高齡化與身心障礙者福利是已開發國家面對之重要課題，根據國家發展委員會預測，我國將於2025年進入超高齡社會。截至2021年11月，衛生福利部統計我國身心障礙人口已達1,200,743人，占總人口數比例5.12%，直轄六都中以臺南市5.32%與高雄市並列第一。因應現有復康巴士人工排班效率不彰，且難以動態調整，本研究欲建立復康巴士自動排班數學模型並應用ALNS演算法求解多場站運輸撥召問題，根據業者數與排班分區方式將營運情境劃分成單一業者與多元業者對應至分區排班與全區排班共4種情境，並以臺南復康巴士為案例，藉由小型測試驗證ALNS演算法與精確解之誤差控制於5%以內。同時藉由營運資料分析歸納出復康巴士每週之營運具高度重現性，結果顯示以ALNS演算法進行自動化排班多能於10分鐘內收斂，總成本相較現有臺南復康巴士人工排班節省12.28%。此外，若將多元業者整併為單一業者全區排班之情境，則可節省16.93%。本研究建立自動化排班數學模型優化復康巴士撥召服務，實證研究顯示自動化排班與資源共享能帶來具體之效益。研究成果亦能夠應用在其他復康巴士系統，以大幅改善目前復康巴士業者人工排班之困境，具體提升營運效率與服務品質。

Aging society and disability welfare are both crucial issues for developed and developing countries in recent years. According to the National Development Council and Ministry of Health and Welfare, the population of the disability is more than 1.18 million in 2018, while Tainan has the highest disability rate among all municipalities. The ageing population is expected to be more than 20% in 2025, which means that Taiwan will become a super-aged society within 4 years. Currently, the government provides Rehabus, long-term care bus, and universal taxi service to guarantee daily mobility for the disabilities. However, due to manual handling on fleet management, each scheduling task takes about 1 to 3 days. This research mainly focuses on the operation of Rehabuses. Analyses and comparisons are conducted for four kinds of operation ways in terms of operator numbers and partitioning methods on scheduling. An automatic scheduling method based on mathematic model with ALNS algorithm is developed and applied for optimizing the operation. Two cases with a small scale have been used to prove the accuracy of solutions obtained by ALNS algorithm and exact method. It is shown that the difference between solution solved by ALNS algorithm and the exact solution is lower than 5%. An empirical case of the Tainan Rehabus has been conducted with its operational pattern of high repeatability. The average compile time of ALNS algorithm is less than 10 minutes, and the solution can improve up to 12.8% of total cost compared with the solution of manual scheduling. In addition, total amount of 16.93% cost can be improved if all the companies are integrated with automatic scheduling and operation. It is also shown that about 70% of the benefits come from releasing the top 20% demands. The results of this study show that automatic scheduling methods can effectively improve the current manual operation situation. The optimization model developed can also be applied to other similar systems for improving accessible mobility.