Evolutionary design of interpretable fuzzy controllers

• Autorzy: Hapke M. , Komosiński M.
• Języki publikacji: EN

Abstrakty

EN

This paper presents an approach that allows to evolve fuzzy controllers that can be expressed as fuzzy rules in human-readable form and interpreted. For comparison, the evolution is also performed on simple neural controllers. The control task considered here is a balancing problem, where a construct made of articulated elastic elements is equipped with sensors and actuators. The goal of the construct is to keep the top heavy part from touching the ground. Evolved controllers are evaluated using computer simulation. Control systems process signals from tilt sensors to actuators fixed in the construct. During evolution, fuzzy controllers (including their fuzzy sets and rules) are reconfigured by genetic operators in order to maximize fitness of the control. The article compares evolvability of neural and fuzzy controllers and demonstrates how additional, comprehensible knowledge can be gained which explains the work of the fuzzy controller. The representation for the fuzzy control system, evolutionary operators, various evaluation functions, and the best evolved control systems are presented. A sample evolved fuzzy control system is analyzed in detail to explain its behavior.

Słowa kluczowe

EN

fuzzy control; pendulum; evolution; optimization; simulation; interpretable rules; neural network

• Wydawca: Wydawnictwo Politechniki Poznańskiej
• Czasopismo: Foundations of Computing and Decision Sciences
• Rocznik: 2008
• Tom: Vol. 33, No. 4
• Strony: 351--367
• Opis fizyczny: Bibliogr. 26 poz.

Twórcy

autor

Hapke M.

autor

Komosiński M.

• Poznan University of Technology, Institute of Computing Science, Piotrowo 2, 60-965 Poznan, Poland,

Bibliografia

• [1]Andrew Adamatzky. Software review: Framsticks. Kybernetes: The International Journal of Systems & Cybernetics, 29(9/10):1344-1351, 2000.
• [2]C.W. Anderson. Learning to control an inverted pendulum using neural networks. Control Systems Magazine, IEEE, 9(3):31-37, 1989.
• [3]J. Casillas, O. Cordon, F. Herrera, and M. J. Del Jesus. Genetic tuning of fuzzy rule-based systems integrating linguistic hedges. In Proc. Joint 9th IFSA World Congress and 20th NAFIPS International Conference, volume 3, pages 1570-1574, 25-28 July 2001.
• [4]M. G. Cooper and J. J. Vidal. Genetic design of fuzzy controllers: the cart and jointed-pole problem. In Proc. Third IEEE Conference on Fuzzy Systems IEEE World Congress on Computational Intelligence, pages 1332-1337, 26-29 June 1994.
• [5]David E. Goldberg. Genetic Algorithms in Search, Optimization and Machine Learning. Addison-Wesley Publishing Co., 1989.
• [6]F. Herrera and M. Lozano. Adaptation of genetic algorithm parameters based on fuzzy logic controllers. Genetic Algorithms and Soft Computing, 125, 1996.
• [7]J. Zizka. Learning control rules for takagi-sugeno fuzzy controllers using genetic algorithms. Proceedings of the Fourth European Congress on Intelligent Techniques and Soft Computing, 2:960-964, 1996.
• [8]J.S.R. Jang. Self-learning fuzzy controllers based on temporal backpropagation. IEEE Transactions on Neural Networks, 3(5):714-723, 1992.
• [9]C.Z. Janikow. A genetic algorithm for learning fuzzy controllers. Proceedings of the 1994 ACM symposium on Applied computing, pages 232-236, 1994.
• [10]Maciej Komosinski. Framsticks: a platform for modeling, simulating and evolving 3D creatures. In Andrew Adamatzky and Maciej Komosinski, editors, Artificial Life Models in Software, chapter 2, pages 37-66. Springer, New York, 2005.
• [11]Maciej Komosinski and Adam Rotaru-Varga. Comparison of different genotype encodings for simulated 3D agents. Artificial Life Journal, 7(4):395-418, Fall 2001.
• [12]M.A. Lee and H. Takagi. Dynamic control of genetic algorithms using fuzzy logic techniques. Proceedings of the 5th International Conference on Genetic Algorithms, pages 76-83, 1993.
• [13]P.J. Mac Vicar-Whelan. Fuzzy sets for man-machine interaction. Int. J. Man-Machine Studies, 8:687-697, 1976.
• [14]L. Magdalena, O. Cordon, F. Gomide, F. Herrera, and F. Hoffmann. Ten years of genetic fuzzy systems: current framework and new trends. Fuzzy Sets and Systems, 141(1):5-31, 2004.
• [15]L. Magdalena and F. Monasterio. Evolutionary-based learning applied to fuzzy controllers. Fuzzy Systems, 1995. International Joint Conference of the Fourth IEEE International Conference on Fuzzy Systems and The Second International Fuzzy Engineering Symposium, 3, 1995.
• [16]E. H. Mamdani. Advances in the linguistic synthesis of fuzzy controllers. International Journal of Man-Machine Studies, 8(6):669-678, 1976.
• [17]E.H. Mamdani and S. Assilian. Application of fuzzy algorithms for control of simple dynamic plant. IEE, 121(12):1585-1588, 1974.
• [18]Chris Melhuish, Andrew Adamatzky, and Brett A. Kennedy. Biologically inspired robots. In Yoseph Bar-Cohen, editor, Smart Structures and Materials 2001: Electroactive Polymer Actuators and Devices, volume 4329, pages 16-27, Newport Beach, CA, USA, 2001. SPIE.
• [19]Z. Michalewicz and D.B. Fogel. How to Solve It: Modern Heuristics. Springer Verlag, 2000.
• [20]C. Moraga and K.H. Temme. Functional equivalence between S-neural networks and fuzzy models. Technologies for Constructing Intelligent Systems, 2002.
• [21]Timothy J. Ross. Fuzzy Logic with Engineering Applications. John Wiley and Sons, 2004.
• [22]S.F. Smith. A learning system based on genetic adaptive algorithms. PhD thesis, Department of Computer Science, University of Pittsburgh, 1980.
• [23]Andrea G. B. Tettamanzi. An evolutionary algorithm for fuzzy controller synthesis and optimization. In In Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, pages 22-25, 1995.
• [24]H.B. Verbruggen and R. Babuska. Fuzzy Logic Control: Advances in Applications.World Scientific, 1999.
• [25]S.D. Whitehead and D.H. Ballard. Active perception and reinforcement learning. Neural Computation, 2(4):409-419, 1990.
• [26] R.R. Yager and D.P. Filev. Foundations of fuzzy control. Wiley, New York, 1994.