The development of an intelligent inventory management system
This thesis is concerned with the development of an intelligent inventory management system. The aim of the system is to bridge the substantial gap between the theory and the practice of inventory management and to help industrial inventory managers to achieve an effective and successful inventory management. The proposed system attempts to achieve this by providing automatic pattern identification and model selection facilities. Such a hybrid knowledge-based inventory system consists of a collection of techniques (or pattern identifier) for identifying demand and lead time patterns and a knowledge base (or rule base) for subsequent selection of a suitable inventory model taking into consideration aspects of the practical situation. There are no previous attempts in the inventory literature to develop such a system to guide model selection. In order to integrate the system into the established computer-based intelligent inventory management system and facilitate the function of the pattern identifier, a data manager has been developed to manipulate the history data required for statistical analysis and to load the data into the system from other applications. In order to establish the system's model base, the study of the modelling of inventory and the features and evolution of expert systems are reviewed. The published models which deal with similar inventory problems have been compared based on its applicability, clarity, and being suitable to be computerised. It was necessary to further develop and amend published models to fill gaps in the model base. The overall structure and salient features of the proposed system and the development of the system using Visual Basic have been described. The system has been tested using real life data supplied by the co-operating companies. Finally, achievements and shortcomings of the system are discussed and some suggestions for further research are outlined.