Research On The Co-Evolutionary Design Method Of Products Based On Swarm Intelligence

Along with the increasingly fierce market competition, how to improve product quality, shorten the product development cycle, and promote the degree of innovation of products are the effective means to increase the competitiveness of products. Since the function and structure of products become more and more complex, it is a very important theoretical and practical significance to improve the efficiency and quality of the traditional design method by modern information technology and propose the methodology of innovative design of products.Using the relevant theoretical research achievements for reference, according to the innovative design of mechanisms, the optimal design of discrete and continuum structures, and the optimal design of tolerances, the Co-Evolutionary Design (CED) method of products is deeply researched in this dissertation by integrating the method of Swarm Intelligence (SI) and the principle of biological co-evolution, which provides the innovative design theory of products with a new research approach.Through the analysis of the existing design model, the three-dimensional description of the innovative design process of products is made up and the principle of the co-evolution among problem space, operator space and solution space is revealed. According to the needs of structural schema design of products, the principle of co-evolution is instantiated to three sub patterns, e.g., the co-evolution between problem space and solution space, the co-evolution between operator space and solution space and the co-evolution between problem space and operator space. Based on the theoretic analysis, this dissertation proposes the framework of the CED method of structure schema of products based on swarm intelligence using the principle of co-evolution as the guidelines and the method of swarm intelligence as the key problem-solving algorithms.Under the sub pattern of the co-evolution of operator space and solution space, for the problems of the innovative design of mechanisms, a method of isomorphism identification of kinematic chains is proposed firstly by adopting the Evolutionary Ant Colony System (EACS) to find the Maximum Structural Code (MSC), the method of digital description of mechanism path is proposed then, and the method of path clustering of mechanisms based on Ant Colony Clustering (ACC) algorithm is presented and the algorithm of path matching is designed finally. The effectiveness of the above methods is invalidated by the isomorphism identification of planar 10-bar mechanisms and the path synthesis of planar 4-bar mechanisms respectively. For the problems of the optimal design of tolerance, the multi-objectives model of the optimal design of technique tolerance is built firstly using the machining cost and quality lost as the design objectives, a Multi-Objective Particle Swarm Optimization (MOPSO) algorithm that can directly find the Pareto solution set is proposed then based on the principle of Pareto optimality, and the modeling and transformation of the optimum design of the parallel tolerances is researched and a Hybrid Swarm Intelligence algorithm for them is proposed. Two engineering examples of the optimal design of tolerance are proposed finally.Under the sub pattern of the co-evolution between problem space and solution space, the co-evolutionary design method of discrete structures is proposed. Firstly, the process of co-evolutionary design of discrete structures is divided to two sub processes with mutual influence, viz., the process in topology space (problem space) and the process in size pace (problem space). Then the models of size optimization and topology optimization are built and the method transforming these models to the standard combinational optimization problems is proposed. The Tailored Ant Colony Optimization for the Sizing of Structures (TACO-SS) and the Tailored Ant Colony Optimization for the Topology of Structures (TACO-TS) are implemented respectively and the parameter setting and the convergence process are analyzed and discussed deeply. The design results of some standard testing examples and the structure of the truss antenna shows that the CED method has the characteristics of high efficiency, global convergence, and robustness and is fit for the solution of large-scale problems. Finally, the method above is applied to the engineering example of the optimal design of the supporting structure of the paraboloid antenna.Under the sub pattern of co-evolution between problem space and operator space, according to the mapping relationship between Cellular Automata (CA) and finite element, the mechanism how the optimal topology emerges based on the local interaction among the discrete elements of continuum structures under the conditions of loads and constraints is studied and a method of topology optimization design of continuum structures based on CA is proposed. According to the topology optimization design of the standard two-dimensional cantilever, the combination of controlling parameters of local rules and the influence of the structure of neighborhood on the process of evolution are discussed. Theoretical analysis and the computational experiments show that the time complexity and the space complexity of this method are very small, it converges very fast, it is fit for the solution of large-scale problems and there is no the phenomena of numerical instability, e.g., the checker boarding and the mesh dependency. Finally, this method is applied to the topology optimization design of the square heat-dispersion panel and the topology optimization design of the airfoil.Based on the theoretical research results above, all the algorithms are programmed, a prototype system for the co-evolutionary design of products based on swarm intelligence developed and its feasibility and effectiveness is also validated.