一种基于薄膜压力传感器和支持向量机的球磨机载煤量检测方法

发布时间：2019-04-21 06:15

【摘要】：磨煤机作为一种煤料粉碎设备,在矿山、冶金、火电、化工等工业行业中往往起到重要的辅助作用,磨煤机的工作效率和这些行业总体的生产效率有着密切的关系。然而,磨煤机存在多变量、强耦合、大惯性等无法回避的特性,若处于不良工况或磨损较大的情况可能会影响磨煤机系统的正常运行,从而提高发生煤粉堵塞、高温超限等多种故障的可能性。在研究提高球磨机工作效率和安全运行的方向中,如何获得当前工作状态下的载煤量是关键性的研究问题,也长期处在该领域的热点研究地位。若能及时、准确地获得当前载煤量,将有效提高对球磨机的自动控制水平、提高运行效率,对行业产生可观的经济效益。论文对于球磨机载煤量的检测,提出了一种根据钢球表面受压缩应力与载煤量关系的新思路,并围绕该思路进行了理论分析、模拟验证、试验论证等环节,探究了该思路的可行性和有效的方法。通过对球磨机运动理论中“抛落式”和“泄落式”运动状态的分析,本文提出对于不同运动状态,钢球表面受压缩应力不同,故可以通过持续检测该信号的数值来对磨煤机当前载煤量进行分类识别。为验证这一理论,建立了球磨机-钢球-煤料模型。在球磨机和设计上,使用Solidworks建立三维模型文件并导入EDEM软件中,通过EDEM特有的“粒子工厂”功能产生钢球和煤料模型,并导入相应的物理参数,进行软件模拟计算。在试验环节,首先设计并制造一个类似球磨机钢球大小的近似球体作为设备外壳,外壳的样式和薄膜压力传感器相适应,由Auto CAD 3D软件绘制完成后交由3D打印厂家完成,之后在其部分表面覆盖一种FSR薄膜压力传感器、在外壳内部设置Arduino电路板和蓝牙无线传感装置,通过无线信号将数据传输到外部计算机中。之后,在软件模拟测试环节设计的球磨机模型文件基础上改造了球磨机模型筒体造型、计算了电机功率、装配元件并装配完成。在数据处理环节,论文使用目前应用广泛的SVM支持向量机人工神经网络作为数据分析的工具,将每一时刻的FSR压力传感器数据作为数据,载煤量作为识别结果形成一个样本,经训练后进行分类识别,取得了较显著的试验成果,证明该方法有可行性,对球磨机相关设计与研究有一定参考意义。
[Abstract]:As a kind of coal comminution equipment, coal mill often plays an important auxiliary role in mining, metallurgy, thermal power, chemical industry and so on. The working efficiency of coal mill is closely related to the overall production efficiency of these industries. However, the coal mill has many characteristics, such as multivariable, strong coupling, large inertia and so on. If it is in bad working condition or wear and tear condition, it may affect the normal operation of coal mill system, so as to improve the occurrence of pulverized coal blockage. The possibility of many faults such as high temperature exceeding the limit. In the direction of improving the working efficiency and safe operation of ball mill, how to obtain the coal carrying capacity under the current working condition is the key research problem, and it is also in the hot research position in this field for a long time. If the current load of coal can be obtained in time and accurately, the automatic control level of ball mill will be improved effectively, the running efficiency will be improved, and considerable economic benefits will be produced to the industry. In this paper, a new idea based on the relation between the compressive stress on the surface of steel ball and the coal carrying capacity is put forward, and the theoretical analysis, simulation verification, test demonstration and so on are carried out around the coal carrying capacity of the ball mill, and some other links, such as theoretical analysis, simulation verification and experimental demonstration, are put forward. The feasibility and effective methods of this idea are explored. Based on the analysis of the "drop" and "release" motion states in the motion theory of ball mill, this paper puts forward that for different motion states, the compressive stress on the surface of steel ball is different. Therefore, the current coal carrying capacity of the coal mill can be classified and recognized by continuously detecting the value of the signal. In order to verify this theory, a ball mill-steel ball-coal model was established. In ball mill and design, the three-dimensional model file is built by Solidworks and imported into EDEM software. The steel ball and coal model are generated by the special "particle factory" function of EDEM, and the corresponding physical parameters are imported to carry on the software simulation calculation. In the test process, a ball similar to the ball mill steel ball size is designed and manufactured as the shell of the equipment. The style of the shell adapts to the thin film pressure sensor. After the drawing by Auto CAD 3D software, it is completed by the 3D printer. Then a FSR thin film pressure sensor is covered on the part of its surface. The Arduino circuit board and Bluetooth wireless sensing device are arranged inside the housing to transmit the data to the external computer by wireless signal. Then, on the basis of the model file of the ball mill designed by the software simulation and test link, the tube shape of the ball mill model is modified, the power of the motor is calculated, the components are assembled and the assembly is completed. In the process of data processing, the SVM support vector machine artificial neural network, which is widely used at present, is used as the tool of data analysis. The data of FSR pressure sensor at every moment is taken as data, and the coal load is used as the identification result to form a sample. After training, the classification and recognition has been carried out, and the remarkable experimental results have been obtained. The method is proved to be feasible and has some reference significance for the design and research of ball mill.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD453

【参考文献】