摘要
论述了自来水投药混凝沉淀过程机理,并根据历史数据分析了影响混凝沉淀过程的主要影响因素。叙述了采集数据预处理过程中常用的数字滤波方法,并根据浊度仪、pH计此二仪表的独特特征,提出了改进的多限幅滤波加滑动平均滤波的复合方法。同时还对不同性质的变量的数据归一化方法进行了论述,提出了全数据范围的改进数据归一化方法。文中通过具体的仿真实验运行证明,上述多限幅滤波加滑动平均滤波的复合方法与改进数据归一化方法是正确的和可行的。这对于混凝沉淀过程的神经网络建模研究提供了真实的样本数据,同时对于其他过程采集数据预处理的方法具有重要的参考价值。
The paper discussed the mechanism of coagulation and sedimentation process in detail,and based on historical data analysis affect the coagulation process of the main factors. The paper describes some digital filtering methods that com- monly used in the process of collecting data preconditioning. Based on the unique characteristics of Turbidity meter and pH Meter,an improved composite method of multi-limiter filter plus moving average filter was proposed.This paper discusses the data normalization method of variables with different nature,and put forward an improved data normalization method of full data range. Through the simulation experiment,it proves that the above-mentioned composite method is correct and feasible. So we can provides a real sample data for the neural network modeling of coagulation and sedimentation process, and this also has important reference value for other data pretreatment methods of the collecting data.
出处
《长春理工大学学报（自然科学版）》
2010年第1期102-107,共6页
Journal of Changchun University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(60704012)
广东省科技攻关资助项目(2005B10201005)
关键词
混凝沉淀
数字滤波
数据归一化
神经网络
coagulating sedimentation system
digital filter
data normalization
neural network
