计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (18): 247-255.DOI: 10.3778/j.issn.1002-8331.1807-0063

• 工程与应用 • 上一篇    下一篇

改进PSO算法优化LSSVM模型的短期客流量预测

陆文星,李楚   

  1. 1.合肥工业大学 管理学院,合肥 230009
    2.智能决策与信息系统技术教育部工程研究中心,合肥 230009
  • 出版日期:2019-09-15 发布日期:2019-09-11

Forecasting of Short-Time Tourist Flow Based on Improved PSO Algorithm Optimized LSSVM Model

LU Wenxing, LI Chu   

  1. 1.School of Management, Hefei University of Technology, Hefei 230009, China
    2.Ministry of Education Engineering Research Center for Intelligent Decision-Making & Information Systems Technologies, Hefei 230009, China
  • Online:2019-09-15 Published:2019-09-11

摘要: 旅游客流量的准确预测为旅游目的地资源优化配置、景区战略计划制定提供有效依据。为了提高景区日客流量的预测精度,提出基于改进粒子群算法(Particle Swarm Optimization,PSO)优化最小二乘支持向量机(Least Squares Support Vector Machine,LSSVM)的预测方法,针对PSO算法的惯性权重在采取线性递减策略时不能满足粒子寻优非线性变化的缺陷,从种群中粒子的聚合程度以及种群进化中粒子适应度同惯性权重的关系出发,利用对数函数非线性变化的特性,提出基于对数函数的惯性权重自适应调整方法(Adaptive Logarithmic Particle Swarm Optimization,ALPSO)。通过改进的PSO算法优化LSSVM的参数,建立山岳型风景区日客流量的预测模型。以黄山风景区2012—2015年景区每日上山人数为例,实验结果证明,与基于标准PSO算法、正弦粒子群算法(Sinusoidal Particle Swarm Optimization,SPSO)和高斯粒子群算法(Gaussian Particle Swarm Optimization,GPSO)优化的LSSVM模型相比,ALPSO-LSSVM模型的预测性能更好,是准确预测景区日客流量的有效方法。

关键词: 粒子群算法, 惯性权重, 最小二乘支持向量机, 客流量预测

Abstract: The accurate prediction of tourist flow provides an effective basis for optimizing the allocation of tourism resources and the development of a strategic plan. In order to improve the forecasting accuracy of daily tourist flow, an improved prediction method of Least Squares Support Vector Machine(LSSVM) optimized by modified Particle Swarm Optimization(PSO) is proposed. In the improved PSO algorithm, considering the degree of aggregation of particles and the relationship between the value of particle fitness and inertia weight in population evolution, in the view of the nonlinear variation of the logarithmic function, an Adaptive Logarithmic Particle Swarm Optimization(ALPSO) is proposed since when the inertia weight adopts a typical linear decrement strategy can not satisfied the defect of non-linear variation characteristics of particle optimization. A prediction model of short-term tourist flow in mountainous scenic spot is established that the modified PSO algorithm is used to optimize the parameters of LSSVM. Given the daily tourist data of Mountain Huangshan from 2012 to 2015 as an example, the results of experiments corroborate that compared with those using LSSVM with standard PSO, LSSVM with Sinusoidal Particle Swarm Optimization(SPSO) and LSSVM with Gaussian Particle Swarm Optimization(GPSO), the ALPSO-LSSVM model presents better predictive performance which is an effective method for accurately forecasting daily tourist flow in scenic areas.

Key words: particle swarm optimization, inertia weight, least squares support sector machine, tourism flow forecasting