APPLICATION OF ARTIFICIAL IMMUNE SYSTEM IN DESIGNING POWER SYSTEMS STABILIZER FREDDY PRASETIA BIN RIDHUAN

APPLICATION OF ARTIFICIAL IMMUNE SYSTEM IN DESIGNING POWER SYSTEMS STABILIZER FREDDY PRASETIA BIN RIDHUAN A project report submitted in partial fulfillment of the requirements for the award of the degree of Master of Engineering (Electrical Mechatronics and Automatic Control) Faculty of Electrical Engineering Universiti Teknologi Malaysia MAY, 2007

iii Dedicated to my beloved parents, for their everlasting support and encouragement to complete the course of this study.

iv ACKNOWLEDGEMENT in Alhamdullillah, I am grateful to ALLAH SWT for His blessing and mercy making this project successful. I wish to express my sincere appreciation to my project supervisor Dr. Hj. Mohd. Fauzi Othman for his effort, encouragement and guidance. In preparing this project report, I did a lot of reading and research on past projects, thesis and journals for my reference. They have given me tips and useful information in order for me to complete my analysis and research. To all the lecturers who have taught me, thank you for the lessons you have delivered. I would also like to thank my friends, thank you for their useful ideas, information and moral support during the course of study. Last but not least, I would like to express my heartiest appreciation to my parents, who are always there when it matters most.

v Abstract Biological Immune system is a control system that has strong robusticity and self-adaptability in complex disturbance and indeterminacy environments. This thesis proposes an appropriate artificial immune system algorithm to develop an immune controller. The idea of immune controller is adept and derived from biological vertebrate immune system. Mimicking and imitating of biological immune system or better known as the artificial immune system is thus developed. Applying and implementing of the algorithm of the artificial immune system is to develop an immune controller. There are various model of artificial immune controller but only the most suitable will be selected. The selected artificial immune controller has the resemblance and similarity of a proportional integral derivative controller. The selected immune controller is to be implemented into the power systems stabilizer. The immune controller is to obtain and achieve system goals in enhancing the performance and stability of power systems. The approach is to prove that an immune controller using artificial immune system algorithm can be used as a controller to obtain steady state outt response.

vi Abstrak Sistem kekebalan biologi merupakan sistem kawalan yang memnyai kebolehgunaan dan penyesuaian diri yang kuat dalam menghadapi gangguan yang kompleks dan persekitaran yang tidak diduga. Tesis ini mencadangkan algoritma sistem kekebalan tiruan untuk membangunkan kawalan kekebalan. Idea kawalan kekebalan diperolehi daripada sistem kekebalan biologi daripada haiwan vetebrata. Meniru gaya sistem kekebalan biologi atau lebih dikenali sebagai sistem kekebalan tiruan boleh dicipta. Menggunakan algoritma daripada sistem kekebalan tiruan untuk membangun kawalan kekebalan. Terdapat pelbagai jenis kawalan kekebalan tiruan tetapi hanya yang paling sesuai akan dipilih. Kawalan kekebalan tiruan yang dipilih memnyai ciri-ciri dan persamaan dengan kawalan pengkamilan, pembezaan dan pendaraban. Kawalan kekebalan yang terpilih akan digunakan kedalam sistem penstabilan kuasa. Kawalan kekebalan bertujuan untuk mencapai matlamat dalam meningkatkan keupayaan dan menstabilkan sistem kuasa. Capaian ini adalah untuk membuktikan bahawa kawalan kekebalan menggunakan algoritma sistem kekebalan tiruan boleh digunakan sebagai kawalan untuk mencapai tindak balas keluaran yang stabil.

vii TABLE OF CONTENTS CHAPTER TITLE PAGE DECLARATION DEDICATION ACKNOWLEDGEMENT ABSTRACT ABSTRAK TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES LIST OF SYMBOLS LIST OF ABBREVIATIONS ii iii iv v vi vii xi xii xv xvi INTRODUCTION. Introduction.2 Objectives 2.3 Scope of Work 3.4 Expected Contribution 4

viii 2 ARTIFICIAL IMMUNE SYSTEM 5 2. Introduction 5 2.2 Innate Versus Acquired Immunity 7 2.2. Innate Immunity 7 2.2.2 Acquired Immunity 7 2.3 Antigens 8 2.4 Immune Cells 9 2.5 B-Cells and Antibodies 9 2.6 T-Cells and Lymphokines 0 2.7 Macrophages 0 2.8 An Overview of the Immune System 0 2.8. Humoral Response 2.8.2 Cell Mediated Response 2 2.9 Analysis of Lines of Defense 3 2.0 Memory Cells 3 2.0. Memory T Cells 4 2.0.2 Memory Helper T Cells 4 2.0.3 Memory B Cells 5 3 LITERATURE REVIEW ON APPLICATION OF AIS 6 3. Introduction 6 3.2 Comter Security 9 3.3 Anomaly Detection in Time Series Data 20 3.4 Fault Diagnosis 22 3.5 Pattern Recognition 23 3.6 Autonomous Agents 25

