基于智能加药的煤泥浮选控制系统研究

发布时间：2018-09-02 08:40

【摘要】：浮选加药作为煤泥浮选过程的重要环节,其药剂量的准确添加对浮选的效果有很大的影响。目前煤泥浮选生产中由于缺少高精度的浮选变量检测设备和有效的加药策略,加药量的多少往往只能凭借工人的生产经验去添加。这种加药的方法不仅加重了工人的劳动强度,而且还会因为主观因素造成浮选精煤灰分的不稳定,从而影响了生产的效益。为了解决上述问题,本文在对赵楼选煤厂煤泥浮选工艺深入分析的基础上,运用智能加药技术对煤泥浮选控制系统进行了研究。制定了以浮选精煤灰分作为系统的主要被控量,以向矿浆预处理器中添加的浮选药剂量作为系统的控制量,以图像识别技术和专家系统作为监督手段的总体方案。首先,通过对该选煤厂煤泥浮选工艺过程和实际的生产记录数据进行了研究分析,确立了对浮选加药的重要过程参数;在此基础上构建了基于GRNN神经网络的浮选加药预测模型,并用交叉验证搜索算法对该神经网络模型的参数进行求解,通过与BP、GA-BP预测模拟的对比,得出其在浮选加药方面的优越性。然后,在对泡沫图像去噪声处理的基础上,通过MIV值评价法对影响精煤灰分的泡沫特征进行了筛选,建立了基于RBF神经网络的浮选精煤灰分预测模型,并用递归最小二乘法对其模型参数进行求解,进而通过与BP网络的比较,得到其在浮选精煤灰分预测方面的优越性。再次,在对浮选精煤灰分预测和泡沫图像特征提取的基础上,建立以专家系统为核心的浮选加药量矫正模型,对浮选预测加药量进行在线修正。最后,在以上研究的基础上,搭建了系统的软硬件模型,并对系统做了仿真运行,验证了其研究的可行性。
[Abstract]:As an important link in the process of coal slime flotation, the accurate addition of flotation agents has a great influence on the flotation effect. At present, due to the lack of high precision flotation variable detection equipment and effective dosing strategy, the dosage of coal slime floatation can only be added by the workers' production experience. This method not only increases the labor intensity of workers, but also results in the instability of ash content of flotation coal due to subjective factors, thus affecting the efficiency of production. In order to solve the above problems, based on the deep analysis of coal slime flotation technology in Zhaolou Coal preparation Plant, this paper studies the slime flotation control system with intelligent dosing technology. The total scheme of using flotation concentrate ash as the main controlled quantity of the system, the amount of flotation reagents added to the pulp preprocessor as the control quantity of the system, the image recognition technology and the expert system as the monitoring means were made. First of all, through the research and analysis of the process of coal slime flotation and the actual production record data, the important process parameters of flotation dosing are established, and the prediction model of floatation dosing based on GRNN neural network is constructed. The parameters of the neural network model are solved by cross-validation search algorithm. Compared with the BP,GA-BP prediction simulation, the superiority of the neural network model in floatation is obtained. Then, on the basis of de-noising processing of foam image, the foam characteristics affecting ash content of clean coal are screened by MIV value evaluation method, and the prediction model of ash content of flotation concentrate coal based on RBF neural network is established. The model parameters are solved by recursive least square method, and its superiority in ash prediction of flotation coal is obtained by comparing with BP network. Thirdly, on the basis of prediction of ash content and feature extraction of foam image, a correction model of flotation dosage is established, which is based on expert system. Finally, on the basis of the above research, the software and hardware models of the system are built, and the simulation of the system is carried out to verify the feasibility of the research.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD94;TD923

【参考文献】