Tolerance Optimization Design Based on Neural Network and Genetic Algorithm
Download as PDF
DOI: 10.23977/meimie.2019.43048
Author(s)
Jinwei Fan, Ning Ma, Peitong Wang, Jian Yin, Hongliang Zhang and Miaomiao Wang
Corresponding Author
Ning Ma
ABSTRACT
Aiming at the characteristics of highly non-linear relationship between tolerance and cost in product manufacturing, a tolerance optimization method based on neural network and genetic algorithm is proposed. This method uses genetic algorithm to obtain global optimal solution with strong robustness by probability search strategy in a wide range of solution space, and the advantages of neural network in solving highly non-linear problems. The function relationship of tolerance cost with black box characteristics is obtained by simulating tolerance cost with neural network. Then genetic algorithm is used in tolerance allocation to minimize total cost, and optimization is carried out under the constraints of meeting assembly tolerance requirements and meeting standard tolerance grade. At the same time, tolerance optimization system is developed based on VC and Matlab, and the object of tolerance allocation is the locker mechanism of aircraft cabin door. The results show that the results of comprehensive allocation using neural network and genetic algorithm are superior to those of traditional methods.
KEYWORDS
Tolerance cost relationship, tolerance optimization, neural network, genetic algorithm