ix 4 PROPOSITION OF ARTIFICIAL IMMUNE 28 CONTROLLER ALGORITHM 4. Introduction 28 4.2 Basic Varela Immune Network Model 29 4.3 Improved Varela Immune Network Model 3 4.4 Design and Analysis of Immune Controller 32 4.5 Sample Simulation Result 36 4.6 Analysis of IVINC parameters 38 5 POWER SYSTEMS STABILIZER BY AIS 40 5. Introduction 40 5.2 Fixed Parameter Controllers 4 5.3 Conventional PSS 4 5.4 Artificial Immune System PSS 43 5.5 The Two Area Test Systems 46 5.6 Result and Analysis 50 5.6. Delta w PSS Controller 5 5.6.2 Multi Band_PSS Controller 59 5.6.3 Comparison IVINC PSS with Delta w PSS 67 5.6.4 Comparison IVINC PSS With Multi Band PSS 85 5.6.5 Comparison IVINC PSS with Delta pa PSS 93 5.6.6 Analysis of IVINC PSS Controller 0 5.6.7 IVINC pa PSS Controller 09 5.6.8 Comparison IVINC pa PSS With No PSS 0 5.6.9 Comparison IVINC pa PSS With Delta w PSS 23 5.6.0 Comparison IVINC pa PSS With Multi Band PSS 39 5.6. Analysis of IVINC pa PSS Controller 47 5.7 Summary of The Analysis 62

x 6 CONCLUSION 59 6. Conclusion 59 6.2 Future Works and Recommendation 60 REFERENCES 6

xi LIST OF TABLES TABLE NUMBER TITLE PAGE 5.5. Parameter of Delta w PSS 5 5.5.2 Parameter of Multi Band PSS 59 5.5.3 Range of Gain K and Gain K3 67 5.5.4 Ideal value of gain k, 0.7<k<.0 3 5.5.5 The effect of changing value k in IVINC pa PSS 50 5.6. Delta w and Multi Band PSS 62 5.6.2 The int of speed deviation with respect 63 of nominal (dw in ) 5.6.3 The int of power acceleration with 64 respect to nominal (pa=pm-pe in )

xii LIST OF FIGURES FIGURE NUMBER TITLE PAGE 2. Antigen Antibody Interaction 8 2.2 Production of Antibodies 9 2.3 An Overview of an Immune System 2.4 Humoral Response 2 2.5 The cell Mediated Response 3 4. M(σ I ) is the mature function of the B i cell 30 4.2 The M(σ) and P(σ) curve of actual control system 34 4.3 A biological immune controller 35 4.4 Artificial immune control system structure 36 4.5 Pulse Generator Int to the System 37 4.6 Simulation result of the outt system response 38 4.7 Simulation result of the outt system response when k3=50 39 4.8 Simulation result of the outt system response when k=0.25 39 5. System Model Used In the PSS Simulation 42 5.2 Improved Varela Immune Network Controller 45 5.3 Implementation of IVINC into the PSS 46 5.4 Test Area System 47 5.5 Generator and 2 of Area and Generator 3 and 4 of Area 2 48 5.6 PSS Controllers 49 5.7 Delta W Controller 5 5.8 a), b), c) and d) 52,53

xiii 5.9 a), b), c) and d) 55,56 5.0 a), b), c) and d) 57,58 5. Multi Band Controller 59 5.2 a), b), c) and d) 60,6 5.3 a), b), c) and d) 63,64 5.4 a), b), c) and d) 65,66 5.5 a), b), c) and d) 68,69 5.6 Speed deviation Difference of Gen and Gen 2 70 5.7 a), b), c) and d) 72,73 5.8 a), b), c) and d) 74,75 5.9 a), b), c) and d) 77,78 5.20 Speed deviation Difference of Gen and Gen 2 79 5.2 a), b), c) and d) 8,82 5.22 a), b), c) and d) 83,84 5.23 a), b), c) and d) 86,87 5.24 a), b), c) and d) 89,90 5.25 a), b), c) and d) 9,92 5.26 a), b), c) and d) 94,95 5.27 a), b), c) and d) 97,98 5.28 a), b), c) and d) 99,00 5.29 a), b), c) and d) 02,03 5.30 a), b), c) and d) 05,06 5.3 a), b), c) and d) 0,02 5.32 a), b), c) and d),2 5.33 a) and b) 4 5.34 a) and b) 5 5.35 a) and b) 6 5.36 a) and b) 7 5.37 a), b), c) and d) 9,20 5.38 a), b), c) and d) 2,22 5.39 a), b), c) and d) 24,25 5.40 a), b), c) and d) 27,28 5.4 a), b), c) and d) 29,30 5.42 a), b), c) and d) 32,33

xiv 5.43 a), b), c) and d) 35,36 5.44 a), b), c) and d) 37,38 5.45 a), b), c) and d) 40,4 5.46 a), b), c) and d) 43,44 5.47 a), b), c) and d) 45,46 5.48 a), b), c) and d) 48,49 5.49 a) and b) 5 5.50 a) and b) 52 5.5 a) and b) 53 5.52 a) and b) 54 5.53 a) and b) 55 5.54 a) and b) 56 5.55 a), b), c) and d) 58,59 5.56 a), b), c) and d) 60,6

xv LIST OF SYMBOLS Ti - quantity of the antibody Bi - quantity of B cell K - mortality of the antibody which is caused by the antibody interaction K2 - natural mortality of the antibody K3 - reproduction rate of antibody which is caused by the mature B cell K4 - mortality of the B cell K5 - reproduction rate of B cell which is caused by the B cell itself K6 - new reproduction rate of B cell which is caused by the bone marrow M(σi) - mature function of the Bi cell P(σi) - reproduction function of which the Bi cells reproduce the Ti antibody Q - reproduction rate of the antigen when the immune process doesn t exist Ke - approximate rate of antigen s being specially eliminate Ag - the reproduction of antigen e(t) - error of the control system u(t) - outt of the immune controller f(e,u) - immune controller G(s) - object controlled by the immune controller r(t) - int signal y(t) - outt response

