基于混沌RBF神经网络的电商市场销量预测模型技术研究  

Research on E-commerce platform sales forecast model based on chaotic RBF neural network

作　　者：刘电威[1] 牛龙龙 LIU Dianwei;NIU Longlong(School of Business,Guangdong Polytechnic of Science and Technology,Zhuhai 519090,China;School of Mathematics and Computer Science,Xiangtan University,Xiangtan 411105,China)

机构地区：[1]广东科学技术职业学院商学院,广东珠海519090 [2]湘潭大学数学与计算科学学院,湖南湘潭411105

出　　处：《湘潭大学学报（自然科学版）》2023年第5期57-64,共8页Journal of Xiangtan University(Natural Science Edition)

基　　金：广东省科技厅项目(KTP20200144);广东省哲学社会科学规划项目(GD20XJY37)。

摘　　要：商品销量受多种因素影响,这种影响难以用公式准确表达.为了获得更精准的预测结果,针对商品销量预测的高度复杂性和非线性,提出了基于混沌RBF神经网络的商品销量预测模型.该模型基于电商市场商品销售历史数据,使用李雅普诺夫指数算法评估商品销售数据混沌序列特性,并通过对数据库相结构的空间优化重构,利用RBF神经网络技术对优化重构后的数据进行训练归纳.同时,通过混沌方法计算网络的连接权值和高斯函数径向基中心,实现了RBF神经网络的优化.在Matlab平台上进行了仿真实验,结果表明,该混沌优化RBF神经网络商品销售预测模型具有较高的精度和较快的速度.There are numerous factors that affect the sales of products,and expressing this effect accurately through formulas is challenging.To overcome the high complexity and nonlinearity of commodity sales forecasting,a model based on a chaotic RBF neural network has been proposed to obtain more precise results.Firstly,the Lyapunov exponent is calculated on historical data to determine the chaotic characteristics of the product sales data series.Then,the historical data is reconstructed in phase space,and the RBF neural network is trained using the reconstructed data.The chaotic method is utilized to calculate the connection weight of the network and the radial basis center of the Gaussian function to optimize the RBF neural network.The model is simulated using the Matlab platform,and the experimental results demonstrate that the chaotic optimization RBF neural network product sales prediction model has higher accuracy and faster speed.

关 键 词：混沌 RBF神经网络 商品销量 预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] F724.6[自动化与计算机技术—控制科学与工程]