摘要: 以优化过程中能量减少最多的方向为系统优化方向,以系统的能量最小而熵最大的平衡态为设计的最终目标。对机械优化中需要圆整的变量事先加以区分,对变量附加两种不同的权值以提高设计的精度。整个混合系统的优化设计以遗传算法为中心并结合模拟退火原理、神经网络技术及有限元分析方法的优点,给出智能混合系统的模型框架及工作原理。通过对Benchmark函数的优化测试表明:该智能混合优化系统具有自适性能力强、优化设计精度高等优点。
关键词:
混合系统,
神经网络,
模拟退火算法,
遗传算法,
自适应参数调整
Abstract: Using the direction of the energy being diminished quickly acts as the systematic optimization direction and searching for the equilibrium state with the minimal energy and the maximal entropy acts as the ultimate design target. Some variables which should be the integer variable, in advance, are differentiated and adopt two kinds of different weights for the variable to improve the design accuracy. The genetic algorithm acts as the core of design combining with simulated annealing principle, neural network technology and finite element analysis method, and the model frame and principle of intelligent hybrid optimization system are given. By means of the function of “Benchmark” to test the hybrid system, results indicate that the intelligent hybrid system has some advantages such as a higher accurate design and a stronger adaptability.
Key words:
hybrid system,
neural network,
simulated annealing algorithm,
genetic algorithm,
self-adaptive parameter adjustment
中图分类号:
刘道华;原思聪;李艳灵;江祥奎. 面向过程的智能混合系统优化设计[J]. 计算机工程, 2008, 34(12): 40-42.
LIU Dao-hua; YUAN Si-cong; LI Yan-ling; JIANG Xiang-kui. Optimization Design of Process-oriented Intelligent Hybrid System[J]. Computer Engineering, 2008, 34(12): 40-42.