This article was downloaded by: [Jawaharlal Nehru Technological University] On: 04 July 2013, At: 04:09 Publisher:

Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Intelligent Automation & Soft Computing

Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tasj20

Introducing Belbic: Brain Emotional Learning Based Intelligent Controller

Caro Lucas , Danial Shahmirzadi & Nima Sheikholeslami
a a b c

Center of Excellence for Control and Intelligent Processing Dep. of Electrical and Computer Eng, University of Tehran, and School of Intelligent Systems, IPM,Tehran, Iran E-mail:
b

Center of Excellence for Control and Intelligent Processing, University of Tehran Tehran, Department of Mechanical Engineering Texas A&M University College Station,Iran, Iran, TX 77843-3123 E-mail:
c

Tehran University of Medical Sciences, Tehran, Iran E-mail: Published online: 01 Mar 2013.

To cite this article: Caro Lucas , Danial Shahmirzadi & Nima Sheikholeslami (2004) Introducing Belbic: Brain Emotional Learning Based Intelligent Controller, Intelligent Automation & Soft Computing, 10:1, 11-21, DOI: 10.1080/10798587.2004.10642862 To link to this article: http://dx.doi.org/10.1080/10798587.2004.10642862

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions

Intelligent Automation and Soft Computing, Vol. 10, No. 1, pp. 11-22, 2004 Copyright 2004, TSI Press Printed in the USA. All rights reserved

INTRODUCING BELBIC: BRAIN EMOTIONAL LEARNING BASED INTELLIGENT CONTROLLER CARO LUCAS Center of Excellence for Control and Intelligent Processing Dep. of Electrical and Computer Eng. University of Tehran, and School of Intelligent Systems IPM, Tehran, Iran. Email: lucas@ipm.ir DANIAL SHAHMIRZADI Center of Excellence for Control and Intelligent Processing University of Tehran Tehran, Iran & Department of Mechanical Engineering Texas A&M University College Station, TX 77843-3123 Email: danial.shahmirzadi@tamu.edu NIMA SHEIKHOLESLAMI Tehran University of Medical Sciences Tehran, Iran Email: nima_sheikholeslami@yahoo.com
ABSTRACTModeling emotions has attracted much attention in recent years, both in cognitive psychology and design of artificial systems. Far from being a negative factor in decision making, emotions have shown to be a strong faculty for making fast satisficing decisions. In this paper, we have adapted a computational model based on the limbic system in the mammalian brain for control engineering applications. We applied the proposed controller (termed BELBIC) for some SISO, MIMO and nonlinear systems. Our results demonstrate excellent control action, disturbance handling and system parameter robustness for BELBIC. Key Words: Reinforcement, Emotion, Satisficing, Control, Robust, Learning, Cognitive, Behavioral

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

1. INTRODUCTION
Biologically motivated intelligent computing has in recent years been successfully applied to solving complex problems [1,2,3,4,5,6,7]. Increasingly, researchers appreciate the limitations of traditional approaches when dealing with uncertainties and complexities associated with real-world situations and the possibilities for overcoming these problems inherent in intelligent approaches [8,9]. Bounded rationality and satisficing have been suggested as alternative approaches to fully rational decision making. The so called curse of dimensionality in dynamic programming is an example of the latter problem [3]. Uncertainties also abound in contemporary decision and control engineering tasks [10]. The process or plant to be controlled is often unknown and even after identification; its characteristics may change due to aging, wear and tear, etc. Furthermore, there are various noises and disturbances present in the system. In recent years, model-based approaches to decision making are being replaced by data-driven and rule-based approaches [11,12]. The success of fuzzy and neurofuzzy control systems also proves that solutions based on approximate reasoning that are less dependent on crisp modeling assumptions are more robust with respect to uncertainties present in the environment [13,14,15,16]. Neural networks, evolutionary strategies and genetic algorithms have also been very popular tools both in decision making and other industrial

