. 19

    Figure 2-9: Conceptual representation of the optimization process with a genetic algorithm. 29

    Figure 2-10: Repeating GA in successive populations. 32

    Figure 11-2: An example single-point intersection in binary representation. 36

    Figure 2-12: Mutation operation in binary representation. 38

    Figure 2-13: Description of PSO velocity and position of Xi particle in two-dimensional search space. 42

    Figure 2-14: Particle swarm optimization flowchart. 43

    Figure 3-1: Block diagram of a fuzzy controller system. 45

    Figure 3-2: Classification of controller design methods 46

    Figure 3-3: Block diagram to show the components of a fuzzy controller. 47

    Figure 3-4: Schematic of applying the angular command to the robot. 50

    Figure 3-5: Block diagram for generating the rotation angle of the robot. 50

    Figure 6-3: Diagram of the angle applied to the robot and the angle returned from the robot. 51

    Figure 4-1: Proportional structure of derived controller and acceleration. 60

    Figure 4-2: Structure of coupled MIMO fuzzy logic controllers. 63

    Figure 4-3: The structure of a fuzzy-neural hybrid controller. 64

    Source:

    [[1]] D.A. Linkens, H.O. Nyongesa, Learning systems in intelligent control: an appraisal of fuzzy, neural and genetic algorithmNyongesa, Learning systems in intelligent control: an appraisal of fuzzy, neural and genetic algorithm control applications, IEE Proc. Control Theory Appl. 143 (3) (1994) 367–386.

    [[1]] R. Bill VT. Jackson, translation: M. Al-Barzi, familiarity with neural networks, third edition, Sharif University of Technology Publications, 138, 1386.

    [[1]] A. Guderzond Chegini, MATLAB Toolbox, first edition, Naqos Publications, Tehran, 223, 2014.

    [[1]] B. Subudhi, A.S. Morris, Fuzzy and neuro-fuzzy approaches to control a flexible single-link manipulator, J. Syst. Control Eng. Proc. IMechE 217 (2003) 387–399.

    [[1]]    Holland, J. H. (1975). Adaptation in natural and artificial systems: An introductory analysis with applications to biology, control, and artificial intelligence. U Michigan Press.

    [[1]]   Renders, J. M., & Flasse, S. P. (1996). Hybrid methods using genetic algorithms for global optimization. Systems, Man, and Cybernetics, Part B: Cybernetics, IEEE Transactions on, 26(2), 243-258.

    [[1]]    Teodorescu, H. N. L., Kandell, A., & Jain, L. C. (Eds.). (1999). Soft-computing in Human-related Sciences (Vol. 8). CRC press.

    [[1]] Mitchell, M. (1998). An introduction to genetic algorithms (complex adaptive systems) IEEE Conf. on Neural Networks, 1942-1948. M. Jamshidi, Autonomous control of complex systems, vol. 120, 10 May 2001. [[1]] G analysis and design of model-based fuzzy control systems, IEEE Trans. Man Cybernetics 14 (2006) 177-187

    [[1]] F.M. Caswara, H. Unbehauen, A neuro-fuzzy approach to the control of a flexible-link manipulator, IEEE Trans. Syst. Man Cybernetics 18 (6) (2002) 932–944

    [[1]]    W.T. Miller, R.S. Sutton, P.J. Werbos, Neural Networks for Control, MIT Press, Cambridge and MA, (1990).

    [[1]]    M. Brown, C. Harris, Neuro-fuzzy Adaptive Modeling and Control, Prentice-Hall Int. Ltd., London, (1994).

    [[1]] S. Omatu, M. Khalid, R. Yusof, Neuro-Control and its Applications, Springer, London, 1996.

    [[1]] M. Brown, C. Harris, Neuro-fuzzy Adaptive Modeling and Control, Prentice-Hall Int. Ltd., London, (1994).

    [[1]] F. Sun, Z. Sun, L. Li, H. Li, Neuro-fuzzy adaptive control based on dynamic inversion for robotic manipulators, Fuzzy Sets Syst. 134 (1) (2003) 117–133.

    [[1]]    W.T. Miller, R.S. Sutton, P.J. Werbos, Neural Networks for Control, MIT Press, Cambridge and MA, (1990).

    [[1]]    G. Feng, A survey on analysis and design of model-based fuzzy control systems, IEEE Trans. Syst. Man Cybernetics 14 (5) (2006) 676–696.

    [[1]] L. Tian, ??C. Collins, Adaptive neuro-fuzzy control of a flexible manipulator, Mecha-tronics 15 (10) (2005) 1305–1320.

    [[1]]    R.-J. Wai, M.-C. Lee, Intelligent optimal control of single-link flexible robot arm, IEEE Trans. Ind. Electron. 51 (1) (2004) 201–220.

    [[1]]    S. Sarraf, A. Fallah, T. Seyedena, A neurofuzzy controller for a single link flexible manipulator, Lecture notes in computer science, Artificial neural networks, ICANN 2007, vol. 4669, Springer, Berlin/Heidelberg, (2007), pp. 621–629.

    [[1]]    R.-J. Wai, M.-C. Lee, Intelligent optimal control of single-link flexible robot arm, IEEE Trans. Ind. Electron. 51 (1) (2004) 201–220.

    [[1]]    H.A. Talebi, K. Khorasani, R.V.

Using soft computing methods in the design of intelligent controllers