CN101493354A

CN101493354A - Material level detecting method for tube mill based on multi-sensor fusing technology

Info

Publication number: CN101493354A
Application number: CNA2009100738626A
Authority: CN
Inventors: 阎高伟; 陈泽华; 谢珺; 续欣莹; 谢刚; 谢克明
Original assignee: Taiyuan University of Technology
Current assignee: Taiyuan University of Technology
Priority date: 2009-03-01
Filing date: 2009-03-01
Publication date: 2009-07-29
Anticipated expiration: 2029-03-01
Also published as: CN101493354B

Abstract

The invention relates to a method for detecting the material level of a drum type ball mill, in particular to a method for detecting the material level of a drum type ball mill based on multi-sensor technology. It solves the problems of low detection accuracy in the existing detection method of the material level of the drum type ball mill. The steps of the method are as follows: (1) respectively collect the noise signal s of the drum of the drum type ball mill, the place where the background noise is relatively high, and the environment. ₁ , s ₂ , s ₃ ; (2) the noise signals s ₁ , s ₂ , s ₃ are processed by conditioning, analog-to-digital conversion, and pre-whitening to obtain signals w ₁ , w ₂ , w ₃ ; (3) signal w ₁ , w ₂ , w ₃ are processed by blind source separation to obtain signals x ₁ , x ₂ , x ₃ ; (4) From the signals x ₁ , x ₂ , x ₃ , identify the signal representing the material level information of the roller ball mill, and Discrete Fourier transform, modulus, and normalization are performed on it to obtain an output signal z(n); (5) According to the output signal z(n), calculate the value of the material level L in the roller ball mill. The method of the invention is reasonable, can effectively reduce the interference of environmental noise and self-background noise, and can accurately and reliably detect the material level of the drum type ball mill.

Description

Material level detection method of roller ball mill based on multi-sensor fusion technology

技术领域 technical field

本发明涉及滚筒式球磨机料位的检测方法，具体是一种基于多传感器融合技术的滚筒式球磨机料位检测方法。The invention relates to a method for detecting the material level of a drum type ball mill, in particular to a method for detecting the material level of a drum type ball mill based on multi-sensor fusion technology.

背景技术 Background technique

滚筒式球磨机是一种工业生产中十分重要的制粉设备，它广泛应用于电力、矿山、冶金、化工和建材等行业，主要作用是将原料或半成品磨制成粉，以供后续工序使用。例如，在火力发电中，原煤必须经过球磨机充分的研磨以达到一定的细度，才能进入炉膛燃烧；在水泥生产过程中，滚筒式球磨机还是生料粉磨和水泥粉磨系统的主要组成设备。Roller ball mill is a very important powder-making equipment in industrial production. It is widely used in electric power, mining, metallurgy, chemical industry, building materials and other industries. Its main function is to grind raw materials or semi-finished products into powder for subsequent processes. For example, in thermal power generation, the raw coal must be fully ground by a ball mill to achieve a certain fineness before it can enter the furnace for combustion; in the cement production process, the drum ball mill is also the main component of raw material grinding and cement grinding systems.