11

12

Intelligent Automation and Soft Computing

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

applications [17,18,19,20,21]. New approaches where intelligence is not given to the system from outside, but is acquired by the system through learning, have proven much more successful [11,22]. A major difficulty associated with development of learning control systems is the fact that we do not know a-priori the desired control action, i.e. the output of the controller. However, we do have measures for evaluating the performance of the overall system. Various methods for overcoming this problem have been used [22]. Identifying the plant or better yet, identifying the inverse plant or pseudo inverse plant, has been tried in several works. Neurofuzzy control, genetically optimized fuzzy control, back-propagation through plant and reinforcement learning have been successfully used in recent years [22,23,24,25]. A more cognitively based version of reinforcement learning has also been developed in which a critic constantly assesses the consequences of actuating the plant with the selected control action in any given state in terms of the overall objectives or performance measures and produces an analog reinforcement cue which in turn directs the learning in the controller block [22]. This cognitive version of the reinforcement signal has been denoted as an emotional cue, for it is indeed the function of emotions like stress, concern, fear, satisfaction, happiness, etc. to assess the environmental conditions with respect to goals and utilities and to provide cues regulating action selection mechanisms [22,26]. Whether called emotional control or merely an analog version of reinforcement learning with critic (evaluative control) , the method is increasingly being utilized by control engineers, robotic designers and decision support systems developers and yielding excellent results [27,28,29,30,31,32]. Although, for a long time, emotion was considered as a negative factor hindering the rational decision making process, the important role of emotions in human cognitive activities is progressively being documented by psychologists [33,34]. It has now become clear that far from being a negative trait in biology, emotions are important positive forces crucial for intelligent behavior in natural as well as artificial systems [26,35].

2. MODELLING
Motivated by the success in functional modeling of emotions in control engineering applications [25,31,32], the main purpose of this research is to use a structural model based on the limbic system of mammalian brain, for decision making and control engineering applications. We have adopted a network model developed by Moren and Balkenius [28], as a computational model that mimics amygdala, orbitofrontal cortex, thalamus, sensory input cortex and generally, those parts of the brain thought responsible for processing emotions. It is noteworthy that the original development of the model was not motivated by any control application and physiological explanability was the sole consideration. This is significant because its successful utilization in control applications demonstrates the usefulness of emotional interpretation in decision making process. There are two approaches to intelligent and cognitive control. In the indirect approach, the intelligent system is utilized for tuning the parameters of the controller. We have adopted the second, so called direct approach, where the intelligent system, in our case the computational model termed BELBIC, is used as the controller block. The model is illustrated in Figure 1. BELBIC is essentially an action generation mechanism based on sensory inputs and emotional cues. In general, these can be vector valued, although in the benchmarks discussed in this paper for the sake of illustration, one sensory input and one emotional signal (stress) have been considered. The emotional learning occurs mainly in amygdala. The learning rule of amygdala is given in formula (1).
Ga = k1. max(0, EC A)

(1)

where Ga is the gain in amygdala connection, k1 is the learning step in amygdala and EC and A are the values of emotional cue function and amygdala output at each time. A drawback of this model is that the formation of the emotional cues from the sensory data is external to the system, since one expects the emotions to be the output of one part of the brain systems that can be used as input by another part that performs the action selection. In future development of the model, we shall try to augment the proposed system so that the emotional cues are generated within the total system and the functional boundary between emotion formation and action selection disappears. However, to form the proper emotions, the system will still need to be fed some other inputs showing goals or intentions. On the positive side, the ability to feed emotive cues externally enhances the flexibility of the system, because one can contextualize the control through modifying emotions to reflect alternative goals or concerns. A reinterpretation of the model could redesignate the signals fed to amygdale as goal signals and claim that emotions are implicitly generated in

Introducing BELBIC: Brain Emotional Learning Based Intelligent Controller

13

this block and used for action selection in the overall system. The term max in the formula (1)

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

Figure 1. A schematic structure of BELBIC.

