REFRIGERATED STORAGE - REAL-TIME CONTROL USING INTELLIGENT PARAMETER PASSING

The conservation of food products without the use of chemicals is an a rea arousing great public interest in the present climate. Conventiona l control of large refrigerated stores can be improved dramatically by the use of artificial intelligence techniques, and particularly exper t systems. One of the requirements of the system developed is for a si mulation of the heat and mass transfer of the store, and this has a hi gh processing time overhead. This paper defines the real-time control problems of the situation, both synchronous and asynchronous, and give s a general overview of methods adopted for dealing with them. The sep aration of the knowledge bases within the whole system has been used t o facilitate the assessment of particular real-time bottlenecks, and t he improvement in performance within these areas by intelligent parame ter passing. Discussion is confined to the two main knowledge base com ponents where this application is critical, the simulation model and t he intelligent data-archiving system. Such an approach promotes the ma chine-learning capabilities of the application in both areas.