论文部分内容阅读
针对自来水厂絮凝剂投加控制问题,提出一套基于矾花数字图像处理的加矾控制决策系统。设计以500万像素工业相机为传感器和以PC为主控器的水下图像采集装置,连续采集水厂处理池中矾花图像,以MATLAB为图像处理编程工具,实时分析矾花形态和分布,比较矾花数字图像各种特征对加矾量与水质关系的表征能力,经大样本试验数据训练,得出SVM算法、BP神经网络和GRNN神经网络相结合的分类器,可用于自动监测水厂净水处理过程中的矾花状态,评判水质参数,实时调节加矾量,减少絮凝过程中检测絮凝效果的时滞。
Aiming at the problem of flocculant dosing control in waterworks, a set of control decision-making system based on alum digital image processing is proposed. A 5-megapixel industrial camera is designed as a sensor and a PC-based underwater image acquisition device is designed to continuously collect the alum images in the water treatment tank of a water treatment plant. MATLAB is used as an image processing programming tool to analyze the morphology and distribution of alum in real time, By comparing the characterization of various characteristics of alum digital images on the relationship between the amount of alum and water quality, the SVM algorithm, BP neural network and GRNN neural network are combined to obtain the characterization ability of large amount of experimental data, which can be used to automatically monitor the water plant The state of alum in the process of water purification, judging the water quality parameters, adjusting the amount of alum in real time and reducing the time delay for detecting the flocculation effect during the flocculation process.