xvi LIST OF ABBREVIATIONS AI - Artificial Intelligence AIS - Artificial Immune System ANNPSS - Artificial Neural Network Power Systems Stabilizer APCs - Antigen Presenting Cells APSS - Adaptive Power Systems Stabilizer AVR - Automatic Voltage Regulator BVINM - Basic Varela Immune Network Model CPSS - Conventional power System Stabilizer DARS - Distributed Autonomous Robotic System FLCPSS - Fuzzy Logic Controller Power System Stabilizer GA - Genetic Algorithm NFPSS - Neuro Fuzzy Power Systems Stabilizer PSS - Power Systems Stabilizer IVINC - Improved Varela Immune Network Controller IVINM - Improved Varela Immune Network Model VINM - Varela Immune Network Model

CHAPTER INTRODUCTION. Introduction The successful operation of a power system depends largely on the engineer s ability to provide reliable and uninterrupted service to load. The reliability of the power supply implies much more than merely being available. Ideally, the loads must be fed at constant voltage and frequency at all times. In practical terms this means that both voltage and frequency must be held within close tolerances so that the consumer s equipment may operate satisfactorily. For example, a drop in voltage of 0-5% or a reduction of the system frequency of only a few hertz may lead to stalling of the motor loads on the system. Thus it can be accurately stated that the power system operator must maintain a very high standard of continuous electrical service. Electrical power systems are among the largest structural achievement of man. Some transcend international boundaries, but others supply the local needs of a ship or an aero-plane. The generators within an interconnected power system usually produce alternating current and are synchronized to operate at the same

2 frequency. In a synchronized system, the power is naturally shared between generators in the ratio of the rating of the generators, but this can be modified by the operator. Systems which operate at different frequencies can also be interconnected, either through a frequency converter or through a direct tie. A direct current tie is also used between system that, while operating at the same nominal frequency, have difficulty in remaining in synchronism if interconnected. Conventional power systems stabilizers contain a phase lag/lead network for phase compensation has played a very significant role in enhancing the stability of power systems. There are various new approaches based on modern control and artificial intelligence techniques to improve the performance of the power systems stabilizer being proposed during the past 30 years. Although it is feasible to develop a satisfactory stabilizer using any one of these techniques, each has its unique strengths and drawbacks. One of the proposed techniques is the application of artificial immune system to power system stabilizer. This paper proposes an optimization algorithm imitating the immune system to design power systems stabilizer in enhancing the stability of power and to improve damping of low frequency oscillations using a suitable artificial immune algorithm..2 Objectives The objectives of this thesis are to study and analyze for the mathematical model and algorithm of artificial immune system. Here are various types of mathematical mode of immune algorithm can be found from books, journals, thesis papers, internet etc. The artificial immune algorithm to be chosen in this analysis must have the similarity or heuristic between the artificial immune controller and the control system itself. By using a selected artificial immune algorithm an immune controller is to be developed. The immune controller is then tested and simulated using MATLAB Simulink to observe its outt response and performance. Once the desired immune controller is obtained, the immune

3 controller is implemented to a power systems stabilizer. The application of this immune controller is to design a power systems stabilizer which optimizes the performance of power systems and enhances the stability of power. The main objective of the immune controller is to enhance the quality of the control system and the damping of low frequency oscillations in the power systems stabilizer. From the various parameters of the IVINC controller, we can conduct analysis from the simulation results to obtain steady state outt response. These parameters will be the guideline or reference for the implementation of further test and analysis of IVINC controllers. The IVINC controller will be implemented into the two area test system of the power systems stabilizer. The IVINC controller will be pair with other conventional controller using various combinations to analyze the systems outt response. The rpose of the analysis is to compare between the IVINC and the conventional controller in obtaining stable outt response. Different combinations of controllers produce different outt response, stability, settling time and peak..3 Scope of Work The scope of work is to study and analyze various mathematical model of immune algorithm in order to design immune controller. The mathematical model of the immune algorithm must have the quality or other relation or characteristics of the control system. With a selected artificial immune system elements and algorithm the rpose of the project is to design an artificial immune controller. The controller then has to be tested and simulated using a MATLAB Simulink. Once the appropriate immune algorithm has been obtained, we can use it to design a power systems stabilizer. The artificial immune system algorithm technique can be used to develop a satisfactory stabilizer so as to enhance the stability of the power system. The immune controller is to be implemented into the power system

4 transfer function using MATLAB Simulink. From there we can observe the outt response. Improvement and adjustment of the immune controller variables need to be conducted from time to time in order to obtain a good result and performance of the outt response of the power system..4 Expected Contribution The artificial immune controller is the first method to be implemented to the two test area system of the power systems stabilizer. Through analysis and simulation it is observed that IVINC controller can perform as well as other controllers in achieving stability. The IVINC Controller is able to produce good simulation result in damping low frequency oscillation in power systems just like other conventional controllers. Furthermore, IVINC controllers can be implemented and applied in other control system applications.

