An expert system approach to discrete-change simulation

Abstract

There are a number of major prerequisites in the planning, design and operation of all manufacturing systems. These include resource planning, production flow analysis and reliability assessment. The application of these methods form an integral part of an overall feasibility study, the result of which leads to a desired system configuration. Analytical techniques have been employed as corrective tools to amend proposed configurations in an iterative manner. Their role is to reduce the host of initial options that are proposed, thereby arriving at the optimum. There are two main shortcomings with these traditional analytical procedures. Firstly, the complex algorithms formulated are inflexible and dedicated to particular measures of performance. Secondly, they are generally used independently and this gives untold problems for integration within the iterative procedure. These problems are further exacerbated by the diversity of interdependent features associated with flexible batch manufacturing environment, and by external perturbations that can break‐down their steady state characteristics. There is a major need for a robust tool to be established that forms a coherent manufacturing control strategy and will minimise the effect of disturbances on the performance of manufacturing systems. The achievement of this in real‐time by an intelligent‐based package is a tall order. The formation of an expert system off‐line, however, is quite feasible. Such a system using discrete‐change simulation is formulated. The method encompasses production flow control as well as reliability prediction analysis. The established methodology is then applied to a real industrial system to configure the most appropriate operational philosophy over any desired production period.