Modeling impact of product variety on performance in mixed-model assembly system: An artificial neural network meta-modeling approach

The increasing variety of products complicates the mixed-model assembly process and affected the assembly system in terms of product quality and productivity. In the paper, variety induced manufacturing complexity with regard to choices that operators have to make for various assembly operations is measured with information entropy of the average randomness in choice processes. The impact of error rate associated with the complexity on the performance of the system is analyzed by means of the investigation on average reaction time and speed-accuracy trade-off. In addition, an established artificial neural network meta-model contribute to modeling the impact of product variety on the system performance. The artificial neural network meta-model has superior performance than a multiple linear regression meta-model in terms of experiment results and appears to be the optimal approach to modeling impact of product variety on performance in mixed-model assembly system.