A self-organized neural network metamodelling and clonal selection optimization-based approach for the design of a manufacturing system

Simulation metamodelling has captured the attention of various researchers due to its ability to estimate the simulation responses corresponding to a given combination of input variables, and to provide broad insights than simulation models alone. In this paper, the metamodelling approach has been extended by proposing a new methodology that incorporates the aid of two soft computing tools: neurofuzzy network and polynomial neural network. To perform post metamodelling analysis, an artificial immune system (AIS) based optimization algorithm has been used. Unlike in most neural-networks, the number of layers and number of nodes in each layer of the proposed hybrid network are not predetermined but can be generated dynamically through a growth process. AIS based optimization algorithm has been utilized to exploit the solution space much more extensively to generate an optimal configuration. The special features of the proposed methodology are the rate of convergence, computational speed, and quality of solution. To validate the procedures developed here, the authors have considered a well-known case study of a PCB manufacturing plant. The obtained results show that the proposed methodology can be applied to solve complex manufacturing system design problems in a more effective and efficient manner.