Research On Multi-Agent System And Its Application Based On Cooperation Mechanism And Intelligent Algorithms

Assignment scheduling problems, which means searching for the best one from large number of feasible solutions, commonly exists in many fields, such as economic planning, engineering design, manufacturing, data storage, information security, etc. To solve these problems effectively will not only have very significant meaning for entire society, but also bring great economic benefits. Task assignment, resource assignment, and production scheduling problems are classical problems in computer science. Some of their sub-problems are still unresolved. Research on these problems is very valuable inboth theory and application. With the development of advanced manufacturing technology, the assignment scheduling problems are becoming more and more complex, which makes the traditional assignment scheduling methods being unable to cope with these complex problems. Therefore, it is an urgent need to find some new assignment scheduling methods.In recent years, Agent and Multi-Agent System (MAS) are growing up to be one of the most important techniques in the practical research on Artificial Intelligence and intelligent software in distributed computing environment. Agent's problem solving process is very close to the way of thinking human being do. Unlike traditional algorithm designing which has to analyze the problem comprehensively, multi-agent computing only needs to assign agents their targets and then keeps free while these agents will automatically achieve client's targets by their active interaction. Therefore, an agent with automatic computing properties such as pervasive, network-oriented, intelljgent, agent-based, and humanized computation can act as a human. Building MAS for large and distributed problems makes computer system more intelligent and further liberates people from their work. And it is convinced that MAS can prosper computer technology.Practivetly, ability to interact aiming at cooperation mechanism is the key issue that distinguishes MAS from other computing modes. So designing patterns for MAS using cooperation mechanisms is very popular. Most of the mechanisms are form economic and society. Intelligent computation originated from observation and simulation on some specific phenomena of nature and biological world. Compared with traditional methods, these new bionic intelligent optimization methods represented by intelligent computation are widely used in scientific research and industrial production because of their higher adaptability, robustness and parallel processing capability for various kinds of complex assignment scheduling problems. This dissertation studied task assignment, resource assignment, and production scheduling problems based on cooperation mechanism and intelligent algorithm in MAS. Some novel models and algorithms are presented for solving these problems and the performance of these algorithms is validated by experiments. The results demonstrate that the proposed algorithms can improve the effectiveness, reduce the resource cost, and overcome premature convergence effectively, while the solution quality is also improved. The main contents of this dissertation are as follows:1. Taking task assignment problem for intrusion detection system as research objective, a multi-agent intrusion detection system (ICNPIDS) is proposed based on immune and contract net protocol. In order to realize the load balance between the detection agents, an assigning strategy is used based on the resource performance value for normal intrusion detection task. Applying the cooperation method of contract net protocol into intrusion detection, the united immune antibodies is also proposed for the task of detecting too difficult intrusion, by analogy with the passive immune antibodies, the memory automatic immune antibodies, and fuzzy automatic immune antibodies. The generation, detection, and updating of the united immune antibodies are also examined, which overcomes the problem what is too difficult to share the analytic experience. The detecting efficiency of intrusion detection model that utilizes the abusively detection and the anomaly detection are improved greatly. Experimental results show that ICNPIDS is a more adaptive and efficient system.2. A mixed strategy for mobile agents network storage scheduling was proposed, which was combined by genetic algorithom and simulated annealing algorithm. The proposed scheduling strategy takes the creation sequence, priority and storage performance value of mobile agents into consider, for reasonablely dispatching multiple mobile agents and ensuring the measurement performance. Moreover, simulated annealing (SA) strategy is introduced to execute refined search of the genetic algorithms (GA) in local area maintains the diversity of the population and avoid premature convergence. Simulation results demonstrate that the proposed algorithm can effectively arrange the order of mobile agents dispatching, where the time delay can be remarkablely decreased and the storage performance can be accordingly prolonged.3. Job shop scheduling problem (JSSP) is a typical production scheduling problem: a number of jobs will be processing on different machines, and each machine is not allowed to process the jobs in the same order. A job shop scheduling model combining MAS with genetic algorithms-tabu search (GATS) is presented. The proposed model is based on the generalized partial global planning mechanism and utilizes the advantages of static GATS with dynamic MAS. A scheduling process from"initialized macro-scheduling"to"repeated micro-scheduling"is desiged for large-scale complex problems to enable to implement an effective and widely applicable prototype system for JSSP. Under this scheme, a set of theoretic strategies in the GPGP are summarized in detail. A multi-objective optimization scheduling is performed and the GPGP is simulated by using simulation software DECAF for the JSSP. Moreover, the memory function of TS is introduced into procedure of genetic GA. A new recombination operator is constructed, while mutation operator is replaced by TS. Therefore, GATS is to improve the particle's searching ability and increase the convergence speed. Meanwhile, those simulation results are compared with CNP and NONE. The results show that the proposed model based on the GPGP not only improves the effectiveness, but also reduces the resource cost.4. For semiconductor fabrication returnable dynamic scheduling problem, a line dynamic scheduling model combining MAS with immune evolutionary strategies algorithms (IESA) is presented in this dissertation. The proposed model is based on the improved GPGP and utilizes the advantages of IESA with dynamic MAS. A scheduling process from"macro-scheduling to micro-scheduling to repeated- scheduling"is designed for large-scale complex problems to enable to implement an effective and widely applicable prototype system for scheduling model. Under this scheme, a set of limitation and improvement of GPGP about its structure and practice are proposed. Furthermore, an IESA is introduced to enhance the local search ability of particle and increase the convergence speed. The improved GPGP and its model are simulated by using simulation software eM-plant. A case study is provided to examine the practicability, flexibility and robustness of the proposed scheduling approach.In recent years, research on cooperation mechanism and intelligent algorithm based MAS as well as its applications has been paid more attention by many scholars at home and abroad, and a lot of improved versions of algorithms and new applications are blooming correspondently. This dissertation mainly focuses on researching on cooperation mechanism and intelligent algorithm based MAS for task assignment, resource assignment and production scheduling problems. Some more effective model and methods are proposed, which has some theoretical significance and application value in the improvement of MAS, cooperation mechanism and intelligence algorithms to more application fields.