工作流程管理系統使企業內部的工作流程自動化，幫助企業降低成本、改善工作效率，以及管理、監控各工作的執行與績效評估。隨著企業工作業務快速增加以及工作流程複雜度不斷提高，工作流程中的活動出現延遲的狀況愈來愈頻繁，這些延遲的活動使得工作時程無法預期完成且無法發揮企業應有的效益。因此本研究從歷史的工作時程記錄中挖掘出有異常時間延遲的活動，提供給企業主管作為改善工作流程的依據。 本研究的目的在於準確地找到工作流程中發生的各種類型時間延遲，因此發展一套模糊特例時間演算法，根據工作流程中重要的活動屬性，以模糊理論為基礎來估算活動應有的執行時間，比較能反應出工作流程實際的狀況，有效地找出各類型的時間延遲。以石材業的全球運籌流程來驗證本方法的準確性，並將本方法與以活動為基礎的特例工作時程演算法計算出的延遲結果比較，結果顯示本方法確實能準確地找到活動的各種類型時間延遲，而且準確率確實高出許多。本方法找出的延遲結果能在企業上被解釋出原因的比例也高出許多，比較能反應企業流程運作的實際狀況。 本研究發展出的模糊特例時間演算法對模糊推論系統作了一些改善，將需由專家判斷的步驟省略，增加開發者在設計模糊推論系統時的彈性。 Workflow management system (WFMS) enables the enterprise processes to be automatic, helping companies reduce costs, improving work efficiency, monitoring the work of the implementation and evaluating performance. With the transactions of business increase rapidly and the complexity of workflow enhance quickly, the activities in the workflow emergence more and more frequent time duration, the time duration make the schedule of activities can not be expected to complete and bring the business’ skill into full play. Therefore, this study mines out the activities of abnormal time duration from the historical records of the workflow, providing these to business executives as the basis for improvement of workflow. Purpose of this paper is to accurately find all types of the time duration occurring in workflow, so we develop a Fuzzy Anomalous Duration Algorithm (FADA), based the important activity properties in the workflow to estimate the proper executive time of activities with fuzzy theory. This algorithm can more able to reflect the actual status of the workflow, and effectively identify all types of the time duration. We use the global logistics process of the stone industry to validate the accuracy of this algorithm, and compared the algorithm’s time duration result with Activity-based Algorithm for Mining Temporal Outlier (ABTO). The results show that this algorithm can correctly find all types of time duration, and the accuracy rate is much higher. The time duration results of this algorithm can be interpreted in the enterprise on a much higher proportion than ABTO, and more able to respond to the actual situation of workflow operations. This paper’s algorithm (FADA) makes some improvements on the fuzzy inference system; it leaves out the step that needs experts’ judge. And it increases the developers’ flexibility in the design of fuzzy inference system.
