| The objective of this research was to investigate the application of expert system technology to chemical process design and operation. This thesis includes two case studies. The first study illustrates the technique of machine learning in an expert system. The expert system provides recommendations for selection of vapor-liquid equilibrium (VLE) calculation methods. VLE calculations are used in the design and simulation of process equipment. The expert system developed here used machine learning to acquire knowledge. Example VLE data points were presented to the system. The system used several common calculation methods to predict the data points. The errors in prediction, along with a classification system based on chemical families, formed the criteria for VLE method selection.; The second expert system was developed to provide advice for the startup of a section of a refinery. It was developed in conjunction with an industrial partner. This expert system contained the startup knowledge from an industrial process operator with many decades of experience. (Abstract shortened by UMI.)... |
