论文部分内容阅读
本文利用数据库技术,针对Y=F(X)数学模型,编写集模型管理、数据样本管理和建模、预报为一体的通用性处理软件。软件自动适应不同分析体系中,自变量数和目标函数的变化,并支持在模型中对每个自变量进行定义,及其物理意义的显示。软件中提供了MLR、PCR、PLS、神经网络和遗传算法-PLS联用5种建模方法,并将交叉验证方法嵌入PCR、PLS方法中,用交叉验证法检验样本中的奇异样本点。利用模拟数据对软件进行了检验。
In this paper, database technology, for Y = F (X) mathematical model, write set of model management, data sample management and modeling, forecasting as one of the general processing software. The software automatically adapts to changes in the number of independent variables and the objective function in different analysis systems and supports the definition of each independent variable in the model and the display of its physical meaning. Five modeling methods, MLR, PCR, PLS, neural network and genetic algorithm-PLS, are provided in the software. The cross validation method is embedded in PCR and PLS methods, and the singular sample points in the sample are tested by cross validation method. The software was tested using simulation data.