台灣四周環海，公部門艦艇遂行保衛海疆任務繁重，需經常執行不特定之任務，裝備常面臨惡劣的使用環境。如何在有限的國家預算資源下，有效提升公艦艇核心裝備的妥善率及壽命是維繫國家安全的首要工作。艦艇在海上航行全部依賴主機所提供的動力，因此主機在運轉時需持續透過各種監控儀器以記錄判斷各項參數值，同時配合例行性之計畫性維護作業以確保主機處於最佳的維運狀態。然而目前維護保養策略是根據設備製造商提供之技術手冊及規範執行船隻修護及保養，此類預防性維護模式仍然無法完全避免艦艇主機產生意外故障。本研究運用羅吉斯迴歸演算法透過分析某船艦104至106年主機監控紀錄及歷史維修保養紀錄，以建立船用主機預測維修模式，可有效辨識主機的失效事件，藉此提供我國公部門在船艦裝備的保養作業及預測維修作為時之決策參考，期盼可以有效降低裝備發生故障的次數，使維修及成本降低，以便在有限的國防預算中提高妥善率。

To secure surrounding seas of Taiwan, ROC Naval fleet often perform ad-hoc missions which severely endanger the equipment. How to maintain the availability and life of ships effectively with the limited budget become the priority in Navy's continuously task. ROCN endeavors to protect the maritime areas and the equipment often faces harsh operating conditions. Sailing on the sea, it relies on the engine's generating power and its reliability act decisive role as performing combating capability. Nowadays, the maintenance operation based on the producer's documents cannot effectively predict and analyze the time when the ships will break down. Based on the monitoring and historical maintenance record of Vessels in public sector and using Logistic Regression Algorithm, this study construct a predictive model to detect fault of marine engine which can provide a feasible solution for public sector. As a result, the number of equipment malfunctions can be effectively reduced and maintenance costs can be reduced while still increasing the availability in the limited national defense budget.