Product concept development and evaluation based on Quality Function Deployment, Axiomatic Design and Reinforcement Learning
| dc.contributor.author | Fazeli, Hamid Reza | |
| dc.contributor.examiningcommittee | Chen, Ying (Biosystems Engineering) | |
| dc.contributor.examiningcommittee | Luo, Yunhua (Mechanical Engineering) | |
| dc.contributor.examiningcommittee | Kwok, Tsz Ho (Mechanical, Industrial and Aerospace Engineering, Concordia University) | |
| dc.contributor.supervisor | Peng, Qingjin | |
| dc.date.accessioned | 2023-08-14T18:03:41Z | |
| dc.date.available | 2023-08-14T18:03:41Z | |
| dc.date.issued | 2023-08-05 | |
| dc.date.submitted | 2023-08-05T20:34:37Z | en_US |
| dc.degree.discipline | Mechanical Engineering | en_US |
| dc.degree.level | Doctor of Philosophy (Ph.D.) | |
| dc.description.abstract | Product concept design plays an important role in searching solutions of the product to meet design requirements based on design constraints and technical measures. Designers have to explore thousands of alternatives to form an optimal product design concept, which is challenging to explore and evaluate all of the possible design concepts efficiently. This thesis proposes new methods in five different research tasks to improve the product concept generation and evaluation. These research tasks address existing challenges to satisfy customer requirements, improve efficiency in domain mapping, form a product concept, evaluate generated design concepts, and automate the design process. In research tasks one and two, novel algorithms are proposed to solve these limitations. A combined method of Decision-making Trial and Evaluation Laboratory and Analytical Network Process approach is proposed in research task one to model interactions in the correlation matrix of House of Quality. Research task 2 builds the relationship matrix of House of Quality using an optimization model based on Best-Worst Method and Full Consistency Methods. In research task 3, a novel domain mapping approach is proposed to translate customer requirements, functional requirements and design parameters into product attributes for analysis, synthesis and evaluation of design concepts. Research task 4 proposes a multi-agent reinforcement learning technique to enable different agents to work and learn collaboratively in a shared design environment and automate the design process. A scalable intelligent design architecture is proposed to use multi-agent reinforcement learning for adaptable and optimum design decisions based on customer requirements. This approach improves time-consuming conventional manual design methods. The effectiveness of the proposed methods is evaluated in designing and prototyping a rehabilitation device in research task 5. The proposed hand rehabilitation device is fabricated and prototyped. | |
| dc.description.note | October 2023 | |
| dc.description.sponsorship | Natural Sciences and Engineering Research Council (NSERC) of Canada University of Manitoba Lab2Market Mitacs | |
| dc.identifier.uri | http://hdl.handle.net/1993/37459 | |
| dc.language.iso | eng | |
| dc.rights | open access | en_US |
| dc.subject | Product Design | |
| dc.subject | Reinforcement Learning | |
| dc.subject | Concept Generation | |
| dc.subject | Systems Engineering | |
| dc.title | Product concept development and evaluation based on Quality Function Deployment, Axiomatic Design and Reinforcement Learning | |
| dc.type | doctoral thesis | en_US |
| local.subject.manitoba | no | |
| project.funder.name | Natural Sciences and Engineering Research Council (NSERC) of Canada |