استفاده از مدلسازي در بهينه‌سازي چندوظيفه‌اي تكاملي براي مسئله مسيريابي وسيله‌ي نقليه

In the modern era, many businesses have moved the location of their goods an‎d services from the door of the business to the door of the home for the convenience of customers. Customer service requires a transportation system an‎d faces challenges. One of these challenges is how to service customers in a way that both satisfies customers an‎d reduces costs. Solving the Vehicle Routing Problem (VRP) can be an answer to this challenge by finding optimal routes, reducing costs, energy consumption, an‎d consequently environmental pollution. There are various methods for solving VRP, one of which is evolutionary algorithms. For each optimization operation, algorithms have to start from scratch. To prevent the entire optimization from being performed in one run, transfer optimization comes to the aid of algorithms an‎d improves the optimization process by transferring the learned knowledge. One of the methods of transfer optimization is the simultaneous optimization of multiple tasks in parallel, called evolutionary multitask optimization (EMTO). In this method, knowledge is exchanged between different tasks (problems) during optimization. In this research, a method for evolutionary multitask optimization is presented, which is tested on a variant of VRP, namely capacitated VRP (CVRP), due to the great diversity of VRP. The proposed method, based on the models obtained from the tasks, makes a mapping between the information of two tasks an‎d then uses this mapping to perform inter-task transfer. By implementing this method on 2 basic methods an‎d a leading method in solving some VRPs, the important components for implementing this method, the criteria of suitable algorithms for implementing them, an‎d how to use this method have been identified, an‎d it has been shown that the proposed method is able to improve the results based on the obtained criteria by an average of 32% without using specific problem data.

5

162261