is for making the learning changes monotonic; implying that the amygdala gain can never be decreased. This rule is for modeling the incapability of unlearning the emotion signal (and consequently, emotional action), previously learned in the amygdala [28,30]. Similarly, the learning rule in orbitofrontal cortex is shown in formula (2).
Go = k2 .( MO EC )

(2)

where Go is the gain in orbitofrontal connection, k2 is the learning step in orbitofrontal cortex and MO is the output of the whole model, where it can be calculated as formula (3):
MO = A O

(3)

in which, O represents the output of orbitofrontal cortex. In fact, by receiving the sensory input, S , the model calculates the internal signals of amygdala and orbitofrontal cortex by the relations in (4) and (5) and eventually yields the output.
A = Ga .S O = Go .S

(4) (5)

Since amygdala does not have the capability to unlearn any emotional response that it ever learned, inhibition of any inappropriate response is the duty of orbitofrontal cortex.

3. IMPLEMENTATION
Controllers based on emotional learning have shown very good robustness and uncertainty handling properties [31,32], while being simple and easily implementable. To utilize our version of the MorenBalkenius model as a controller, we note that it essentially converts two sets of inputs into the decision signal as its output. We have implemented a closed loop configuration using this block (termed BELBIC) in the feed forward loop of the total system in an appropriate manner so that the input signals have the proper interpretations. The block implicitly implemented the critic, the learning algorithm and the action selection mechanism used in functional implementations of emotionally based (or generally reinforcement learning based) controllers, all at the same time [25,31,32]. The structure of the control circuit we implemented in our study is illustrated in Figure 2. The functions we used in emotional cue and sensory input blocks are given in (6) and (7),

14

Intelligent Automation and Soft Computing

EC = W1 .e + W 2 .CE

(6)

Figure 2. Control system configuration using BELBIC. SI = W3.PO + W4 . PO

.

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

(7)

where EC , CE , SI and PO are emotional cue, control effort, sensory input and plant output and the W1 through W4 are the gains must tuned for designing a satisfactory controller. It is to be noted that the externality of emotion formation means that through choosing the emotional cues, we can implicitly decide our control goals. Here we have added the goal of keeping the control effort as low as possible to the usual goal of tracking the set point so as to implement control that is not cheap (Control systems whose performance measures are independent of the plant input are called cheap). The selected performance measures behave as negative emotions like stress, because their increase shows that the system is not performing satisfactorily. As for sensory input, a PD input mechanism has been selected so as to make the system more responsive to changes. Quicker response is an important characteristic of the limbic system as compared to the cortical areas of brain.

4. SIMULATIONS
We confirmed the capability of BELBIC by performing some simulations. It must be mentioned that in the all simulations outlined below, we implemented the set-point control strategy with the desired value of 1. The descriptions of simulations are given below:

4.2 Linear SISO System: Submarine Model

In this simulation, we considered a simple model of a submarine. The model was supposed to reach the desired depth underwater. The quantitative model is represented via (8).
G(s) = 0.1( s + 1) 2 s ( s 2 + 0.09) = 0.1s 2 + 0.2 s + 0.1 s 3 + 0.09 s

(8)

We implemented the control circuits in MATLAB SIMULINK package. The output of the system with a simple feedback and the output of the system with a BELBIC controller are given in Figure 3. The variation of learning weights in amygdala and orbitofrontal cortex are illustrated in Figures 4 and 5.

Introducing BELBIC: Brain Emotional Learning Based Intelligent Controller

15

Figure 3. Output of the submarine system (a)without controller (b)with a BELBIC.

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

Figure 4. Amygdala learning process.

Figure 5. Orbitofrontal Cortex learning process.

As expected, the learning curve in Figure 4 is monotonic. This is because amygdala can only learn and never unlearn but, orbitofrontal cortex can both inhibit and excite, depending on the contexts previously leaned. Due to strong learning algorithms used in BELBIC, we anticipate a satisfactory robustness of BELBIC, in comparison to similar common controllers, e.g. PID. The responses of the system with appropriate tuned BELBIC and PID controller are illustrated in Figure 6 and the performance indices of the responses are given in Table I.
Table I. Submarine response performance indices with BELBIC and PID controller.

