{"id":{"repo_id":"birmingham","oai_identifier":"oai:etheses.bham.ac.uk:14309"},"canonical_url":"https://search.ndltd.org/etd/birmingham/oai:etheses.bham.ac.uk:14309","repository":{"repo_id":"birmingham","name":"University of Birmingham","base_url":"https://etheses.bham.ac.uk/cgi/oai2"},"display":{"title":"Surrogate-assisted evolutionary algorithms for computationally expensive optimisation problems","abstract":"Surrogate-assisted evolutionary algorithms (SAEAs) are designed to solve black-box optimisation problems that are computationally expensive. These optimisation problems could be multi-objective or/and constrained. In industry, many real-world optimisation problems, such as engine calibration problems, allow only a few evaluations due to their high cost and therefore limited budgets. To save these valuable budgets, this thesis focuses on developing effective and efficient SAEAs to solve computationally expensive optimisation problems. This thesis makes three major contributions: First, to achieve better optimisation results within limited evaluations, it proposes a domination-based ordinal regression surrogate for SAEAs. This ordinal regression surrogate approximates the ordinal landscape of multiple objectives. Two surrogate management strategies are also proposed to cooperate with the ordinal regression surrogate. The resulting SAEAs outperform the state-of-the-art algorithms on a set of multi-objective optimisation problems. Second, it develops a bilevel SAEA to handle constrained optimisation problems. In the algorithm, decision variables are divided into either upper-level or lower-level variables according to their impacts on solution feasibility. The lower-level optimisation focuses on lower-level variables to make candidate solutions feasible, while the upper-level optimisation adjusts upper-level variables to optimise objective(s). The algorithm finds more feasible solutions and better optimisation results in an engine calibration problem when compared with the algorithms used in the engine industry and some state-of-the-art algorithms. Third, for the sake of saving more evaluations, it proposes an experience-based SAEA framework to learn experience from related tasks and then use the learned experience in a new optimisation task. The experience is learned through a novel meta-learning technique and they represent the domain-specific features among related tasks. Assisted by the learned experience, accurate surrogates can be learned with very few evaluated samples in an efficient way. Experimental studies have demonstrated the effectiveness of experience learning and that of using experience in an existing SAEA. Competitive optimisation results are achieved while fewer evaluations are used than before, which saves valuable evaluation budgets considerably.","abstract_html":"Surrogate-assisted evolutionary algorithms (SAEAs) are designed to solve black-box optimisation problems that are computationally expensive. These optimisation problems could be multi-objective or/and constrained. In industry, many real-world optimisation problems, such as engine calibration problems, allow only a few evaluations due to their high cost and therefore limited budgets. To save these valuable budgets, this thesis focuses on developing effective and efficient SAEAs to solve computationally expensive optimisation problems. This thesis makes three major contributions: First, to achieve better optimisation results within limited evaluations, it proposes a domination-based ordinal regression surrogate for SAEAs. This ordinal regression surrogate approximates the ordinal landscape of multiple objectives. Two surrogate management strategies are also proposed to cooperate with the ordinal regression surrogate. The resulting SAEAs outperform the state-of-the-art algorithms on a set of multi-objective optimisation problems. Second, it develops a bilevel SAEA to handle constrained optimisation problems. In the algorithm, decision variables are divided into either upper-level or lower-level variables according to their impacts on solution feasibility. The lower-level optimisation focuses on lower-level variables to make candidate solutions feasible, while the upper-level optimisation adjusts upper-level variables to optimise objective(s). The algorithm finds more feasible solutions and better optimisation results in an engine calibration problem when compared with the algorithms used in the engine industry and some state-of-the-art algorithms. Third, for the sake of saving more evaluations, it proposes an experience-based SAEA framework to learn experience from related tasks and then use the learned experience in a new optimisation task. The experience is learned through a novel meta-learning technique and they represent the domain-specific features among related tasks. Assisted by the learned experience, accurate surrogates can be learned with very few evaluated samples in an efficient way. Experimental studies have demonstrated the effectiveness of experience learning and that of using experience in an existing SAEA. Competitive optimisation results are achieved while fewer evaluations are used than before, which saves valuable evaluation budgets considerably.","abstract_has_math":false,"creators":["Yu, Xunzhao"],"institution":"University of Birmingham","degree_name":"d_ph","degree_level":"d_ph","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2023,"date_issued":"2023-12","date_published":"2023-12","updated_at":"2026-11-19T09:32:33Z","subjects":["QA75 Electronic computers. Computer science"],"languages":["eng"],"rights":[],"rights_urls":[],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":null,"outbound_label":null,"outbound_source":null,"record_url":null,"record_label":null,"record_kind":null,"record_source":null,"download_url":"http://etheses.bham.ac.uk//id/eprint/14309/1/Yu2023PhD.pdf","download_label":"Download PDF","download_source":"dc:identifier.uri","access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://etheses.bham.ac.uk/cgi/oai2?verb=GetRecord&metadataPrefix=uketd_dc&identifier=oai%3Aetheses.bham.ac.uk%3A14309","prefix":"uketd_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.sponsor","label":"Sponsor","values":["other"]},{"key":"dc:creator","label":"Author","values":["Yu, Xunzhao"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2023-12"]},{"key":"dc:date.issued","label":"Date","values":["2023-12"]},{"key":"dc:publisher.department","label":"Dc Publisher Department","values":["College of Engineering & Physical Sciences","School of Computer Science"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["University