基于假设检验和专家系统的神经网络预测

根据硅渣形成的机理及数据特点，利用统计分析的假设检验来增加数据的可靠性，并结合长期积累的专家经验知识以及采用产生式规则知识的表示形式，建立了基于主元分析的数据分类专家系统。然后针对硅渣机理的特点，构造出神经网络结构来预测硅渣的准确值。最后将该系统用计算机实现仿真实验，仿真结果表明，本文提出的基于假设检验和专家系统数据预处理的神经网络预测方法是一种可行的数据预测策略。

A neural network of hypothesis testing and expert system

According to the formative mechanism of white residue and the feature of data, the hypothesis testing method of statistical analysis is adopted to increase the data reliability. The knowledge representation of production is introduced into the system, and the expert system of data classification is built based on the principal component analysis and the accumulation of the empirical knowledge. Then, the neural network is designed to predict the accurate value of the white residue. At last, it is simulated by the computer. The simulation result shows that the neural network of hypothesis testing and expert system is a feasible method. It proved that the strategy of the neural network of hypothesis testing and expert system for data prediction is pratical.