BELBIC PID via:

POS 6.38 5.15

Rise Time 0.02 0.27

Settling time 0.40 1.55

S-S Error % 0.00 0.00

For evaluating the robustness of the controllers, we change the plant given in (8) to the one described
0.1s 2 + 0.1s + 0.2 1.1s 3 + 0.12 s + 1

G(s) =

(9)

The deteriorations in performance indices are given in Table II, which demonstrate that BELBIC is more robust to system parameters variations than similar PID.
Table II. Submarine response performance indices with BELBIC and PID controller under variation in system parameters.

BELBIC PID

POS 5.15 6.26

Rise Time 0.02 0.27

Settling time 0.93 4.10

S-S Error % -1.65 -6.63

Also, we are interested in analyzing the robustness of BELBIC with respect to disturbances. For this purpose, we add a constant disturbance signal to the plant input in (8) and obtain the response of previous BELBIC and PID controllers. The performance indices of these responses are demonstrated in Table III.
Table III. Submarine response performance indices with BELBIC and PID controller under constant external disturbance.

BELBIC PID

POS 09.27 15.66

Rise Time 0.02 0.27

Settling time 0.51 1.68

S-S Error % 1.92 9.95

16

Intelligent Automation and Soft Computing

We can better demonstrate the disturbance robustness of BELBIC in comparison to similar PID, by imposing the above disturbance for a short period of time in the steady state condition of the systems responses. The results are demonstrated in the Figure 7.

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

Figure 6. Submarine model outputs with (a) BELBIC and (b) PID.

Figure 7. Deterioration of submarine model steady state response with a (a) BELBIC and (b) PID

5. NONLINEAR SISO SYSTEM: ONE DOF ROBOT MODEL

For demonstrating generally efficient applicability of BELBIC, we utilized it for a nonlinear model of a robot where the input is the signal of the D.C motor attached to the arm of the robot and the output is its angular position [22]. The differential equation, governing the system is given in (10).
=
..

u 39.2 cos 3.67

(10)

The closed loop response of the system in the absence of any controller is given in Figure 8 and the responses of an appropriately tuned BELBIC and PID controllers are given in Figure 9, respectively. The robustness of BELBIC to external disturbances is also notable for this system. Figure 10 demonstrates the responses given in Figure 9, in the presence of a constant external disturbance during the interval 6 to 9 s.

Figure 8. The closed loop response of the nonlinear model of the one DOF robot with no controller.

Figure 9. The closed loop response of the nonlinear model of the one DOF robot with a (a) BELBIC and (b) PID controller.

Introducing BELBIC: Brain Emotional Learning Based Intelligent Controller

17

6. STRONGLY COUPLED LINEAR MIMO SYSTEM: GAS TURBINE GENERATOR MODEL

In this section, we examine the BELBIC on a multi input-multi output system. The system is a strongly coupled linear model of a gas turbine generator where it has two inputs and two outputs [22]. The inputs are reflux Fuel Pump Excitation and Nozzle Actuator Excitation and the outputs are Gas Generator Speed and Inter-turbine Temperature, respectively. The matrix form of the transfer function of the system is given in (11).

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

0.806 s + 0.264 2 s + 1.15 s + 0.202 G(s) = 1.95 s 2 + 2.12 s + 4.9 3 2 s + 9.15 s + 9.39 s + 1.62

4 3 2 s + 20.8 s + 116.4 s + 111.6 s + 188 s 3 + 12.8 s 2 + 13.6 s + 2.36 7.14 s 2 + 25.8 s + 9.35

(15 s + 1.42)

(11)

In Figures 11 through 13, we demonstrate the results of the above system, without any controller, with 2 BELBIC controllers and with 2 PID controllers, respectively. We can easily observe the appropriate performance characteristics of the BELBIC, especially its very fast responses; however these fast responses cost somewhat high overshoots.

