Research on solving multi scheme design problems w

  • Detail

A study on the problems of solving multiple scheme design by the use of genetic algorithm Ouyang MiaoAn (South China University of science and technology, postal workstation of Foshan enterprises) Abstract: This paper put forward a genetic algorithm for solving the problems of multiple scheme design in the design of intelligence. It discussed the key technique, including a kind of an improved genetic algorithm, for solving the problems of multi-scheme in genetic algorcthm and thus avoided a too early convergence at a partial optimum solution happened in the standard genetic algorithm and also accelerated the convergent speed in an overall optimization. An example of making solutions for multiple schemes in the modular synthesis of machine tools is presented. The genetic algorithm can set problems effectively on the selection of modules and the "combined explos analysis reason: ion" in issues of combination

Key words: Intelligent design, Genetic algorithm, Multiple schemede design, Modular synthesis of machine tool.

fig 2 tab 0 ref 3 "Jixie sheji" 81721 introduction

intelligent design is a computer application technology in the stage of knowledge intensive industry. In the field of design automation, it aims to pursue the highest efficiency to obtain the optimal solution of engineering system or technical problems. The intelligent design of engineering problems inevitably faces the problem of multiple schemes. The solution of multiple schemes is a common problem in engineering problems, and it is also one of the key problems of intelligent design. Without multiple schemes, there will be no optimization of the scheme, and without the optimal scheme, there will be no real significance of intelligent design. However, so far, most intelligent design topics focus on knowledge acquisition, knowledge representation, knowledge processing, human-machine interface and so on. At the same time, the previous intelligent system basically adopts a principle synthesis method when solving the design of multiple schemes, such as morphological matrix method and bond graph method. These methods generally enumerate and combine the optional schemes from the perspective of functional analysis, but the number of schemes of these combination methods is too many, and it is difficult to evaluate and optimize. In view of the above reasons, It is necessary to explore a new processing mechanism to solve the problem of multi scheme design in intelligent design

genetic algorithm (GA) is a branch of Computational Intelligence (CI). Like artificial neural networks, genetic algorithm masters nature and gets inspiration and Enlightenment from nature's masterpiece - "biological evolution". The evolution process of nature always objectively follows the law of "things compete with nature, and the fittest survive". In 1961, Bledsoe had proposed to use some concepts in biology for systematic analysis and research. In the late 1960s, John Holland of the University of Michigan proposed the concept of genetic algorithm. Since then, the law of "natural selection" has been developed in engineering applications. Adaptation in natural and artificial system, written by John Holland in 1975, lays the theoretical foundation for the guidance and analysis of the comprehensive application of biology, cybernetics and artificial intelligence. The research of genetic algorithm is of great significance to the development of Computational Intelligence in artificial intelligence, which lays a foundation for the new discipline of computational intelligence

genetic algorithm has a wide range of applications. It has been used in neural network structure optimization design, neural network weight training, alternative to traditional optimization design, fuzzy logic controller matrix parameter optimization design, decision support system, salesman path optimization, VLSI Digital Circuit Design. In the field of manufacturing, it is used in a wide range of manufacturing intelligent systems, such as optimal design of mechanism parts and cutting tools, production scheduling, identification of nonlinear systems, control and measurement systems, and so on. This paper attempts to use genetic algorithm in intelligent design to solve multi solution problems. 2 algorithm description of multi scheme solution

2.1 mathematical description of genetic algorithm

suppose there is a problem to be optimized:

f=f (x, y, z), f ∈ R, (x, y, z) ∈ C (1), where: X, y, Z - independent variables, which can be numerical quantities, logical quantities, or even symbolic quantities. Each set of Xi, Yi, Zi ∈ C constitutes a solution of the problem

c -- it can be regarded as the definition domain of independent variables X, y, Z, as well as the constraint conditions of TPV widely used in the fields of automobile, construction, transportation, electronics and electrical appliances, or the solution space composed of all solutions that meet the constraint conditions

f -- a real number belonging to the real number field R, which can also be regarded as a measure of the quality of a group of solutions (Xi, Yi, Zi) ∈ C

f -- represents a mapping from the solution space C to the number field r of shishanying paper ( in the first quarter of 2019. The only requirement for it is that it must have a definition, that is, for a certain solution (Xi, Yi, Zi) ∈ C, it can calculate a certain fi corresponding to it

the optimization goal is to find (x0, Y0, Z0) ∈ C, so that f=f (x0, Y0, Z0) → max. when solving multi scheme problems, we should find a series of solutions, get the optimal solution, sub optimal solution, etc

2.2 solution strategy of genetic algorithm for solving multi scheme problems

genetic algorithm adopts the global search technology of population. The generation of initial population means the generation of multiple initial schemes. The performance of these initial schemes is generally not good. Their quality is measured by fitness calculation, and the schemes are sorted according to the size of fitness. The solution of genetic algorithm is realized by three main operators, namely seed selection, hybridization and mutation. The purpose of the operator is to produce a new generation of groups, that is, to produce better schemes. This operator will stop for a period of time after continuous production of prepreg semi-finished products until one or more schemes are satisfactory to users. The solution strategy of genetic algorithm is shown in Figure 1

Copyright © 2011 JIN SHI