| The natural immune system is a very comlpex system with several organs, many tissues and billions of cells and molcules. The main role of the immune system is to recognize and eliminate foreign antigens (e.g., bacteria, virus, etc) and act as a defensive barrier. The immune system is regarded as a second brain of human and other mammals, and is a subject of great research interest because of its powerful information process capabilities, such as pattern recognition, learning, memory, adaptability, specificity, diversity, etc.The theory of clonal selection is commonly accepted by most immunologists. In this dissertation, the models of artificial immune reponse based on clonal selection theory are studied first. Then we propose a new algorithm, Artificial Immune Algorithm (AIA), to solve the question of optimization. The convergency of AIA is also analysised in mathematical style. Simulation result demonstrated the validity of the algorithm. The application of the AIA to industrial and other areas are presented in this dissertation.In Chapter One, the history of immunology is reviewed and the principle of clonal selection is presented, then we provide the advances in the research area of Artificial Immune Systems (AIS), and finally list the content of this dissertation. In Chapter Two, we propose a simplified model of artificial immune response, which have ability of learning and memorizing. In Chapter Three, we propose AIA based on clonal selection theory, and investigate the influence of coding to its performance. Markov chain analysis is applicated to AIA in Chapter Four. The result demonstrates that AIA can converge to its global optimal point whatever the initial distributing of solutions is. In this chapter, we also propose new means to evaluate the speed of convergence. In Chapter Five, several kinds of test functions are provided to test the performance of AIA. It performs very well in most of the tests. Quadrature experiments are used to investigate the influence of parameters on its convergence speed. In Chapter Six, AIA is applied to solve the problem of Optimal Power Flow (OPF) of electrical systems. Tests on IEEE-30 Bus System shows that AIA performs better than traditional Primary-Dual Inner Point Method (PDIPM) and Elitist Selection Genetic Algorithm. In Chapter Seven, a hybrid architecture that integrates deliberated planning and behavior-based controlling is proposed for autonomous mobile robots. Simulation result demonstrated the feasibility of the system and AIA.The main contributions of this dissertation are summarized as follows:(1) By analyzing the principle of clonal selection in immunology, a simplified model of artificial immune response is proposed, the model has powerful informationprocess capabilities, such as pattern recognition, learning, memory, etc.(2) A new optimization algorithm, Artificial Immune Algorithm (AIA) is proposed in this dissertation. Compared to the previous immune optimization algorithms, the algorithm in this dissertation can simulate the natural immune system more generally and truly. And AIA has good performance in simulation test on De Jong's test functions.(3) Stochastic theory and other correlative theories are used to analysis AIA, the global convergence of AIA is proved. Meanwhile, A new method to evaluate the speed of convergence is proposed.(4) Quadrature experiments are used to investigate the influence of parameters on its convergence speed. Several guidances on parameters of AIA are provided in this dissertation.(5) Optimal Power Flow in electrical engineer is a complex nonlinear planning, artificial immune algorithm is proposed tackle this question. Tests on IEEE-30 Bus System shows that AIA performs better than other algorithm.(6) A new hybrid architecture that integrates deliberated planning and behavior-based controlling is proposed for autonomous mobile robots. A new path planning method is proposed based on AIA. Simulation result demonstrated the feasibility of the system and AIA.
|
PDF Full Text Request