Expert Operation Assistance System

Abstract

Abstract Cepsa has implemented an Expert Operation Assistance System, designed by and for users, based on a Knowledge-based system that implements production rules. This implementation allows, through production rules defined in a knowledge extraction session, to make multiple comparisons with data taken from Cepsa’s datalake, and to generate a series of recommendations. The objective of these recommendations is to facilitate Operator’s work by warning of situations that may arise as an early warning system. An expert system is a program that imitates human behavior, optimizing the information provided to give an opinion on a specific topic. It has some specific knowledge about an area using information from human knowledge for decision making. Expert systems are the first operational result of artificial intelligence because they manage to solve problems through knowledge and reason in the same way that a human expert does. Cepsa has developed an Expert System platform where business unit specialists can generate their modules implement their rules and get results. This platform allows to scale as many modules as business units need. Many instances of Expert Systems rules have been implemented in some Cepsa units (Crude units, FCC, pumps, furnaces, vacuum unit …) with great acceptance by users and high revenues. Converting human knowledge to the knowledge of AI, expert systems operate as the assistant of expert people increasing supervisory effectiveness. Cepsa has been researching numerous options in the Expert Systems market, but we have chosen to develop our own platform for several reasons: Expert systems have been used mainly for other businesses. It allows giving recommendations on degrees of freedom in those units in which Machine Learning is not able to act due to there are no instrumented manipulated variables (manual valves…) These systems allow to optimize our plants by handling this type of variables