Take-out Delivery Research Based On Generalized Multi-offspring Genetic Algorithm

With the development of online ordering platform,take-out delivery become important and important in the operation of online ordering platform.Improving efficiency of take-out delivery become a necessary choice of online ordering platform.The rationality of the transport route directly affects the delivery speed,distribution costs and customer satisfaction,especially the process of determining the distribution line of the multi-restaurant,multi-customer,multi-vehicle complex distribution network is indeed a complex system engineering.Select an appropriate vehicle path,you can speed up the respond speed of customer demand,improve service quality,enhance customer satisfaction with the distribution process and improve operational efficiency.The existing decision-making method of take-out distribution is artificial decision-making.Due to the limitation of human computing ability,it is impossible to manually order the orders of different restaurants,so that the multi-restaurant cannot be fed together and the overall distribution efficiency can be reduced.In this paper,the issue of take-away distribution is transformed into vehicle routing problem.Then,transform the solution of vehicle routing problems to solution of take-off distribution.Because the vehicle routing problem is solved by computer,it can solve the problem of large-scale vehicle routing in a limited time.Therefore,the vehicle routing problem model can solve the multi-restaurant distribution problem,reducing the waste of repeated distribution path and imbalance of load between restaurants,improving efficiency.The experimental results show that the vehicle routing problem distribution method has obvious advantages over the manual distribution method,the average distribution distance is short,the maximum delivery time is short,and the average waiting time is short.Vehicle routing problem is the famous NP-hard problem,which has a wide range of engineering and realistic background.Many practical problems can be transformed into vehicle routing problem,such as the study of the take-away distribution,which can be transformed into a series of vehicle routing problem.Because the vehicle routing problem is a NP-hard problem,the exact solution of large-scale vehicle routing problem cannot be obtained in polynomial time,and the heuristic algorithm is needed to find the approximate solution.In this paper,genetic algorithm is used to solve the VRP problem.Genetic algorithm is a parallel and random adaptive search algorithm for simulating biological selection and evolution mechanism.It is suitable for dealing with complex and nonlinear problems which are not easy to solve for traditional search algorithm.It can parallel search in the global search domain,with a high efficiency.However,the demand for delivery is very high,and the algorithm is required to give the result as soon as possible.This requires a fast convergence algorithm,this paper use multi-offspring genetic algorithm for this task.However,this multi-offspring genetic algorithm requires that the number of offspring is an integer multiple of the number of the parent,so the flexibility is not enough.The value of the number of offspring is likely to miss the optimal value.This paper extends the multi-offspring genetic algorithm,improving the method to generate offspring,proposing generalized multi-offspring genetic algorithm.The method to generate offspring of generalized multi-offspring genetic algorithm is more flexible and easier to obtain the optimal value,improving efficiency of the algorithm.