5 CHAPTER 2 INTRODUCTION TO ARTIFICIAL IMMUNE SYSTEM 2. Introduction The main function of the immune system is to protect the body from pathogens and cancer. Vertebrate immune systems are more complex than the invertebrates. They are characterized by two important properties, which are memory and specificity. In the case of invertebrate, the immune system consists mainly of Phagocytes which are nonspecific. This means that it will not remember any previous antigen, and will use the same attacking strategy each time. Phagocytes has no receptors for specific pathogens, which means that these cells will engulf and try to kill any pathogen. On the other hand, the vertebrate host has evolved more specialized cells called Lymphocytes. These Lymphocytes are pathogen specific, which means that they have distinct receptors to interact with different pathogens. To combat antigens, nature has provided us with the immune system. The blood, lymph nodes, and bone marrow act with the liver, spleen, thymus, and tonsils to produce and deliver specialized cells, including B- lymphocytes, T lymphocytes, and phagocytes. These cells limit the severity and duration of colds, Fight infections in the nose and throat, help wounds to heal, destroy some cancers, and much more.

6 There are two types of immune models [3,5]: ) Immune model based on the immune system theory (mainly clones choice theory nowadays). a) The somatic theory describes that somatic recombination and mutation contribute to increasing the diversity of antibody. b) The network hypothesis describe that a mutual recognition network among antibody contributes to control of the proliferation of clones. 2) Immune network model based on the immune network theory. a) All the continuous immune network models at present are the ordinary differential equation of time, which conforms to the real control system. b) The discrete immune network model is not the common discrete model based on time control system, but it means that the immune cells or molecules are separated among each others. Figure 2.0: Types of Immune Models

7 2.2 Innate Versus Acquired Immunity There are two types of immunity, innate immunity and adaptive or acquired immunity. Also, the immune system response can be divided into humoral immunity, and cell mediated response. 2.2. Innate Immunity The innate immunity can be regarded as natural resistance of the host to foreign pathogens. There are a number of external and internal lines of defenses in the innate immunity. As an examples we find Lysozymes in tears, and skin inflamation as a resistance to a peneterating pathogen. The innate immunity is the first line of defense against the foreign pathogens, and it uses the non-specific strategy while attacking it. Phagocytes engulf the foreign pathogen, and try to kill it. Some examples on the same line of defense are Monocytes, Macrophages, and Neutrophils. There are other types of cells that is called Natural killer cells NKcells that also use non-specific response to protect the host against the foreign pathogen. 2.2.2 Acquired Immunity In contrast to the innate immune system, the acquired immune system uses a specific response to pathogens. The important advantage of the acquired immunity is the use of memory through lymphocytes. After getting rid of the foreign pathogen the lymphocytes change into memory cells. These memory cells will recognize rapidly the same pathogen when it evades the host again, and eliminate it before causing any damage. The two major types of lymphocytes are T-cells, and B-cells. B-cells have direct contact with the antigen when interacting

8 with it. On the other hand, T-cell can bind to the antigen only after it is processed and presented by other cells. B-cells are the basic building block of the humoral immunity through the production of antibodies. Cell mediated immunity is contributed by T-cells mediated response. Tcells have many forms like the helper T-cell which helps either B-cells, or phagocytic macrophages. Another form that the T-cell can be is the cytotoxic T-cells, which recognize cells infected by virus or cancer, and eliminate them. 2.3 Antigens An antigen (Ag) can be defined as a substance that triggers specific immune response. In vertebrates, the host system does not respond to its own proteins, and that is called tolerance. T-cells and B-cells that are capable of recognizing self-cells are eliminated during maturation phase,. An antigen may carry several epitops, and consequently this will trigger the production of several antibodies, see Figure 2.. Generally, T or B cells do not recognize all of these epitopes, instead they recognize part of it. So, a single Ag may attract the attention of several T or B cells. Also, two different antigens may carry the same crossreactive epitopes, which means that an antibody produced for that antigen can interact with another one. Figure 2. Antigen Antibody Interactions 2.4 Immune Cells

9 Cells destined to become immune cells are produced in the bone marrow. The descendants of some stem cells become lymphocytes, while others develop into a second group of immune cells known as phagocytes. The two major classes of lymphocytes are B cells and T cells. B cells complete their maturation in the bone marrow. On the other hand, T cells migrate to the thymus; an organ that lies high behind the breastbone. Each lymph node contains specialized compartments that house a great number of B lymphocytes, T lymphocytes, capable of presenting antigen to T cells. Thus, the lymph node brings together the several components needed to start an immune response. 2.5 B-Cells and Antibodies B-Cell is one of the major arms of the immune system mechanisms, and it is responsible for the humoral response. The name humoral comes from these fluids that circulate around the body known as humors. Each B cell is programmed to make one specific antibody. When a B cell encounters its triggering antigen, it produces many large plasma cells. Every plasma cell is a factory for producing antibody. Each of the plasma cells descended from a given B cell produces millions of identical antibody molecules and pours them into the bloodstream, see Figure 2.2. A given antibody matches an antigen as a key matches a lock, and marks it for destruction. Figure 2.2 : Production of Antibodies 2.6 T-Cells and Lymphokines

