基于改进PSO-BP算法的快递业务量预测  

Prediction of package volume based on improved PSO-BP

作　　者：许荣斌[1,2,3] 王业国 王福田[2,3] 何明慧 汪梦龙 谢莹[1,2] XU Rongbin 1,2,3 , WANG Yeguo 1,2 , WANG Futian 2,3 , HE Minghui 2, WANG Menglong 2, XIE Ying 1,2(1.Key Laboratory of Intelligent Computing ＆ Signal Processing, Ministry of Education, Anhui University, Hefei 230039, China; 2.School of Computer Science and Technology, Anhui University, Hefei 230601, China; 3.Co-Innovation Center for Information Supply ＆ Assurance Technology, Anhui University, Hefei 230601, Chin)

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室,安徽合肥230039 [2]安徽大学计算机科学与技术学院,安徽合肥230601 [3]安徽大学信息保障技术协同创新中心,安徽合肥230601

出　　处：《计算机集成制造系统》2018年第7期1871-1879,共9页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(61602005);教育部人文社科青年基金资助项目(14YJCZH169);安徽省自然科学基金资助项目(1608085MF130;1808085MF199);安徽高校自然科学研究重点项目(KJ2018A0016);安徽大学博士启动基金资助项目~~

摘　　要：为了有效监控快递运输过程,对日常快递业务量进行预测,以保证快递包裹能够按时到达。将大量快递包裹运输过程抽象建模以构造多流程实例;提出改进惯性权重的粒子群优化算法和反向传播神经网络的组合模型(IPSO-BP)来预测物流公司日常快递业务量;进而动态申请合适数量云资源以处理变化的业务需求。大量仿真实验证明,在神经网络参数选择合理的情况下,IPSO-BP模型比其他传统方法有更好的预测效果。To effectively monitor package delivery process, the daily package volume was predicted for ensuring that all packages could reach their destinations on time. A large number of package delivery processes were modeled to construct multi-process instances, and a novel model by combining Improved Particle Swarm Optimization （IPSO） with Back Propagation Neural Network （BPNN） that named IPSO-BP was proposed to predict logistics companies’ daily package volume. A suitable number of cloud resources were dynamically allocated for dealing with changing business needs based on the predicted package volume. A large number of experiments indicated that IPSO-BP model had better prediction effect than other conventional methods when neural network parameters were chosen properly.

关 键 词：物流运输 工作流 粒子群优化算法 反向传播神经网络 快递业务量预测 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]