Figure 10. The closed loop response of the nonlinear model of the one DOF robot with a (a) BELBIC and (b) PID controller, in the presence of a external disturbance during the interval 6 through 9 s.

Figure 11. Responses of 2-input, 2-output model of a gas turbine generator without any controller.

Figure 12. Responses of 2-input,2-output model of a gas turbine generator with 2 BELBIC controllers.

Figure 13. Responses of 2-input, 2-output model of a gas turbine generator with 2 PID controllers.

18

Intelligent Automation and Soft Computing

7. AN INDUSTRIAL IMPLEMENTATION: SWITCH RELUCTANCE MOTOR

In order to demonstrate how the proposed algorithm works in a more complex setting, as well as compare it with a nonlinear control algorithm, we show how we can adapt BELBIC to control a switch reluctance electromotor designed as an industrial prototype, the production line of which is currently being investigated [36,37,38]. An optimized nonlinear controller based on QFT design has been reported in [36]. The problem is a nice example for cases where emotional approach is called for. There are several performance considerations like efficiency optimization through feedback, control effort reduction, and set point tracking, which lead to multi-criteria control. Figure 14 shows velocity response results for the novel nonlinear control method reported in [36], while Figure 15 shows similar results obtained through using BELBIC with the emotional cues selected as [39]:

EC = W1 .e + W 2 . CE 2 .dt
Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

(12)

Figure 14. Speed response of the SRM with nonlinear QFT controller.

Figure 15. Speed response of the SRM with BELBIC controller.

8. CONCLUDING REMARKS
In this paper, we introduced an innovative intelligent controller, BELBIC, based on mammalian limbic emotional learning algorithms. We demonstrated the performance of BELBIC by applying it in some simulations and comparing the results of the system with the results of similar classical controllers e.g. PID. The results have shown that the proposed BELBIC has satisfactory control performance. Especially, it was very powerful in disturbance handling and robust with respect to parameter changes. This was due to appropriate learning ability that BELBIC had. We strongly thought that an algorithm for optimum tuning of BELBIC could make it so much more efficient. The proposed network, though very simple and preliminary, could already show the advantages of the approach in control engineering and it can be hoped that a more complete model reflecting the anatomical characteristics of the emotional processing units in mammalian brain will be useful for modeling such important processes as habituation and attention modulation in control engineering and decision support applications. When we consider the performance of BELBIC in the simple examples furnished for showing its usefulness; and even in the more complex industrial example where its performance was compared to that of a custom optimized nonlinear (and not just a linear PID) controller, we can see that it has been able to achieve a lot more than the claimed satisficing control. In fact, its performance has in many respects superceded that of the non-emotional controller. However, the real advantage of BELBIC is not its optimality in any particular application, but its generality and flexibility as well as ability to handle multiple performance criteria, and uncertainty. On the modelling level, the main achievement has been to show its utility in control applications despite the fact that the computational model has initially been developed with purely descriptive motivations. Future extensions of the model can result in a more explicit feeding of belief, desire, intention, and other mental signals, alongside the sensory input to amygdala, and option to get emotion signals as outputs in addition to the selected actions, which at

Introducing BELBIC: Brain Emotional Learning Based Intelligent Controller

19

any rate, are the outputs expected from the controller. The emotion signals can then serve as useful ad hoc performance measures reflecting the cognitive assessments of the controller.

