ニューロコンピューティングに基づく定期補修計画問題の一解法

DOI Web Site Web Site 被引用文献1件

書誌事項

タイトル別名
  • A Practical Method for Solving Maintenance Scheduling Problem based on Neurocomputing
  • ニューロコンピューティング ニ モトズク テイキ ホシュウ ケイカク モンダイ

この論文をさがす

抄録

In this paper, we propose a practical method of solving a reliability objective maintenance scheduling problem based on neurocomputing approach. The objective of this problem is to levelize reserve over planning periods. In the proposed method, the Dynamical Canonical Network for nonlinear programming, which developed by Kennedy and Chua, is modified to apply the neural network to 0-1 integer programming.<br>However, it is difficult to obtain good solutions using neurocomputing. Therefore, we observe the effectiveness of ‘excess-bias method’ called and apply it to neural network for 0-1 integer programming so that we get good solutions. Nevertheless, its algorithm needs considerable time, and so we don't think it is an efficient algorithm. An efficient algorithm is desirable to rapidly obtain good solutions because we put neurocomputing as fast-approximation method. Hence, we propose to improve the algorithm of the excess-bias method to rapidly find good solutions.<br>The proposed algorithm is examined on three types of model system, small scale, medium scale, and real scale. In consequence of the computation, the effectiveness of the neurocomputing has been verified. And, the proposed algorithm has been found to be in a middle situation between normal and excess-bias method for the accuracy and the speed of the computation.

収録刊行物

被引用文献 (1)*注記

もっと見る

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