- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Theses and Dissertations /
- Artificial neural network-assisted repair technique...
Open Collections
UBC Theses and Dissertations
UBC Theses and Dissertations
Artificial neural network-assisted repair technique for handling constraints in structural optimization Cai, Yuecheng
Abstract
Structural design typically involves nonconvex criteria that need effective optimization algorithms which can find the global optimum or Pareto optima. Constraints create complex hyperspaces that are difficult to navigate, and traditional constraint handling techniques (CHTs) might not be capable of steering the search. Repair techniques are one type of CHTs that can be very effective but have a few limitations that restrict their use. We here present a new repair-based CHT that addresses these issues by being: (i) adaptive to the share of infeasible solutions in a population and (ii) free of problem-specific heuristic for repair that a user typically needs to provide. Only the best performing infeasible solutions are repaired, to balance the normal operating procedure of the optimization algorithm with CHT, i.e., minimizing objectives and satisfying constraints. A procedure is proposed to apply artificial neural network (ANN) to automate the definition of problem-specific knowledge by identifying and ranking the most significant variables that influence each constraint. The proposed CHT approach is implemented in single-objective swarm algorithm PSO and multi-objective evolutionary algorithms NSGA-II and MOEA/D. The following test cases are considered: mathematical benchmark problem, truss optimization and structural optimization of a chemical tanker’s main frame. Trained ANN is used as surrogate model in the latter case. In comparison to the original algorithms, a few state-of-the-art algorithms and CHTs, all modified algorithms show significantly better performance.
Item Metadata
Title |
Artificial neural network-assisted repair technique for handling constraints in structural optimization
|
Creator | |
Supervisor | |
Publisher |
University of British Columbia
|
Date Issued |
2022
|
Description |
Structural design typically involves nonconvex criteria that need effective optimization algorithms which can find the global optimum or Pareto optima. Constraints create complex hyperspaces that are difficult to navigate, and traditional constraint handling techniques (CHTs) might not be capable of steering the search. Repair techniques are one type of CHTs that can be very effective but have a few limitations that restrict their use. We here present a new repair-based CHT that addresses these issues by being: (i) adaptive to the share of infeasible solutions in a population and (ii) free of problem-specific heuristic for repair that a user typically needs to provide. Only the best performing infeasible solutions are repaired, to balance the normal operating procedure of the optimization algorithm with CHT, i.e., minimizing objectives and satisfying constraints. A procedure is proposed to apply artificial neural network (ANN) to automate the definition of problem-specific knowledge by identifying and ranking the most significant variables that influence each constraint. The proposed CHT approach is implemented in single-objective swarm algorithm PSO and multi-objective evolutionary algorithms NSGA-II and MOEA/D. The following test cases are considered: mathematical benchmark problem, truss optimization and structural optimization of a chemical tanker’s main frame. Trained ANN is used as surrogate model in the latter case. In comparison to the original algorithms, a few state-of-the-art algorithms and CHTs, all modified algorithms show significantly better performance.
|
Genre | |
Type | |
Language |
eng
|
Date Available |
2022-01-12
|
Provider |
Vancouver : University of British Columbia Library
|
Rights |
Attribution-NonCommercial-NoDerivatives 4.0 International
|
DOI |
10.14288/1.0406265
|
URI | |
Degree | |
Program | |
Affiliation | |
Degree Grantor |
University of British Columbia
|
Graduation Date |
2022-05
|
Campus | |
Scholarly Level |
Graduate
|
Rights URI | |
Aggregated Source Repository |
DSpace
|
Item Media
Item Citations and Data
Rights
Attribution-NonCommercial-NoDerivatives 4.0 International