REFRENCES
[1] [2] [3] [4] A. Crameri, S. A. Raillard, E. Bermudez, and W. P. C. Stemmer. DNA Shuffling of a Family of Genes from Diverse Species Accelerates Directed Evolution Nature, 391(6664), January 1998, PP: 288-291 M. Sipper and M. Tomassini. Convergence to Uniformity in A Cellular Automaton via Local Coevolution International Journal of Modern Physics C, Vol. 8, No. 5, 1997, PP: 1013-1024 C. Lucas, D. Shahmirzadi, and M. Biglarbegian. Co-evolutionary Approach for Graph-Coloring Problem Upcoming in the Amirkabir University of Technology Press. S. Hofmeyr and S. Forrest. Architecture for an Artificial Immune System Evolutionary Computation Journal, 2000 J. Chen, E. Antipov, B. Lemieux, W. Cedeno, and D. H. Wood. DNA Computing Implementing Genetic Algorithms Preliminary Proceedings DIMACS Workshop on Evolution As Computation, PP: 39-49, DIMACS, Piscataway NJ, January 1999 D. Shahmirzadi. Proposing A Genetic Algorithm For Solving The Map-Coloring Problem Proceeding of 4th student conference on intelligent systems, March 5-7, 2002, K.N Toosi University of Technology, Tehran, Iran M. Fatourechi, C. Lucas, and A. Khaki Sedigh. An Agent-based Approach to Multivariable Control Proceedings of IASTED International Conference on Artificial Intelligence and Applications, pp: 376-381, Sept 4-7, 2001, Marbella, Spain H. V. D. Parunak. Go to the Ant: Engineering Principles from Natural Multi-Agent Systems Annals of Operations Research, 75 (Special Issue on Artificial Intelligence and Management Science), 1997, PP: 69-101 R. D. Smallwood and E. J. Sondik. The Optimal Control of Partially Observable Markov Processes Over a Finite Horizon Operational Research, (21), 1973, PP:1071-1088 J. H. Steele. Approximation and Validation of Models with Uncertainty: a closed-loop perspective Ph.D. thesis, August 2001, Control Group Department of Engineering, University of Cambridge A. A. Savinov. An Algorithm for Induction of Possibilities Set-Valued Rules by Finding Prime Disjunctions 4th On-line World Conference on Soft Computing in Industrial Applications (WSC4), September 21-30, 1999 A. Ott. The Rule Set Based Access Control (RSBAC) The 8th International Linux Kongress, Enschede, November 28-30, 2001 J. R. Jang. ANFIS: Adaptive-Network-Based Fuzzy Inference System IEEE Transaction on Systems, Man and Cybernetics, Vol. 23, No. 3, 1993 C. Lucas and D. Shahmirzadi. An Interpolative Fuzzy Inference Procedure Using Least Square Principle IASTED International Journal of Control and intelligent systems, Vol. 31, No. 2, June 2003 H. Jeschke. Fuzzy Multi Objective Decision Making on Modeled VLSI Architecture Concepts International Symposium on Circuits and Systems 1998, June 1-3, Monterey, CA, USA C. Lucas, A. Fatehi, D. Shahmirzadi, and M. M. Takami. Implementation of A Genetically Optimized Interpolative Fuzzy Inference Engine in Controlling A Ball-Plate Laboratory Setup ASME 2003 International Design Engineering Technical Conferences, IDETC03, Proceeding of 19th Biennial Conference on Mechanical Vibration and Noise (VIB), Symposium on Smart Dynamical Systems, September 2-6, 2003, Chicago, Illinois C. Lucas, D. Shahmirzadi, M. Nikkhah Bahrami, and H. Ghafoorifard. A FEM-Based Quasi-Static Neuro-Model for Acoustic Noise in Switch Reluctance Mototrs International Journal of Computational Acoustics, Vol. 11, No. 4, December 2003 D. Shahmirzadi, C. Lucas, M. Nikkhah Bahrami, and H. Ghafoorifard. A Computational Approach To Acoustic Noise In Switch Reluctance Drives Proceeding of 3rd International Conference on Mathematical and Computational Applications, ICMCA2002, September 4-6, 2002, Konya, Turkey D. Shahmirzadi, C. Lucas, M. Farshad, and R. Pedrami. Training Neural Network For Modeling The Acoustic Behavior of Switch Reluctance Motor Neuroscience events International conference,

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]

[17] [18] [19]

20

Intelligent Automation and Soft Computing

[20] [21]

[22] [23]