0 T-Cells play two rolls in the immune system defense. B cells cannot make antibody against most substances without regulatory T-cell help. On the other hand, Cytotoxic Tcells, directly attack body cells that are infected. Another important regulatory T cells are "helper" cells. Typically identifiable by the T4 cell marker, helper T cells activate B cells and other T cells as well as natural killer cells and macrophages. Another subset of T cells contributes by turning off or "suppress" these cells. T cells work by secreting cytokines or, Lymphokines which are considered to be chemical messagers. 2.7 Macrophages Macrophages are responsible for carrying the initial attack against an invasion launched by antigens. Macrophages are distributed throughout body tissues, and they rid the body of worn-out cells and other debris. Foremost among the cells that "present" antigen to T cells, having first digested and processed it, macrophages play a crucial role in initiating the immune response. As secretory cells, Monocytes and Macrophages are essintial to the regulation of immune responses and the development of inflammation; they produce an array of powerful chemical called Monokines including enzymes, complement proteins, and regulatory factors such as interleukin-. Sometimes antigens change themselves, and that is why we continue to get sick. 2.8 An Overview of the Immune System When foreign antigen enters the body, it triggers B-cells to produce antibodies, which bind to the antigen and clear it from the body; this is called Humoral immune response. The cell-mediated response involves helper T-cells and T cytotoxic (CTL) cells. Helper T-cells (Th) can be divided into two sub

fields: Th and Th2. Th cells help B-cells, where Th cells activate macrophages. CTL cells kill virtually infected or Cancer cells, see Figure 2.3. Figure 2.3 : An Overview of an Immune System 2.8. Humoral Response When the B-cell proliferates, all of its descendants will make this uniquely rearranged set of antibodies. B-cells continue to multiply, various mutants arise; these allow for the natural selection of antibodies that provide better and better "fits" for antigen elimination. The result of this entire process is that a limited number of B-cells can respond to an unlimited number of antigens. Antibodies are triggered when a B-cell encounters its matching antigen, and digest it. Antigen fragments are displayed on B-cell distinctive markers. The combination of antigen

2 fragments, and marker molecules attract the mature matching helping cells. T-Cells secrete Lymphokines allow B-cells to multiply and mature into antibody producing Plasma cell. Antibodies are released into the blood stream, and they lock into matching antigens. These antigen-body complexes are soon overcome either by the complement cascade, or by the liver and spleen, see Figure 2.4. Figure 2.4 : Humoral Response 2.8.2 Cell Mediated Response Machrophages initiate the cell mediated response, or by other antigenpresenting cell. The antigen-presenting cell digest the antigen, and then displays antigen fragments on its own surface. Bound to the antigen fragment is an MHC molecule. These fragments capture the T cell's attention. A T cell whose receptor fits this antigen binds to it. This bond stimulates the antigen-presenting cell to secrete Interleukins required for T cell activation and performance, see Figure 2.5.

3 Figure 2.5 : The cell Mediated Response 2.9 Analysis of Lines of Defense The human immune system attempts to quickly control the spread of antigens once they have been identified. There are several other lines of defense against antigens besides the immune system. The first line of defense is the skin, which prevents the invasion of most micro organisms. The proteins and acidity of the saliva in the mouth and stomach digest harmless microorganisms. However, if there is a cut in the skin, or fluid transmission occurs, pathogens can invade the body. The second line of defense is the cell-mediated response of the immune system. Macrophages are circulating throughout the body that destroy the invading microorganisms by phagocytosis. The last line of defense is known as the humoral immune response. Many types of immune cells are triggered to move into the affected area, and a great deal of antibodies and phagocytes destroy the invading antigen. 2.0 Memory Cells Some of the lymphocytes activated during the primary immune response remain dormant and keep circulating in the immune system for a long time. These lymphocytes carry the memory of the encountered antigen, and therefore these long-lived cells are called memory cells. Memory can also be maintained by longlived antigen (not necessarily by a polation of long-lived distinct memory cells).

4 Whenever T cells and B cells are activated, some of the cells become "memory" cells. Then, the next time that an individual encounters that same antigen, the immune system is primed to destroy it quickly. The degree and duration of immunity depend on the kind of antigen, its amount, and how it enters the body. An immune response is also dictated by heredity; some individuals respond strongly to a given antigen, others weakly, and some not at all. 2.0. Memory T Cells Memory T cells are formed during an immune response. As the term implies, memory T cells remember past attacks by antigens, and can respond with increased strength during subsequent invasions by a particular pathogen. Memory T cells are long lasting immune cells, and react to particular antigens. Unlike T cells that recirculate in the blood and lymph, memory T cells often circulate throughout the entire body, especially in the site they were originally activated. Memory T cells rely on memory helper T cells for launching a global immune response. Look below for links 2.0.2 Memory Helper T Cells Memory helper T cells are also known as memory effector T cells. Memory helper T cells are used by memory T cells to launch an immune response against an attack by pathogens. Memory helper T cells react in much the same way as helper T cells, except that they are stimulated by memory T cells. Memory helper T cells can differentiate into a cytotoxic T cell that attacks abnormal cells, or into a helper T cell that stimulates an immune response from B cells. Look below for links to other immune cells.

5 2.0.3 Memory B Cells Little is currently known about memory B cells. However, memory B cells are probably similar to memory T cells in that they retain a strong affinity to low concentrations of antigen, and are able to launch a strong immune response following stimulation by a particular antigen that they are sensitive to. Like normal B cells, memory B cells circulate throughout the entire body. However, they are significantly longer lived, on scales of a few months to years. Look below for links to other immune cells.

