海軍成功級艦平時主要負責海域偵巡任務；戰時則依戰況執行要域防衛、聯合反封鎖及聯合截擊作戰等任務。與其他工程系統相同，必須維持艦載系統的正常運作才能發揮艦艇應有功能，以滿足平時與戰時的需求。艦艇除大修（數年一次，依艦型而定）及定期保養（約每年一次）之外，其餘時間均透過泊港期間艦力自修，執行各系統的維修。礙於泊港期間可供維修的時間、人力與預算有限，無法全面檢修，因此艦力自修的內容影響次一任務的執行甚深。 選擇性維修的目的是系統在兩次運行之間的中斷間隔（暫時不需運作），進行維修活動，由於有限的時間及資源，通常無法對所有元件全面維修，因此，必須選擇適當的元件實施適當級別的維修，以維持或提升系統的可靠度。本研究以成功級艦動力系統為例，建構多次任務型態的選擇性維修模型，維修手段包含更換新品、重點維修及一般性維修等。由於模型複雜度依元件數量、任務次數及維修項目等組合呈指數成長且為非線性規劃，因此以簡群演算法為基礎，客制化可有效率求解問題的進化式演算法。

In peacetime, the Navy's Cheng Kung class ships are mainly responsible for the sea area investigation and patrol tasks; in wartime, they are mainly responsible for the defense of important areas, joint anti blockade and joint interception operations according to the combat conditions. Like other engineering systems, the normal operation of shipboard system must be maintained in order to give full play to the ship's functions and meet the needs of peacetime and wartime. In addition to the overhaul (once a few years, depending on the ship type) and regular maintenance (about once a year), the rest of the time is through self-repair during berthing to carry out the maintenance of various systems. Due to the limited time, manpower and cost available for maintenance during berthing, it is impossible to carry out comprehensive maintenance. Therefore, the content of self-repair of ship power has a profound impact on the implementation of the next task. The purpose of selective maintenance is to carry out maintenance activities at the time interval between the two operations of the system. Due to the limited time and resources, it is usually impossible to carry out comprehensive maintenance on all components. Therefore, it is necessary to select the appropriate components to carry out the appropriate level of maintenance in order to maintain or improve the reliability of the system. In this study, the power system of the success class warship is taken as an example to construct a multi mission selective maintenance model. The maintenance strategy covers corrective and preventive maintenance. Since the complexity of the model grows exponentially with the combination of the number of components, the number of tasks and maintenance items, and it is a nonlinear programming, an evolutionary algorithm based on simple group algorithm can be customized to solve the problem efficiently.