[24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39]

proceeding of 4th Irano-Armenian workshop on neural networks, May 20-22, 2002, School of Intelligent Systems, Institute for studies in theoretical physics and mathematics, Tehran, Iran E. J. Davis and G. Kendall. An Investigation, using Co-Evolution, to Evolve an Awari Player Proceedings of Congress on Evolutionary Computation (CEC2002), PP: 1408-1413, Hilton Hawaiian Village Hotel, Honolulu, Hawaii, May 12-17, 2002 D. Eby, R. C. Averill, W. F. Punch III, and E. D. Goodman. Evaluation of Injection Island GA Performance on Flywheel Design Optimization, January 15, 1998, Proceedings of Third Conference on Adaptive Computing in Design and Manufacturing, Plymouth, England, April, 1998, Springer Verlag C. Lucas, S. A. Jazbi, M. Fatourechi, and M. Farshad. Cognitive Action Selection with Neurocontrollers Third Irano-Armenian Workshop on Neural Networks, August 2000, Yerevan, Armenia M. H. Voigt. Fuzzy evolutionary algorithms June 1992, Technical Report 92-038, International Computer Science Institute (ICSI), 1947 Center Street, Suite 600, Berkeley, CA, 94704 A. Bergman, W. Burgard, and A. Hemker. Adjusting parameters of genetic algorithms by fuzzy control rules New Computer Techniques in Physics Research III, World Scientific Press, 1994, PP: 235--240 X. Xu, H-G He, and D. Hu. Efficient Reinforcement Learning Using Recursive Least-Squares Methods Journal of Artificial Intelligence Research 16, 2002, PP: 259-292 K. Inoue, K. Kawabata, and H. Kobayashi. On a Decision Making System with Emotion 5th IEEE International Workshop on Robot and Human Communication, 1996, PP: 461-465 J. S. Bay. Behavior Learning in Large Homogeneous Populations of Robots IASTED International Conference on Artificial Intelligence and Soft Computing, PP: 137-140, July 1997 J. Moren and C. Balkenius. A Computational Model of Emotional Learning in The Amygdala From animals to animals 6: Proceedings of the 6th International conference on the simulation of adaptive behaviour, Cambridge, Mass., 2000. The MIT Press K. Miyazaki, N. Araki, E. Mogi, T. Kobayashi, Y. Shigematsu, M. Ichikawa, and G. Matsumoto. Brain Learning Control Representation In Nucleus Accumbens 1998 Second International Conference On Knowledge-Based Intelligent Electronic Systems, PP:21-23, April 1998, Australia C. Balkenius and J. Moren. A Computational Model of Emotional Conditioning in the Brain Proceeding of the workshop on Grounding Emotions in Adaptive Systems, Zurich, 1998 M. Fatourechi, C. Lucas, and A. Khaki Sedigh. Reducing Control Effort by means of Emotional Learning Proceedings of 9th Iranian Conference on Electrical Engineering, (ICEE2001), PP: 41-1 to 41-8, May, 2001, Tehran, Iran M. Fatourechi, C. Lucas, and A. Khaki Sedigh. Reduction of Maximum Overshoot by means of Emotional Learning Proceedings of 6th Annual CSI Computer Conference, PP: 460-467, February, 2001, Isfahan, Iran R. Neese. Emotional Disorders in Evolutionary Perspective British Journal of Medical Psycology (1998), 71, PP: 397-415 J. D. Greene, R. B. Sommerville, L. E. Nystorm, J. M. Darley, and J. D. Cohen. An fMRI Investigation of Emotional Engagement in Moral Judgment Science, Vol. 293, 14 Sep. 2001, PP:2105-2108 M. Toda. The Urge Theory of Emotion and Cognition Emotion and urges, SCCS Technical Report, No.93-1-01, 1993 C. Lucas, M. Modir Shanechi, P. Asadi, and P. Melati Rad. A Robust Speed Controller for Switched Reluctance Motor with Nonlinear QFT Design Approach IEEE IAS 2000 Annual Meeting/World Conference on Industry Applications, October 2000, Rome, Italy, PP: 1573-1577 C. Lucas and S. A. Jazbi. Intelligent Motion Control of Electric Motors with Evaluative Feedback International Conference on Large High Voltage Electric Systems, CIGRE (37th Session), 30 August5 September, 1998, Paris, France, PP: 11-104 B. Mirzaeian, M. Moallem, V. Tahani, and C. Lucas. Multi-Objective Optimization Method Based on a Genetic Algorithm for Switch Reluctance Motor Design IEEE Transaction on Magnetics, 38 (3), May 2002, PP: 1524-1527 C. Lucas, B. N. Araabi, O. Namaki, and D. Shahmirzadi. Speed Control of A Switched Reluctance Motor Using BELBIC (Brain Emotional Learning Based Intelligent Controller) Submitted

