2024-03-29T05:38:24Zhttps://ipsj.ixsq.nii.ac.jp/ej/?action=repository_oaipmhoai:ipsj.ixsq.nii.ac.jp:000123472022-10-21T05:24:51Z00581:00690:00699
目的戦略志向協調推論方式の開発と昇降機保全技術者割付けシステムへの適用Development of Goal Intended Coordination Inference Technique and Its Application for Elevator Maintenance Engineer Scheduling Systemjpn論文http://id.nii.ac.jp/1001/00012347/Journal Articlehttps://ipsj.ixsq.nii.ac.jp/ej/?action=repository_action_common_download&item_id=12347&item_no=1&attribute_id=1&file_no=1Copyright (c) 2000 by the Information Processing Society of Japan人工知能株式会社日立ビルシステム株式会社日立ビルシステム株式会社日立ビルシステム株式会社日立製作所システム開発研究所本間, 正喜中村, 晴久中野, 明男鶴田, 節夫エレベーターなどの保守産業が急速に発展し,保守台数が急増する中,昇降機保全技術者割付けの自動化が重要になってきた.ところが,本問のような人員割付けは人や社会的条件が絡むため自動化は容易ではない.1)本問題を分析すると,500都市以上の大規模巡回セールスマン問題に匹敵するうえ,主要制約だけでも勤務条件や顧客条件など100近い複雑・例外的で多分に感覚的な条件を持つ難問であった.2)そこで,知識を目的とその分割・実行・調整・統合により達成する戦略の多階層木として表現し問題を解く目的戦略志向協調推論方式を開発し,適用した.3)全国の代表的な営業所での現場検証の結果,複雑な制約条件を満足する効率の良いスケジュールの作成が手修正時間も含めて従来の50%以下の時間で可能となった.これにより本割付けシステムの実用性ならびに目的戦略志向協調推論方式の有効性を確認した.Increase of the number of elevators causes the deficiency of their maintenance engineers. Under such circumstances, automatic scheduling of above engineers is becoming more essential, as a very promising way to improve both customer satisfaction and working conditions. However, personnel's scheduling is usually a hard problem, involving personal/social conditions. Above automatic scheduling problem is analyzed 1), solved 2), and the technique is estimated 3), as follows:1) The system is at least equal in complexity to a large scale travelling salesman problem rounding over 500 cities. Besides, it involves about 100 complex, exceptional or sensuous conditions such as working conditions and customer satisfaction.2) In order to overcome these difficulties, a technique called ``goal intended strategic coordination inference" is developed and applied. Here, knowledge is represented as a multiple level hierarchical tree that includes nodes comprising goals and their strategies to divide/execute, adjust, and integrate.3) The validation at representative maintenance- bases showed that the system could make efficient and satisfactory schedules with scheduling time approximately 50% less than before, including manual modification time. This proved that the developed system was useful and applied inference technique was effective.AN00116647情報処理学会論文誌414105410622000-04-151882-77642009-06-29