INTELLIGENT SYSTEM FOR SAFE PROCESS STARTUP

An intelligent system has been developed to assist the process startup in a section of a refinery. The system was developed in conjunction with an industrial partner. This intelligent system codifies the knowledge from industrial process operators who have many years of experience. The project aims at investigating automation of the process startup operation by using an integrated distributed intelligent system concept, as well as capturing important expertise and knowledge from experienced operators. Testing and utilizing the intelligent system within the plant indicates that the system can provide operational advice at the level of human experts. Another element in its success is the automation of ''look-up'' functions in the written manual. Hypertext is used to provide rapid and efficient access to the manual. The hypertext system is integrated with the knowledge base to provide an intelligent selection of appropriate manual sections. An important result of this research is the development of a new conceptual process model, which represents the chemical and mechanical states of the unit for startup. The model is characterized by the mechanical states, modified by the operators, to manipulate the chemical states. Then, the chemical states are driven towards the goal states: the normal operation of the unit. This methodology provides a new approach to improving the industrial manufacturing environment using AI technology.