of Birmingham"]},{"key":"dc:relation.isreferencedby","label":"Dc Relation Isreferencedby","values":["http://etheses.bham.ac.uk//id/eprint/14309/"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["d_ph"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["d_ph"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["QA75 Electronic computers. Computer science"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.dcterms_iso639_2","label":"Language (ISO 639-2)","values":["eng"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["http://etheses.bham.ac.uk//id/eprint/14309/1/Yu2023PhD.pdf"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Surrogate-assisted evolutionary algorithms (SAEAs) are designed to solve black-box optimisation problems that are computationally expensive. These optimisation problems could be multi-objective or/and constrained. In industry, many real-world optimisation problems, such as engine calibration problems, allow only a few evaluations due to their high cost and therefore limited budgets. To save these valuable budgets, this thesis focuses on developing effective and efficient SAEAs to solve computationally expensive optimisation problems. This thesis makes three major contributions: First, to achieve better optimisation results within limited evaluations, it proposes a domination-based ordinal regression surrogate for SAEAs. This ordinal regression surrogate approximates the ordinal landscape of multiple objectives. Two surrogate management strategies are also proposed to cooperate with the ordinal regression surrogate. The resulting SAEAs outperform the state-of-the-art algorithms on a set of multi-objective optimisation problems. Second, it develops a bilevel SAEA to handle constrained optimisation problems. In the algorithm, decision variables are divided into either upper-level or lower-level variables according to their impacts on solution feasibility. The lower-level optimisation focuses on lower-level variables to make candidate solutions feasible, while the upper-level optimisation adjusts upper-level variables to optimise objective(s). The algorithm finds more feasible solutions and better optimisation results in an engine calibration problem when compared with the algorithms used in the engine industry and some state-of-the-art algorithms. Third, for the sake of saving more evaluations, it proposes an experience-based SAEA framework to learn experience from related tasks and then use the learned experience in a new optimisation task. The experience is learned through a novel meta-learning technique and they represent the domain-specific features among related tasks. Assisted by the learned experience, accurate surrogates can be learned with very few evaluated samples in an efficient way. Experimental studies have demonstrated the effectiveness of experience learning and that of using experience in an existing SAEA. Competitive optimisation results are achieved while fewer evaluations are used than before, which saves valuable evaluation budgets considerably."]},{"key":"dc:format","label":"Dc Format","values":["text"]},{"key":"dcterms:accessrights","label":"Dcterms Accessrights","values":["public"]},{"key":"dc:title","label":"Title","values":["Surrogate-assisted evolutionary algorithms for computationally expensive optimisation problems"]}]}],"canonical_facts":{"dc:contributor.sponsor":["other"],"dc:creator":["Yu, Xunzhao"],"dc:date":["2023-12"],"dc:date.issued":["2023-12"],"dc:description.abstract":["Surrogate-assisted evolutionary algorithms (SAEAs) are designed to solve black-box optimisation problems that are computationally expensive. These optimisation problems could be multi-objective or/and constrained. In industry, many real-world optimisation problems, such as engine calibration problems, allow only a few evaluations due to their high cost and therefore limited budgets. To save these valuable budgets, this thesis focuses on developing effective and efficient SAEAs to solve computationally expensive optimisation problems. This thesis makes three major contributions: First, to achieve better optimisation results within limited evaluations, it proposes a domination-based ordinal regression surrogate for SAEAs. This ordinal regression surrogate approximates the ordinal landscape of multiple objectives. Two surrogate management strategies are also proposed to cooperate with the ordinal regression surrogate. The resulting SAEAs outperform the state-of-the-art algorithms on a set of multi-objective optimisation problems. Second, it develops a bilevel SAEA to handle constrained optimisation problems. In the algorithm, decision variables are divided into either upper-level or lower-level variables according to their impacts on solution feasibility. The lower-level optimisation focuses on lower-level variables to make candidate solutions feasible, while the upper-level optimisation adjusts upper-level variables to optimise objective(s). The algorithm finds more feasible solutions and better optimisation results in an engine calibration problem when compared with the algorithms used in the engine industry and some state-of-the-art algorithms. Third, for the sake of saving more evaluations, it proposes an experience-based SAEA framework to learn experience from related tasks and then use the learned experience in a new optimisation task. The experience is learned through a novel meta-learning technique and they represent the domain-specific features among related tasks. Assisted by the learned experience, accurate surrogates can be learned with very few evaluated samples in an efficient way. Experimental studies have demonstrated the effectiveness of experience learning and that of using experience in an existing SAEA. Competitive optimisation results are achieved while fewer evaluations are used than before, which saves valuable evaluation budgets considerably."],"dc:format":["text"],"dc:identifier.uri":["http://etheses.bham.ac.uk//id/eprint/14309/1/Yu2023PhD.pdf"],"dc:language.dcterms_iso639_2":["eng"],"dc:publisher.department":["College of Engineering & Physical Sciences","School of Computer Science"],"dc:publisher.institution":["University of Birmingham"],"dc:relation.isreferencedby":["http://etheses.bham.ac.uk//id/eprint/14309/"],"dc:subject":["QA75 Electronic computers. Computer science"],"dcterms:accessrights":["public"],"dc:title":["Surrogate-assisted evolutionary algorithms for computationally expensive optimisation problems"],"dc:type":["Thesis"],"dc:type.qualificationlevel":["d_ph"],"dc:type.qualificationname":["d_ph"]},"updated_at":"2026-11-19T09:32:33Z"}