عنوان مقاله :
تقاضاي تصادفي با ماهيت درون زا در مسئله مسيريابي وسايل نقليه
عنوان به زبان ديگر :
ENDOGENOUS UNCERTAIN DEMANDS IN VEHICLE ROUTING PROBLEM
پديد آورندگان :
هوشمند خليق، فرناز , ميرحسني، علي دانشگاه صنعتي اميركبير - دانشكده رياضي و علوم كامپيوتر
كليدواژه :
برنامه ريزي تصادفي چندمرحله يي , عدم قطعيت درون زا , درخت سناريوي وابسته به تصميم , قيود عدم ناهماهنگي , شناسايي قيود زائد , مسيريابي وسيله نقليه
چكيده فارسي :
برنامه ريزي تصادفي با عدم قطعيت درون زا مبحثي جديد است كه در آن فرايند تصادفي مسئله تحت تاثير تصميمات قرار دارد. در اين مقاله، توسعه ي نويني از مسئله ي مسيريابي وسايل نقليه به عنوان كاربردي جديد از اين مبحث معرفي مي شود كه در آن، تقاضاي هر مشتري غيرقطعي است و عدم قطعيت در لحظه ي ملاقات محقق مي شود. لذا، زمان محقق شدن عدم قطعيت وابسته به تصميمات و عدم قطعيت داراي ماهيت درون زاست. ابتدا يك مدل برنامه ريزي تصادفي چندمرحله يي با عدم قطعيت درون زا براي اين مسئله ارائه مي شود. سپس، از آنجا كه قيود عدم ناهماهنگي به صورت شرطي هستند و درصد بالايي از كل قيود را تشكيل مي دهند، چندين رويكرد كارا براي كاهش اين قيود ارائه و اثبات مي شود كه تاثير چشم گيري در كاهش اندازه ي مسئله و زمان حل خواهد داشت. نتايج محاسباتي عملكرد مدل پيشنهادي و روش هاي كاهش قيود را روي چند نمونه ي تصادفي نشان مي دهند.
چكيده لاتين :
Stochastic programming under endogenous uncertainty is a new topic in which the time of uncertainty realization is affected by optimization decisions. Development of application areas in this field has been of interest to researchers. Therefore, this paper addresses a novel variant of the vehicle routing problem as a new application of this topic. In this problem, a single vehicle with a limited capacity must serve a set of customers whose demands are uncertain and the actual demands are revealed upon the vehicles arrival to customers. Under some circumstances, the vehicle may be depleted throughout the route before all customers are satisfied; in this case, it needs to return to the depot for replenishment. The objective is to find a policy to satisfy customers demand with the minimum total expected cost. The main point is that this problem is considered from a dynamic viewpoint. Indeed, the route is constructed gradually over multiple stages according to the realized information. For example, the primary decision determines a customer who should be visited in the first stage. Upon the vehicle visiting this customer, his/her actual demand becomes known; then, depending on the realized information, the next decision is to determine a customer who must be visited next, whether directly or after the replenishment at the depot. This process is repeated until all customers are visited. Finally, the vehicle returns to the depot. Clearly, the time of uncertainty realization is decision-dependent and hence, uncertainty is of endogenous nature. Therefore, the scenario tree becomes decision-dependent and nonanticipativity of decisions must be ensured by conditional constraints. This paper formulates the problem as a multi-stage stochastic programming model with endogenous uncertainty. Then, since nonanticipativity constraints constitute a considerably large portion of the total constraints, efficient approaches are proposed to significantly reduce the problem size and improve the solution time. Computational results evaluate the proposed model and approaches on some randomly generated test problems.
عنوان نشريه :
مهندسي صنايع و مديريت شريف
عنوان نشريه :
مهندسي صنايع و مديريت شريف
