动态粒子群算法优化灰色神经网络的旅游需求预测模型研究

Research on Tourism Demand Forecasting Model Based on Gray Neural Network Optimized by Dynamic PSO

在已有研究成果的基础上,本文提出一种经过优化的灰色神经网络模型用于旅游需求预测。首先,将灰色神经网络进行数学建模,以便于优化算法的应用。其次,在标准粒子群算法基础上,通过对惯性权重与学习因子的动态设置,构造一种动态粒子群算法,并利用动态粒子群算法对灰色神经网络参数进行优化。最后,通过优化的灰色神经网络预测模型对旅游需求进行预测,并比较了本文所建立的灰色神经网络模型与目前流行的旅游需求预测模型的预测效果。结果表明,动态粒子群算法优化的灰色神经网络的预测精度相比已有方法具有更高的预测精度。

On the basis of existing researches, aiming at the problems of low efficiency and local optimum and retardation of parameter modification during grey neural network evolution process, this paper establishes a new grey neural network model based on dynamic PSO. Firstly, this paper proposes a mathematical grey neural network. Secondly, through the dynamic setting of inertia weight and learning factor, this paper proposes a dynamic PSO based on standard PSO. And then this paper establishes a new grey neutral network by optimization of neural network parameter using dynamic PSO. Finally, this paper compares the prediction accuracy of BP neural network, grey neural network, ARIMA-BP neural network and dynamic PSO grey neural network. The simulation results show that the new grey neural network based on dynamic PSO has better forecast performance.