6 CHAPTER 3 LITERATURE REVIEW ON APPLICATION OF ARTIFICIAL IMMUNE SYSTEM 3. Introduction Recently researchers have begun to argue that intelligent behavior and cognition are much more about effective interaction between agent and environment, rather than an agent s capability to handle abstract world models internally. Based on these influences the field of behavior-oriented AI has emerged, which unlike its traditional counter part, is mainly concerned with the study of autonomous agents, situated in and interacting with an environment. Typical criticisms of conventional artificial intelligent systems are that these systems show brittleness for environmental changes, and required much comting time for mapping complex sensory ints into complex internal models before action can be taken. Therefore, in recent years much attention has been focused on the reactive planning systems (e.g., behavior-based Al), which have demonstrated robustness and flexibility against dynamically changing world. On the other hand, biological information processing systems have many interesting functions and are expected to provide various feasible ideas to engineering fields, especially robotics. Biological information processing systems in living organisms can be mainly classified into the following four systems: () brain-nervous system, (2) genetic system, (3) endocrine system, and (4) immune system. Nervous and genetic systems have already been applied to engineering fields by modeling as neural

7 networks, and genetic algorithms [8], and they have been widely used in various fields. Immune system, in particular, have various interesting features such as immunological memory, immunological tolerance, micro-pattern recognition, nonhierarchical distributed structure, and so on that can be applied to many engineering fields. In the following lines we will brief some of the basic features of the immune system. Recognition: The immune system can recognize and classify different patterns and generate selective responses. Recognition is achieved by intercellular binding the extent of this binding is determined by molecular shape and electrostatic charge. Self-non-self discrimination is one of the main tasks of the immune system deals with during the recognition process. Feature Extraction: Antigen Presenting Cells (APCs) interpret the antigenic context and extract its features, by processing and presenting antigenic peptides on its surface. These APC servers as a filter and a lens: a filter that destroy molecular noise, and a lens that focuses the attention of the lymphocyte receptors. Diversity: It uses combinatorics, usually done by a genetic process for generating a diverse set of lymphocyte receptors to ensure that at least some lymphocytes can bind to any known or unknown antigen. Learning: It learns, by experience, the structure of a specific antigen. Changing Lymphocyte concentration is the mechanism for learning and takes place during the primary response of Ag interception. So the learning ability of the immune system lies primarily in the mechanism which generates new immune cells on the basis of the current state of the system (also called clonal selection mechanism). Memory: When lymphocytes are activated, a few of each kind become special memory cells which are content-addressable, and continues to circulate in the blood. The life time of immune memory cells is dynamic and requires stimulation by antigens. The immune system keeps an ideal balance between economy and performance in conserving a minimal but sufficient memory of the past, and this is done normally by using short-term and long-term memory mechanisms.

8 Distributed Detection: The immune system is inherently distributed. The immune cells, in particular lymphocytes, circulate through the blood, lymph, lymphoid organs, and tissue spaces. As lymphocytes recirculate, if they encounter antigenic attacks, they stimulate specific immune responses. Self-regulation: The basic mechanisms of immune responses are self-regulatory in nature. There is no central organ that controls the functions of the immune system. The regulation of immune responses can be either local or systemic, depending on the route and property of the antigenic challenge. Co-stimulation: Activation of B cells are closely regulated through costimulation. The second signal coming from helper T cells helps to ensure tolerance and judge the invader is dangerous, harmless, or false alarm. Dynamic protection: Clonal expansion and somatic hyper-mutation allow generation of high-affinity immune cells which are called affinity maturation. This process dynamically balances exploration versus exploitation in adaptive immunity. Dynamic protection increases the coverage provided by the immune system over time. There are other features like adaptability, specificity, selftolerance, differentiation etc., and they perform important functions in immune response. All these remarkable information-processing properties of the immune system can be utilized several important aspects in the field of comtation. Recent studies have clarified that the immune system does not only detect and eliminate the non-self materials, but plays important roles to maintain its own system against dynamically changing environments. Therefore, immune system would provide a new paradigm that is suitable for dynamic problem dealing with unknown environments rather than static problem. However, the immune system has little been applied to engineering fields in spite of its productive characteristics. In the following sections we will scan some of the applications in the literature on the immune system. Then we will elaborate to our research, and its importance.

9 3.2 Comter Security There are many problems encountered while trying to apply comter security, such activities as detecting unauthorized use of comter facilities, keeping the integrity of data files, and preventing the spread of comter viruses. Forrest et al viewed these protection problems as instances of the more general problem of distinguishing self as legitimate users, corrupted data, etc., and from non-self as unauthorized users, viruses, etc. They introduce a change-detection algorithm that is based on the way that natural immune systems distinguish self from other. Mathematical analysis of the expected behavior of the algorithm allows them to predict the conditions under which it is likely to perform reasonably. Based on this analysis, they also reported preliminary results illustrating the feasibility of the approach on the problem of detecting comter viruses. They demonstrate that the algorithm can be practically applied remains an open problem, and finally, they suggest that the general principles can be readily applied to other comter security problems. Kephart et al anticipated that with in the next few years, the Internet will provide a rich medium for new breeds of comter viruses capable of spreading faster than today s viruses [8]. To counter this threat, They have developed an immune system for comters that senses the presence of a previously unknown virus, and within minutes automatically derives and deploys a prescription for detecting and removing it to other PC's in the network. Their system was integrated with a commercial anti-virus product, IBM Anti- Virus. Their immune system algorithm consists of the following steps: ) Discovering a previously unknown virus on a user s comter. 2) Capturing a sample of the virus and sending it to a central comter. 3) Analyzing the virus automatically to derive a prescription for detecting and 4) removing it from any host object.

