提高旅游风景区日客流量的预测精度,对旅游风景区的日常运营管理和 旅游资源的保护有重要意义. PSO-BP被广泛应用于预测中,针对PSO算法的惯性权重采 取线性动态变化时无法满足粒子多样性和易陷入局部极值等缺陷,文章提出一种利用改 进后的PSO-BP方法,利用粒子适应度值对惯性权重进行动态非线性变化,同时结合粒子 迭代周期增加位置扰动,对粒子群算法进行改进. 将改进后的PAPSO算法(particle swarm optimization algorithm with position disturbance and adaptive inertia weight, PAPSO)对BP神经网络 的初始权值和阈值进行优化,建立黄山风景区日客流量的Matlab预测模型,对黄山旅游客流量数 据进行实验,结果表明文章提出的基于PAPSO算法优化BP神经网络的预测模型有效地提升了预测精度.

Improving the forecasting accuracy of daily passenger traffic in scenic spots is significant to the daily operation management of scenic spots and the conservation of tourism resources. PSO-BP has been applied widely to the prediction. Based on the fact that the inertia weight of PSO algorithm cannot meet the requirements of particle diversity and local extremum when it changes linearly, this paper proposes to use the fitness value of particles to change the inertia weight dynamically and nonlinearly. In this paper, the particle fitness value is used to change the inertia weight by the improved method of PSO-BP. Meanwhile, the particle swarm optimization algorithm is improved by adding position perturbation in the iteration period of particles and combining with the particle iteration period to increase the position disturbance. The improved PAPSO (particle swarm optimization algorithms with position disturbance and adaptive inertia weight, PAPSO) is used to optimize the initial weights and thresholds of BP neural network. The researchers established a Matlab prediction model of the daily passenger flow in Huangshan scenic area and conducted an experiment on the tourist flow data of Huangshan. The results show that the prediction model based on PAPSO algorithm to optimize BP neural network improves the prediction accuracy effectively.