RAILWAY TIMETABLE RESCHEDULING OF AUTOMATED PEOPLE MOVER USING MAX-PLUS ALGEBRA AND ROBUST MODEL PREDICTIVE CONTROL

• Main Author: Aditya Hidayat, Andri
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/54889
• Institution: Institut Teknologi Bandung

Railway transportation is a mode of transportation that has great efficiency, reliability and capacity. The presence of Communication-Based Train Control (CBTC) yields a good impact in many ways such as more accurate train scheduling and train information easily obtained by passenger. The enthusiasm and satisfaction of the public towards this transportation have an impact on accumulation of passenger number at certain stations during rush hours. Increasing headway of train makes disturbances in the form of delays and other more prone to occur. If this delay is not immediately resolved by rescheduling, it can propagate to the next stations and interfere passenger satisfaction. Rescheduling is the main solution to reduce the impact of train departure delay. CBTC technology in Indonesia is currently develope, for example Automated People Mover Systems (APMS) skytrain at Soekarno Hatta International Airport, make it possible for railway timetable to be rescheduled quickly and precisely. In the initial stage, temporary regulation issued by the Indonesian Ministry of Transportation limits train speed to 30 km/h in straight areas and 20 km/h in switching areas. Therefore, to reach the fastest headway, the maximum speed of the train will always be reached so the solution that can be done is to reduce the dwelling time. Train dwelling time has an uncertainty due to the number of departure and arrival train passengers. Robust optimal controller is required to handle these problems. In this study, train departure delay will be solved by train dwelling time minimization. The main objective of this research is to achieve railway timetable rescheduling that can handle dwelling time uncertainty due to the number of passenger arrival and departure on station so that robust optimal railway timetable rescheduling obtained. Max-Plus Algebra is used for APMS train network modelling. Model Predictive Control used to recover train departure delay until railway timetable returns to normal conditions. The polyhedral uncertainty set is chosen in determining the dwelling time using robust optimization. The results showed that the rail network modeling was successfully created using Max-Plus Algebra. The response of Model Predictive Control is able to follow trajectory reference of train departure time. Dwelling time uncertainty problem due to the number of departure and arrival passengers can be formulated mathematically using polyhedral uncertainty sets. Although the recovery time without Robust Optimization is up to 16.35% faster, the solution to the uncertainty problem using Robust Counterpart has advantage that it computationally tractable guarantee and can be solved in polynomial time with a global optimal solution. Keywords: Max-Plus Algebra, Model Predictive Control, Robust Optimization, Polyhedral Uncertainty, Train Timetable. ?