漸增式資源配置之決策支援系統－以警用巡邏箱配置為例

Abstract

一般的資源配置問題大多是在給定條件下追求具有最大效益或最低成本的資源配置，一旦有新資源加入（條件異動）就必須重新分配，所以新增資源後配置結果能否與之前配置結果相容將對整體成本有重要的影響。過去資源配置方法欲從m處預定地選出n處配置資源時，大多無法在舊有配置結果上新增資源，而且還可能因人爲疏忽在同一地點設置多個資源，造成配置後的資源不是相距太近就是相隔太遠的缺點，所以需要更有效的資源配置方法。爲解決此類資源配置問題，本文首先分析此類資源配置問題特性；其次根據決策者需求來設計合適的目標函數；最後以基因演算法與地理資訊系統爲基礎建構一決策支援系統提供決策者合適的資源配置建議。文中實驗結果除能顯示本文所提方法確實可以避免上述不當配置之外，或許也可以提供其他領域中類似的資源配置問題參考之用。

General resource allocation problems usually aim to pursue the maximum benefits or the lowest cost under given conditions. Once some extra resources are available (i.e., condition changed), we need to re-allocate these resources. Therefore, whether a re-allocation remains consistent with the previous one becomes a decisive factor in cost reduction. When choosing n proper places out of m, traditional resource allocation schemes may suffer from the following drawbacks. They cannot add new resources directly to a previous allocation and thus the previous allocation will be utterly repudiated. Moreover, they might assign multiple resources almost to the same place due to culpable disregard. Consequently, two allocated resources may be too distant from or too close to each other. A more practical method is thus called for. In this paper, a practical method is proposed to solve such kind of problems. First, the properties of the problem are analyzed. Second, a suitable objective function is defined according to decision makers' needs. Finally, a decision support system is implemented to provide proper resource allocation suggestions, which is based on Genetic Algorithm (GA) and Geographic Information System (GIS). The experimental results show that the proposed method can prevent inappropriate resource allocations and also suggest that it may be useful to some similar cases in other fields.