| With the rapid development of e-commerce in recent years,the competition in the logistics industry has become increasingly intense.How to choose the optimal route to deliver goods determines the delivery time and cost of the service.Hangzhou CN company has developed several vehicle routing algorithms,which combines strategies such as large-scale neighborhood search and hyper-heuristic strategies.In order to enable these algorithms to serve other outer users,we decided to develop a decision system that fits the vehicle route scheduling scenario.The purpose of this research is to develop a decision system for vehicle route scheduling,abstracting requests with different requirements into algorithm input,using multiple algorithms developed with a uniform interface to solve the problem.The optimal result is presented to the user after it is taken from all the candidate solutions.In the route scheduling scenario,the user wants use the system to deliver goods to various stations using several vehicles with lowest cost.The goods have many different attributes,such as weight,volume,etc.The definition of the cost function can also be defined by the user to support various cost pattern.The vehicle route decision system implemented in this thesis uses the microservice architecture to divide different functional modules into independent microservices.Microservices communicate through MetaQ or Restful API.The former is used for requests that require immediate results,while the latter is used for asynchronous communications and notifications.The advantage of microservice architecture is its cohesive and reduced service scale that needs to be designed.While being implemented,each microservice is based on Spring Boot framework.The system uses HBase and Memcached for persistence or cache.It also involves auxiliary libraries such as Protocol Buffer.At present,the system has successfully went online and served real users,providing algorithm staff and logistics users a way to increase work efficiency. |
PDF Full Text Request