基于改进BP神经网络的价格预测模型研究  

Study on price prediction model based on improved BP neural network

作　　者：孙红敏[1] 吴静婷[1] 李晓明[1] 

机构地区：[1]东北农业大学电气与信息学院,哈尔滨150030

出　　处：《东北农业大学学报》2013年第8期133-137,共5页Journal of Northeast Agricultural University

基　　金：国家自然科学基金(70771033);黑龙江省研究生创新科研基金项目(YJSCX2012-007HLJ)

摘　　要：人工神经网络是一种人工智能算法,具有强大功能,可任意逼近非线性连续函数。面对畜产品价格变化的复杂因素,文章运用MATLAB实现各种BP神经网络的设计和训练,利用改进的神经网络算法即在权值中引入动量项,输入层至隐含层的传递函数采用S型曲线,隐含层至输出层的传递函数采用线性函数,对东北地区畜产品价格进行预测。结果显示模拟数据与实测数据拟合性很好、预测精度较高、泛化能力较好,可为畜产品价格预测提供一种全新思路和方法。Artificial neural network is an artificial intelligence algorithm which has powerful function, it can be arbitrary nonlinear continuous function approximation. According to the complex factors of animal products price changes, this paper uses MATLAB to realize all kinds of design and training of BP neural network, using the improved neural network algorithm by introducing the momentum in the weights of items, the transfer function of the input layer to hidden layer uses the S type curve, the transfer function of hidden layer to output layer uses linear function, livestock product price forecasting of northeast. Results show that the simulated data and experimental data fitting, high forecasting accuracy, generalization ability is higher. This study provides a new thought and method for animal products price prediction.

关 键 词：人工神经网络 畜产品价格 预测 

分 类 号：F323.7[经济管理—产业经济]