20 5) Delivering the prescription to the user s comter, incorporating it into the 6) anti-virus data files, and running the anti-virus product to detect and remove 7) all occurrences of the virus. 8) Disseminating the prescription to other comters in the user s locale and to the 9) rest of the world. Dasgupta et al conducted a research that focuses on investigating immunological principles in designing a multi-agent system for intrusion detection and response in networked comters []. In this approach, the immunity-based agents roam around the machines (nodes or routers), and look for changes such as malfunctions, faults, abnormalities, misuse, deviations, intrusions, etc. These agents can mutually recognize each other's activities and can take appropriate actions according to the underlying security policies. Their activities are coordinated in a hierarchical fashion while sensing, communicating and generating responses. Such an agent can learn and adapt to its environment dynamically and can detect both known and unknown intrusions. Their research is the part of an effort to develop a multi-agent detection system that can simultaneously monitor networked comter's activities at different levels (such as user level, system level, process level and packet level) in order to determine intrusions and anomalies. Their proposed intrusion detection system is designed to be flexible, extendible, and adaptable that can perform real-time monitoring in accordance with the needs and preferences of network administrators. 3.3 Anomaly Detection in Time Series Data Detecting anomalies in time series data is a problem of great practical interest in many manufacturing and signal processing applications. Dasgupta et al presented a novel detection algorithm inspired by the negative-selection mechanism of the immune system, which discriminates between self and non-self []. Self is defined to be normal data patterns and non-self is any deviation

2 exceeding an allowable variation. Experiments with this novelty detection algorithm are reported for two data sets: simulated cutting dynamics in a milling operation and a synthetic signal. The results of the experiments exhibiting the performance of the algorithm in detecting novel patterns were reported. Anomaly detection in a system or a process behavior is very important in many real world applications such as manufacturing, monitoring, signal processing etc. Dasgupta et al presented an anomaly detection algorithm inspired by the negative-selection mechanism of the immune system, which discriminates between self and other. Here self is defined to be normal data patterns and non-self is any deviation exceeding an allowable variation. Experiments with this anomaly detection algorithm are reported for two data sets: time series data, generated using the Mackey-Glass equation and a simulated signal. Compared to existing methods, this method has the advantage of not requiring prior knowledge about all possible failure modes of the monitored system. Results are reported to display the performance of the detection algorithm. Ishida et el proposed a new information processing architecture which is extracted from the immune system. By focusing on informational features of the immune system (i.e. specificity, diversity, tolerance, and memory), an immune algorithm is proposed. The algorithm proceeds in three steps: diversity generation, establishment of self-tolerance, and memorizing non-self. The algorithm may be used to model the system by distributing agents. In this case, the system (the self) as well as the environment (the non-self) are unknown or cannot be modeled. Agent-based architecture based on the local memory hypothesis and networkbased architecture based on the network hypothesis is discussed. Agent-based architecture elaborated with the application to adaptive system where the knowledge about environment is not available. Adaptive noise neutralizer is formalized and simulated for a simple plant.

22 D haeseleer et al presented a new achievements on a distributable changedetection method inspired by the natural immune system. A weakness in the original algorithm was the exponential cost of generating detectors. Two detectorgenerating algorithms are introduced which run in linear time. The algorithms are analyzed, heuristics are given for setting parameters based on the analysis, and the presence of holes in detector space is examined. The analysis provides a basis for assessing the practicality of the algorithms in specific settings, and some of the implications are discussed. 3.4 Fault Diagnosis The body s immune system is impressively good at coping with external and internal errors, usually known as bacteria and viruses. The body is able to distinguish the hemoglobin found in blood from the insulin secreted by the pancreas from the vitreous humor contained in the eye from everything else. It must manage to repel innumerable different kinds of invading organisms and yet not attack the body. Tyrell posed a question which is can we mimic these mechanisms in the design of our comter systems?. He gave some details on how the body actually performs this amazing feat and gives some suggestions as to how this might inspire the design of comter systems increasing their reliability. Braddly et al proposed a novel approach to hardware fault tolerance that takes inspiration from the human immune system as a method of fault detection and removal. The immune system has inspired work within the areas of virus protection and pattern recognition yet its application to hardware fault tolerance is untouched. Their paper introduces many of the ingenious methods provided by the immune system to provide reliable operation and suggests how such concepts can inspire novel methods of providing fault tolerance in the design of state machine hardware systems. Through a process of self/non-self recognition the proposed hardware immune system will learn to differentiate between acceptable and

23 abnormal states and transitions within the immunized system. Potential faults can then be agged and suitable recovery methods are invoked to return the system to a safe state. A production line of semiconductor is a large scale and a complex system. A control system of the line is considered to be difficult to control because there exist lots of malfunctions such as maintenance of equipment, equipment break down disturbance in the production of wafers in the semiconductor production system. Fukuda et al have been exploited some methods and systems using simulations or expert systems approach to solve these disturbances. The semiconductor production systems had been large and complex and the environments of the systems have been changing dynamically, so that it is hard to exploit a perfect control system of semiconductor production by using only conventional methods. Research conducted by Ishiguro et al did focus on chemical and nuclear plant. In these systems, once a certain device (unit) in a plant system becomes faulty, its influence propagates through the whole system, and then causes a fatal situation. To enhance safety and reliability of plant systems, an efficient fault diagnosis technique is desired. On the other hand, biological systems such as human beings can be said to be the ultimate information processing system, and are expected to provide feasible ideas to engineering fields. Among the information processing systems in biological systems, immune systems work as on-line fault diagnosis systems by constructing large-scale networks, called immune networks (idiotypic networks). In this study, the researchers tried to apply these immune networks to fault diagnosis of plant systems, and the feasibility of their proposed method is confirmed by simulations.

