Doctor of Philosophy
Department of Mechanical Engineering
Shivathaya, Seetaram Sahadev, Material design in steel making utilising mathematical modelling, knowledge-based and fuzzy logic approaches, Doctor of Philosophy thesis, Department of Mechanical Engineering, University of Wollongong, 1997. http://ro.uow.edu.au/theses/1586
Development of knowledge-based system for material design is a complex task, due to the interrelationship of many factors in steel making process. In addition to this, design specifications vary frequently and material design knowledge is held in largely intuitive undefined format. This thesis discusses a material design system which deals with the determination of the steelmaking aim chemistry. If an attempt is made to design aim chemistry only based on a mathematical approach of utilising the empirical models between various design parameters, it would result in unrealistic design because relationships between various design parameters are not always linear. Therefore it is inevitable to apply knowledge-based methods along with the mathematical approach to deal with this complex task. The approach put forward in this thesis is a hybrid approach, where the knowledge-base is applied at every stage of the design process to utilise the expert as well as the heuristic knowledge of metallurgists to obtain the designs which are realistic and which take into account various limitations and constraints encountered in steel making. The material design is also characterised by extensive utilisation of the grade history database which contains performance data for various steel grades and thickness combinations. The inputs to the system are through interactive dialogue sessions and the inputs consist of the material standards, size, quantity, tonnage, end use and the customer special requirements. These inputs along with the numerous rules in the knowledge-bases as well as the mathematical modelling enable the effective design of the steelmaking aim chemistry.
Knowledge Elicitation (KEL) is the most important stage, but it is often the principal bottleneck in the development of knowledge-based systems. Due to the difficulties faced in the knowledge elicitation process, development of a knowledge-based system for material design in steel making industry is a complex task. An attempt is made in this thesis to present a novel approach to deal with knowledge elicitation for material design problems in steel making industry. This research centres around the human aspects and is based on practical expelience gained while developing a knowledge-based system for material design at BHP Steel, Australia. This approach involves codification of the customer special requirements to identify the knowledge sources involved in the design process. This is followed by the use of paper models to improve the efficiency of KEL process. The second stage of the structured interviews is based on the customer special requirement codes for eliciting the missing information and for clarifying any ambiguities or inconsistencies. This research also focuses on the use of non-interviewing techniques to elicit the expert knowledge in order to reduce the expensive interview time. The knowledge representation scheme developed for the material design system aims at reducing the search time and storage space by utilising the codification scheme to classify various knowledge sources into appropriate categories.
The thesis then presents the application of fuzzy logic to the material design system to rank the alternative steel making aim chemistries according to the degree which will satisfy the customer's requirements of chemistry and mechanical properties, with due consideration given to the economic aspects and the complexity involved in the production. Statistical data regarding the performance of the grades produced in the past are also utilised in this process. The development of the membership functions for the material design fuzzy logic based system, which is an important task, is presented in this thesis. Four factors considered in the development of the membership functions include chemistry, mechanical properties, relative cost and the complexity involved in the production of the steel material.
Finally the thesis presents the development of an interactive graphical user interface for a material design knowledge-based system based on a three character alphanumeric codification scheme for customer special requirements. This user interface makes the material design system more user friendly and enables error free and fast input of the basic information and the customer special requirements, corresponding to any customer order for steel plates. This is achieved through two interactive dialogue sessions utilising two input screens. In addition to making the system more user friendly and visually appealing, the interface developed also adds flexibility and sophistication to the prototype knowledge-based system for designing steel plates.