{"created":"2024-05-21T01:34:38.335799+00:00","id":2000219,"links":{},"metadata":{"_buckets":{"deposit":"fbef574a-6d41-4345-9674-e042b2739950"},"_deposit":{"created_by":20,"id":"2000219","owner":"20","owners":[20],"pid":{"revision_id":0,"type":"depid","value":"2000219"},"status":"published"},"_oai":{"id":"oai:muroran-it.repo.nii.ac.jp:02000219","sets":["41:227"]},"author_link":[],"control_number":"2000219","item_81_date_granted_17":{"attribute_name":"学位授与年月日","attribute_value_mlt":[{"subitem_dategranted":"2024-03-22"}]},"item_81_degree_grantor_10":{"attribute_name":"学位授与機関","attribute_value_mlt":[{"subitem_degreegrantor":[{"subitem_degreegrantor_language":"ja","subitem_degreegrantor_name":"室蘭工業大学"},{"subitem_degreegrantor_language":"en","subitem_degreegrantor_name":"Muroran Institute of Technology"}],"subitem_degreegrantor_identifier":[{"subitem_degreegrantor_identifier_name":"10103","subitem_degreegrantor_identifier_scheme":"kakenhi"}]}]},"item_81_degree_name_11":{"attribute_name":"学位名","attribute_value_mlt":[{"subitem_degreename":"博士(工学)","subitem_degreename_language":"ja"}]},"item_81_description_7":{"attribute_name":"抄録","attribute_value_mlt":[{"subitem_description":"進化アルゴリズムは,生物の進化過程からインスパイアを得ており,交叉,変異戦略,自然選択といった進化操作により最適化を実現している。進化計算は,その高い汎用性および頑健性から,伝統的な数学的計画に比べて幅広い問題に対処することができるという特徴を持つ。特に最近では,進化計算における多点探索という特徴を活かした多目的最適化問題への応用研究が大きな注目を集めている。本研究は,多目的最適化のための効果的な進化計算手法として,新規個体生成のためのメカニズムとしてCX(Cross-over)およびED(Estimation of distribution)戦略を組み合わせた新たな手法を開発し,その有用性を検証した。\n多目的最適化は,単目的最適化よりも複雑であるため,探索性能を向上させるためには,より体系的かつ適応的な手法が必要となる。本研究では,代表的な多目的最適化のための進化計算手法であるMOEA/D(Multi-Objective Evolutionary Algorithm Based on Decomposition)フレームワークに焦点を当て,そのフレームワークに新たなメカニズムを導入することによる性能改善を試みた。一般的に,進化計算における新規個体生成では,幅広い範囲を対象とする探索と局所的な範囲を対象にする探査をバランスさせることが重要とされているが,事前の取り組みを通じて,単一の演算子のみでは両者を同時に実現することが困難であることが判明した。そのため,本研究では複数の個体生成の戦略を適応的に切り替える方法を実現することで,探索と探査を自動的にバランスさせ,効率的な探索を実現する手法について検討を行った。\n本研究で着目したMOEA/Dフレームワークでは,多目的問題を複数の単目的問題に分割し探索を行う。そのため,単目的問題において大域的な探索を指向する戦略と局所的な探索を指向する戦略を組み合わせることで,多目的最適化においても適応的に両方の戦略を切り替える手法を実現することができる。本研究では,この考えに基づきMOEA/D-EFモデルおよびMOEA/D-HHモデルとしてMOEA/Dを効率化した新規手法を提案した。これらのモデルはMOEA/Dフレームワークに適応的に複数の演算子を切り替えるメカニズムを導入することで探索の効率化を実現している。\n論文の第二章では,MOEA/Dフレームワーク,CX戦略およびED戦略に基づく進化演算子,多目的最適化問題で使用される2つの性能メトリクスについて検討しており,第三章では,本研究で導入された進化演算子に焦点を当て,IDE,JADE,DE-IDEAL,およびCMAESについて解説している。第四章では,提案するMOEA/D-EFモデルについて説明し,第五章ではEfficiency Inspectionに基づく演算子切り替えメカニズムを組み込んだMOEA/D-HHモデルについて焦点を当て説明している。","subitem_description_language":"ja","subitem_description_type":"Abstract"},{"subitem_description":"Evolutionary algorithms draw inspiration from the natural process of biological evolution, encompassing gene encoding, crossover, mutation strategies, and mechanisms of natural selection. Due to their high robustness and self-learning characteristics, evolutionary computing has emerged as an advanced global optimization technique for handling complex problems, proving more effective than traditional mathematical planning. In recent years, research has increasingly focused on utilizing evolutionary computingfor both single and multi-objective optimization problems. Our study specifically concentrates on effectively addressing multi-objective optimization problems, with a particular emphasis on individual generation methods based on the CX (Crossover) and ED (Estimation of Distribution) strategies in evolutionary algorithms.\nMulti-objective optimization problems, more intricate than single-objective ones, demand systematic approaches. We direct our attention to the MOEA/D (Multi-Objective Evolutionary Algorithm Based on Decomposition) framework. However, attempting to design new operators or modify classical ones to enhance overall algorithm efficiency revealed that a single operator cannot handle all search scenarios. Operator search capabilities are typically represented by exploration and exploitation methods, and concurrently possessing both capabilities is often challenging. To overcome this issue and enhance algorithmic search efficiency, it is necessary to combine multiple operators with different search characteristics into a hybrid algorithm and introduce an adaptive operator switching mechanism.