摘要
分析了影响我国钢材价格的客观因素,基于BP神经网络建立钢材价格预测的模型。采用Levenberg-Marquardt算法对BP神经网络的权值进行优化。使用MATLAB语言编写程序,用1990-2008年的数据对模型进行训练得出预测结果。结果表明,预测值与真实值较吻合,所建立的神经网络模型有较准确的预测精度。
This paper made a detailed analysis of the objective factors that influencing our country's steel prices,and built a prediction model of steel prices based on BP neural network.After that,the weights of BP neural network were optimized by Levenberg-Marquardt algorithm.Then programmed in MATLAB language and obtained a result of the prediction by using data from 1990-2008.The results show that the predicted values are good agreement with the true values,and the established neural network model has a accurate precision in prediction.
出处
《自动化与仪表》
北大核心
2010年第12期7-10,共4页
Automation & Instrumentation