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

Introducing BELBIC: Brain Emotional Learning Based Intelligent Controller

21

ABOUT THE AUTHORS

Caro Lucas received the Msc degree from the University of Tehran, Iran, in 1973, and the PhD degree from the university of California, Berkeley, in 1976. He is a Professor and the Director of Center of Excellence for Control and Intelligent Processing at the Department of Electrical and Computer Engineering, University of Tehran, Iran, as well as a Researcher at the School of Intelligent Systems (SIS), Institute for Studies in Theoretical Physics and Mathematics (IPM), Tehran, Iran. He has served as the Director of SIS (19931997), Chairman of the ECE Department at the University of Tehran (19861988), Managing Editor of the Memories of the Engineering Faculty, University of Tehran (1979-1991), Reviewer of Mathematical Reviewers ( since 1987), Associate Editor of Journal of Intelligent and Fuzzy systems (1992-1999), and Chairman of the IEEE, Iran section (1990-1992). He was also a Visiting Associate Professor at the University of Toronto (summer, 1989-1990), University of California, Berkeley (1988-1989), an Assistant Professor at Garyounis University (1984-1985), University of California at Los Angeles (1975-1976), a Senior Researcher at the International Center for Theoretical Physics and the International Center for Genetic Engineering and Biotechnology, both in Trieste, Italy, the Institute of Applied Mathematics Chinese Academy of Sciences, Harbin Institute of Electrical Technology, a Research Associate at the Manufacturing Research Corporation of Ontario, and a Research Assistant at the Electronic Research Laboratory, University of California, Berkeley. He is the holder of Patent on Speaker Independent Farsi Isolated Word Neurorecognizer. His research interests include biological computing, computational intelligence, uncertain systems, intelligent control, neural networks, multiagent systems, data mining, business intelligence, financial modeling and knowledge management. Professor Lucas has served as the chairman of several International Conferences. He was the founder of the SIS and has assisted in founding several new research organizations and engineering disciplines in Iran. He is the recipient of several research grants at the University of Tehran and SIS. Danial Shahmirzadi studied in National Organization for Developing Exceptional Talents (Allame Helli Tehran) as his guidance and secondary school, Tehran (1990-1997). He received his BSc degree in mechanical engineering from university of Tehran in 2002. He has been a researcher in several industrial projects e.g. Crank Shaft Balancing Project in IKCO and Design, Construction and Intelligent Control of SR motors Project in SIS, IPM. Recently, his researches are mainly focused on developing brain emotional learning algorithms and adopting them to engineering applications. He is interested in Intelligent Control Systems, Fuzzy systems, Soft computing, Computational intelligence, and Mechanical vibration and noise. He is a student member of ISME (Iranian Society of Mechanical Engineering) and IEEE and has contributed in several national and international journals and conferences. He is now with the Mechanical Engineering Department of Texas A&M University doing his MSc.

Downloaded by [Jawaharlal Nehru Technological University] at 04:09 04 July 2013

Nima Sheikholeslami studied in National Organization for Development of Exceptional Talents (Allame Helli Tehran) as his guidance and secondary school, Tehran (1992-1999). Currently, He is with the Tehran University of Medical Sciences toward his MD degree. He is interested in multidisciplinary research areas.