\nFurthermore, the adaptability of operators within the framework is crucial. Many evolutionary operators were initially designed to mimic the evolution in nature and may not be well-suited for multi-objective optimization problems. In this regard, the MOEA/D framework has caught our attention. This framework can decompose multi-objective optimization problems into multiple sub-problems, allowing for the introduction of classical evolutionary operators. Simultaneously, the MOEA/D framework introduces the concept of the neighborhood of sub-problems, enabling information sharing within the neighborhood during the evolutionary process. Considering this unique feature, our primary research goal is to extend classical evolutionary operators to handle multi-objective optimization problems.\nTo achieve this goal, we conducted detailed research and analysis on advanced evolutionary operators with different search characteristics and strategies. Subsequently, we proposed the MOEA/D-EF model and MOEA/D-HH model, introducing adaptive operator switching mechanisms to align with the MOEA/D framework.\nIn the second chapter of the paper, we discuss the MOEA/D framework, evolutionary operators based on CX and ED strategies, and two performance metrics widely used in multi-objective optimization problems. The third chapter focuses on the introduced evolutionary operators, including IDE, JADE, DE-IDEAL, and CMA-ES. The fourth chapter introduces the MOEA/D-EF model, while the fifth chapter delves into the MOEA/D-HH model, emphasizing the operator switching mechanism based on efficiency inspection.","subitem_description_language":"en","subitem_description_type":"Abstract"}]},"item_81_dissertation_number_13":{"attribute_name":"学位授与番号","attribute_value_mlt":[{"subitem_dissertationnumber":"甲第532号"}]},"item_81_identifier_registration":{"attribute_name":"ID登録","attribute_value_mlt":[{"subitem_identifier_reg_text":"10.15118/0002000219","subitem_identifier_reg_type":"JaLC"}]},"item_81_text_12":{"attribute_name":"学位の種別","attribute_value_mlt":[{"subitem_text_language":"ja","subitem_text_value":"課程博士"}]},"item_81_text_14":{"attribute_name":"報告番号","attribute_value_mlt":[{"subitem_text_language":"ja","subitem_text_value":"甲第532号"}]},"item_81_text_15":{"attribute_name":"学位記番号","attribute_value_mlt":[{"subitem_text_language":"ja","subitem_text_value":"博甲第532号"}]},"item_81_text_16":{"attribute_name":"研究科・専攻","attribute_value_mlt":[{"subitem_text_language":"ja","subitem_text_value":"工学専攻"}]},"item_81_version_type_24":{"attribute_name":"著者版フラグ","attribute_value_mlt":[{"subitem_version_resource":"http://purl.org/coar/version/c_970fb48d4fbd8a85","subitem_version_type":"VoR"}]},"item_access_right":{"attribute_name":"アクセス権","attribute_value_mlt":[{"subitem_access_right":"open access","subitem_access_right_uri":"http://purl.org/coar/access_right/c_abf2"}]},"item_creator":{"attribute_name":"著者","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Han Jiayi","creatorNameLang":"en"},{"creatorName":"ハン ジャイ","creatorNameLang":"ja"}],"familyNames":[{},{}],"givenNames":[{},{}]}]},"item_files":{"attribute_name":"ファイル情報","attribute_type":"file","attribute_value_mlt":[{"accessrole":"open_access","date":[{"dateType":"Available","dateValue":"2024-05-21"}],"filename":"A532.pdf","filesize":[{"value":"2.5 MB"}],"format":"application/pdf","mimetype":"application/pdf","url":{"objectType":"fulltext","url":"https://muroran-it.repo.nii.ac.jp/record/2000219/files/A532.pdf"},"version_id":"992b4a86-6301-4fb3-aea7-ae0625ba4821"},{"accessrole":"open_access","date":[{"dateType":"Available","dateValue":"2024-05-23"}],"filename":"A532_summary.pdf","filesize":[{"value":"358 KB"}],"format":"application/pdf","mimetype":"application/pdf","url":{"objectType":"abstract","url":"https://muroran-it.repo.nii.ac.jp/record/2000219/files/A532_summary.pdf"},"version_id":"272bbf56-e752-47eb-a150-377ee0764f3e"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"eng"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourcetype":"doctoral thesis","resourceuri":"http://purl.org/coar/resource_type/c_db06"}]},"item_title":"多目的最適化のためのハイブリッド適応進化アルゴリズムに関する研究","item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"多目的最適化のためのハイブリッド適応進化アルゴリズムに関する研究","subitem_title_language":"ja"},{"subitem_title":"A Study of Hybrid Adaptive Evolutionary Algorithm for Multi-objective Optimization","subitem_title_language":"en"}]},"item_type_id":"81","owner":"20","path":["227"],"pubdate":{"attribute_name":"PubDate","attribute_value":"2024-05-21"},"publish_date":"2024-05-21","publish_status":"0","recid":"2000219","relation_version_is_last":true,"title":["多目的最適化のためのハイブリッド適応進化アルゴリズムに関する研究"],"weko_creator_id":"20","weko_shared_id":-1},"updated":"2024-05-27T06:57:10.896846+00:00"}