应用神经网络专家系统预测浅层地下水水质  

Application of Neural Network Expert System on Predicting the Quality of Sub-groundwater

作　　者：李云先[1] 

机构地区：[1]上海工程技术大学艺术设计学院,上海200336

出　　处：《淮南职业技术学院学报》2005年第2期92-95,共4页Journal of Huainan Vocational Technical College

摘　　要：在系统介绍神经网络专家系统基本原理和工作步骤的基础上,利用开发出的神经网络专家系统,总结分析出影响淮南市区浅层地下水水质的主要参数,并通过B-P网络对大量实例的学习,自动生成符合实际的专家知识库;对实际问题,只要输入区块的特征参数,系统就能由推理机根据专家知识库给出预测区的水质,权值不受人因素为干扰.并以淮南市区16已知水样点的评价参数及水质级别为学习样本,对6预测水样点的水质进行了预测,结果表明效果良好;从而指出运用神经网络专家系统可以克服指数法和聚类法等不易准确确定参数的隶属度和权重分配等局限性,保证了预测结果的精确度和可靠性;只要测得预测目标的特征参数后,输入系统,很快就能得出预测结果,从而提高了预测速度和工作效率。After introducing the fundamental of the neural network expert system and the steps of how it works in the paper, then by making use of the developed neural network system to summarize the primary factors influencing on the quality of sub-ground water in urban area of Huainan. After studying many examples, the system can create the corresponding expert repository automatically. It only needs us to input the characteristic parameters to get the quality of sub-groundwater in the predicting area. The weight distribution of parameters is not affected by artificial factors. To investigate the effectiveness of our system, we predicted the quality of six predicting points successfully by taking the parameters of sixteen known sub-area as study samples. From this, the paper indicts that the neural network expert system can overcome the difficulty of regular statistical methods and fuzzy methods to accurately decide the subordination and the weight distribution of the parameters, so the system can ensure the precision and liability of the forecast. Because the neural network expert system creates the depository after learns the study samples, it can give the predicting results rapidly after entering characteristic parameters that enhances the work effectiveness.

关 键 词：神经网络专家系统 浅层地下水水质 B-P算法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TD163[自动化与计算机技术—控制科学与工程]