作　　者： ; ; ; ; (邹振裕）; (李展峰）; (罗永恒）;

机构地区： 华南理工大学自动化科学与工程学院

出　　处： 《工业仪表与自动化装置》 2010年第6期8-10,30,共4页

摘　　要： 通过分析某水厂现有的投药控制,基于机理法及神经网络建模法建立了混凝剂投加量的2种数学模型,并对2种模型进行了检验和比较。结果表明:2种模型都较为准确有效,但机理建模能较直观清晰地反映具体的数学解析式以及变量之间的关系,为该项目的整体优化提供条件;而神经网络建模精度则相对较高,能对水厂投药量作出及时准确预测,为操作工人提供操作指导。实际生产中可根据需要将2种模型结合运用,具有较好的实际意义。 By analyzing the dosage control of a waterworks,two different mathematical models of the coagulant dosage have been established by means of the mechanism method and BP neural network,and the results of two models have been verified and compared.The experimental results show that the two kinds of models are relatively accurate.Mechanism modeling can more directly reflect the specific mathematical formulas and the relationship between variables clearly,provide condition for the whole optimization of this project.But the accuracy of neural network modeling is much higher,accurately obtains the prediction of the coagulant dosage and provides instruction for operators.According to the need of the actual production,both models can be combined;it will have a better practical significance.

关 键 词： 机理法 差分进化算法 神经网络建模 数学模型 混凝剂投加量

领　　域： [自动化与计算机技术] [自动化与计算机技术]