Human-Computer Cooperative Co-Evolutionary Design Method And Its Applications

This dissertation mainly studies the human-computer cooperative layout design method, the co-evolutionary algorithms, the Pro/E-based layout optimization system and their applications to engineering complex layout problems. The engineering application backgrounds included layout design of equipments of a sort of spacecraft (satellite, airship) module and cutter layout design of full face rock tunneling boring machine(TBM). And this study was supported by the National Natural Science Foundation of China, the "863" Hi-Tech Research and Development Program of China and the "11-5" Science Technology Key Project, layout problem belongs to the Packing and Combination optimization problem with behavior constraints in mathematics, the complex layout scheme design in mechanics design, the human-computer cooperative problem in artificial intelligence. Its main difficulties lie in computing complexity, engineering complexity and practical application complexity. Studying this kind of problem is valuable in science and practical application fields.Firstly, a co-evolutionary genetic algorithm with heuristic coordinate strategies is presented by the analysis of the convergence direction of the subpopulations. This algorithm aims at reducing the computing complexity of the complex layout problem, and is based on a co-evolutionary framework, Then borrowing the idea of human-computer cooperation, this study propose a human-computer cooperative co-evolutionary design method to reduce the engineering complexity of the problem, where the human experience can be cooperated with algorithm solutions on the level of algorithm gene and system design. Finally, several key technologies of Pro/E-based layout system for a sort of spacecraft module layout design problem are studied to make the above-mentioned method be practical. Meanwhile, the proposed methods are applied to a cutter layout design of TBM. The main contents of this dissertation are as follows:In first chapter, the engineering background of this dissertation is introduced, then the solution strategies of this dissertation are surveyed, finally the research contents and the organization structure are given.In second chapter, the previous work of the complex layout problems is summarized. Then the research of human-computer cooperation, genetic algorithm, co-evolutionary method and heuristic method are surveyed in detail. The above-mentioned study provided the theory basis of this dissertation.In third chapter, layout design of the satellite module problem and the TBM butter layout design problem are introduced, then this dissertation takes an international satellite (INTELSAT III) module, Europe Meteorological satellite (MSG) module and cutter layout design of Robbis TBM MB264-310 as the engineering layout instances, formulates the mathematical model of the above-mentioned layout problems, and analyzes their characters, finally proposes the solving layout problems of this dissertation.In fourth chapter, a co-evolutionary genetic algorithm with heuristic coordinate strategies is presented, in which two heuristic coordinate strategies are studied. One of heuristic coordinate strategies is based on the global coupled constraints and used to keep each sub-problem at the same searching directions during the whole solving process. Another heuristic coordinate strategy include the extended pattern search strategy and the combination-rotation strategy, which is used to reduce the coupling complexity among the sub-problems and advance the convergence speed of each sub-problem during the latter searching time and determine the original layout scheme. The numerical experimental results showed that the proposed algorithm was superior to cooperative co-evolutionary algorithm and the all-at-once algorithm.In fifth chapter, a human-computer cooperative co-evolutionary design method is proposed. In this method, firstly the artificial reference set is constructed by the artificial schemes designed by "human" during the on-line or off-line design process, then based on the artificial reference set, the diversity reference set is constructed by the diversity layout schemes created by the diversity methods. During the sub-problems co-evolution, the artificial schemes of the artificial reference set, the diversity schemes of the diversity reference set and the algorithm schemes of algorithm population are cooperated to create new cooperative scheme by a unified code. The proposed human-computer cooperative method can combine human intelligence with algorithm on the level of algorithm gene.In sixth chapter, several key technologies of Pro/E-based layout system for spacecraft module layout problem are studied and tested by several instances, the key technologies mainly include as follows: (1) the layout optimization interference computation model on Pro/E platform; (2) layout optimization mass property computation model on Pro/E platform. The layout system has been tested and accepted by the user.In seventh chapter, the satellite module layout instance and the TBM cutter layout instance are solved by the above-mentioned methods to demonstrate their affectivity and feasibility. The experimental results showed that the proposed human-computer cooperative method could solve the complex engineering layout problem effectively. Another had to be point out that the above-mentioned experiments only include the finite layout tests and satellite module layout applications and it can not draw a general conclusion.In eighth chapter, the main research content of this dissertaion is summarized, and then several future work are given. Another hope this study can be helpful to the research of "Man-machine synergy", and also helpful to the spacecraft module layout practical application and the full face rock TBM cutter layout practical application.