A fuzzy analytic hierarchy process approach for multi-objective molecular design problem

• Main Authors: Ooi, Jecksin, Promentilla, Michael Angelo B., Tan, Raymond Girard R., Ng, Denny K.S., Chemmangattuvalappil, Nishanth G.
• Format: text
• Published: Animo Repository 2017
• Subjects: Molecules—Models—Computer-aided design; Chemical Engineering
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/2902
• Institution: De La Salle University

This paper presents a novel Computer Aided Molecular Design (CAMD) algorithm for managing the subjectivity issues in multi-objective molecular design problem. In most of the CAMD problems, physicochemical properties have often been selected as target properties. However, molecules with good functionalities do not imply that they are safe to the environment and consumers. Therefore, a trade-off should be performed between their performance and safety, health and environmental properties. One of the major challenges in such multi-objective CAMD problems is the subjectivity involved in allocating the weighting factors to each property. It is particularly important to address this issue as different weighting factor of each property will lead to the generation of different optimal molecules. Moreover, performance targets which belong to different categories cannot be compared on a common scale. In this work, a systematic methodology is developed to design molecules which can satisfy multiple target properties including physicochemical, safety, health and environmental properties through Fuzzy Analytic Hierarchy Process (FAHP) approach. FAHP is an extension of classical Analytic Hierarchy Process (AHP) approach where it uses triangular fuzzy numbers instead of exact numerical values to express the pairwise comparison of decision elements. Via FAHP, the inability of classical AHP approach to deal with the uncertainties arising from ambiguity involved during value judgement elicitation is addressed. A case study on solvent design for oil extraction from palm pressed fibre is solved to illustrate the proposed methodology. © 2017 Elsevier B.V.