科研之友
微信
新浪微博
Facebook
分享链接
摘要
The conventional cost-tolerance model is constructed by linear or non-linear regression analysis based upon empirical data from all frequently used production processes. These approaches suffer relatively large model-fitting errors and also fail to consider the varying manufacturing environment due to the simplicity of the mathematical models used. In the present study, a novel, multi-parameter cost-tolerance model is developed based on a fuzzy neural network (FNN) that has tolerance and a cost influence coefficient as inputs and manufacturing cost as output. First, the cost influence coefficient is introduced based on analysis of the manufacturing environment. Then, a multi-input, FNN-based cost-tolerance model is constructed. Analysis of the model is conducted and comparison is made with existing models. Analytical results show that the proposed model yields better performance in controlling the average fitting error and is flexible to varying manufacturing environment. It can provide more reliable results and reduce the chances of error in the tolerance design.