24 3.5 Pattern Recognition Forrest et al described an immune system model based on binary strings. The rpose of the model is to study the pattern recognition processes and learning that take place at both individual and species levels in the immune system. Genetic algorithm is a central component of their model. The paper reports simulation experiments on two pattern recognition problems that are relevant to natural immune systems. Finally, it reviews the relation between the model and explicit fitness sharing techniques for genetic algorithms, showing that the immune system model implements a form of implicit fitness sharing. Dasgupta et al described a technique based on immunological principle, for a novel pattern detection method. it is a probabilistic method that uses a negative selection scheme, complement pattern space, to detect any change in the normal behavior of monitored data patterns []. The technique is compared with a positive selection approach, Implemented by an ART neural network, which uses the self pattern apace for anomaly detection. Hunt et al described an artificial immune system (AIS) which is based upon models from the natural immune system. This natural system is an example of an evolutionary learning mechanism which possesses a content addressable memory and the ability to forget little used information. It is also an example of an adaptive non-linear network in which control is decentralized and problem processing is efficient and effective. As such, the immune system has the potential to offer novel problem solving methods. The AIS is an example of a system developed around the current understanding of the immune system. It illustrates how an artificial immune system can capture the basic elements of the immune system and exhibit some of its chief characteristics. They illustrate the potential of the AIS on a simple pattern recognition problem. Then, they apply the AIS to a real world problem: the recognition of promoters in DNA sequences. The results obtained are consistent with other approaches, such as neural networks and are better than the nearest

25 neighbor algorithm. They concluded that the primary advantages of the AIS are that it only requires positive examples, and the patterns it has learnt can be explicitly examined. In addition, because it is self-organizing, it does not require effort to optimize any system parameters. Cooke et al have developed an artificial immune system AIS which is based on the human immune system. The AIS possesses an adaptive learning mechanism which enables antibodies to be used for classification tasks. In their paper, they described how the AIS has been used to evolve antibodies which can classify promoter containing and promoter negative DNA sequences. The DNA sequences used for teaching were 57 nucleotides in length and contained procaryotic promoters. Their system classified previously unseen DNA sequences with an accuracy of approximately 90%. 3.6 Autonomous Agents In recent years much attention has been focused on behavior-based artificial intelligence, (Al) which has already demonstrated its robustness and flexibility against dynamically changing world. Watanabe et al developed an approach in which the followings problems have not yet been tackled: ) How to construct an appropriate arbitration mechanism, and 2) How to prepare appropriate competence modules (behavior primitives). One of the promising approaches to tackle the problems is a biologically inspired approach. The Watanabe group focused on the immune system, since it is dedicated to self-preservation under hostile environment, based on the fact that autonomous mobile robots must cope with dynamically changing environment. They constructed a new decentralized behavior arbitration mechanism inspired by the biological immune system. Then, they applied it to the garbagecollecting problem of autonomous mobile robot that takes into account the concept of self sufficiency. To verify the feasibility of their method, they carried out some experiments using a real robot. In addition, they investigated

26 two types of adaptation mechanisms to construct an appropriate artificial immune network without human intervention. Immunized Comtational Systems combine a priori knowledge with the adapting capabilities of immune systems to provide a powerful alterative to currently available techniques for intelligent control [8]. This was the basic idea that Krishnakumar et al presented on various levels of intelligent control and relate them to similar functioning in human immune systems. A technique for implementing immunized comtational systems as adaptive critics was presented then applied to a flight path generator for level 2, non-linear, full-envelope, intelligent aircraft control problem. Conventional artificial intelligent (Al) systems have been criticized for their brittleness under hostile /dynamic changing environments [6]. Therefore, recently much attention has been focused on the reactive planning systems such as behavior-based AI. However, in the behavior-based Al approaches, how to construct a mechanism that realizes adequate arbitration among competence modules is still an open question. Ishigura et al proposed a new decentralized consensus-making system inspired from the biological immune system. They applied their proposed method to a behavior arbitration of an autonomous mobile robot as a practical example. To verify the feasibility of their method, we carry out some simulations. In addition, they proposed an adaptation mechanism that can be used to construct a suitable immune network for adequate action selection. Lee et al proposed a method of cooperative control (T-cell modeling) and selection of group behavior strategy (B -cell modeling) based on immune system in distributed autonomous robotic system (DARS). The immune system is a living body s self protection and self-maintenance system. Thus these features can be applied to decision making of optimal swarm behavior in dynamically changing environment. For the rpose of applying immune system to DARS, a robot is regarded as a B cell, each environmental condition as an antigen, a behavior strategy as an antibody and control parameter as a T-cell respectively. The executing process of the proposed method is as follows: When the environmental