滚筒式球磨机属于一种高能耗、低效率的设备，每天消耗着大量的电能。滚筒式球磨机利用低速回转的滚筒将作为研磨体的大量钢球带到一定高度，然后大量钢球抛射下落，将物料击碎。滚筒式球磨机所装载的钢球和物料的重量比往往达到十比一乃至更高，因此大量的能源被球磨机自身消耗。据统计，在水泥厂，70％以上的用电量是消耗在磨机上，而在磨机所消耗的电能中只有不到10％的能量被真正用于原料的磨碎，而90％以上的能量被转化为振动、发热和噪声。而在火力发电厂，滚筒式球磨机的能源消耗占到厂用电的三分之一。由于滚筒式球磨机工作在旋转状态下，一直以来缺少直接检测其内部料位的设备，经常发生饱磨、空磨和堵磨等现象，不得不经常停料或停磨，造成生产效率降低，还影响产量和质量的提高，同时对设备的寿命也会产生不良后果。据不完全统计，全国现有近三万台大型球磨机应用在电力、石化、冶金、化工、矿山、水泥等各种行业。根据试验研究，球磨机都有至少10％以上的节能潜力和节约9％以上钢材原材料的潜力，对这些潜力的挖掘需要能够准确测量滚筒式球磨机的料位，并进行有效控制。因此，可靠测量滚筒式球磨机料位，并控制滚筒式球磨机稳定运行在最佳工作状态，对于提高其运行的稳定性和经济性和节能降耗有着重要的意义。The roller ball mill is a kind of equipment with high energy consumption and low efficiency, which consumes a lot of electric energy every day. The drum ball mill uses a low-speed rotating drum to bring a large number of steel balls as grinding bodies to a certain height, and then a large number of steel balls are thrown down to crush the materials. The weight ratio of the steel balls and materials loaded in the drum ball mill often reaches ten to one or even higher, so a large amount of energy is consumed by the ball mill itself. According to statistics, in cement plants, more than 70% of the electricity consumption is consumed on the mill, and less than 10% of the energy consumed by the mill is actually used for grinding raw materials, while more than 90% Energy is converted into vibration, heat and noise. In a thermal power plant, the energy consumption of the roller ball mill accounts for one-third of the power consumption of the plant. Since the roller ball mill works in a rotating state, there has always been a lack of equipment to directly detect its internal material level, and phenomena such as full grinding, empty grinding, and blocked grinding often occur, and the feeding or grinding has to be stopped frequently, resulting in reduced production efficiency. It affects the improvement of output and quality, and also has adverse consequences on the life of the equipment. According to incomplete statistics, there are nearly 30,000 large-scale ball mills used in various industries such as electric power, petrochemical, metallurgy, chemical industry, mining, and cement in the country. According to experimental research, ball mills have at least 10% energy-saving potential and the potential to save more than 9% steel raw materials. The excavation of these potentials requires accurate measurement of the material level of the roller ball mill and effective control. Therefore, it is of great significance to reliably measure the material level of the drum type ball mill and control the stable operation of the drum type ball mill in the best working state for improving its operation stability, economy, energy saving and consumption reduction.

目前，对滚筒式球磨机料位的检测方法主要有差压法、功率法、测振法和测音法。1、差压法利用球磨机出入口的差压来表示球磨机内部的料位。虽然在理论上可以用差压表示存料量，但由于球磨机出入口差压不仅与球磨机的存料量有关，还受球磨机的钢球装载量和通风量等因素的影响。因此，在实际的使用中差压法并不能准确地反映球磨机的料位信息。2、功率法通过检测球磨机运行时电机电流的变化来间接反映存料量的一种方法，由于球磨机的电机电流主要受钢球装载量的影响，空载与满载时磨电流的变化不大，而且不是单值关系，因此该方法的精度不高。3、测振法通过钢球磨煤机运行过程中振动的频率和幅值来确定存料量，但是该方法滞后较大，容易受环境中其它设备的振动影响，此外由于传感器直接与振动的球磨机机体接触使传感器容易损坏，维护成本高，不利于系统的连续运行。4、测音法俗称电耳法，它利用球磨机运行过程中钢球与球磨机内壁的衬板以及钢球之间碰撞产生的声音信号来反映球磨机的料位。当存煤量少时，钢球之间以及钢球与衬板之间的撞击比较剧烈，声音比较大，声音随存煤量的增加而减小。由于是非接触测量，该方法获得了较为广泛的应用。但是，该方法也存在缺陷，主要有如下几方面：第一，相邻球磨机的噪音会对测量产生较大的影响；第二，在高料位运行时，由于球磨机内部物料填充度高，钢球碰撞产生的噪音幅度较低，而球磨机的运转部分和电机传动机构等设备的背景噪声相对较高，造成高料位测量灵敏度低，不能准确反映料位。At present, the detection methods for the material level of the roller ball mill mainly include differential pressure method, power method, vibration measurement method and sound measurement method. 1. The differential pressure method uses the differential pressure at the inlet and outlet of the ball mill to indicate the material level inside the ball mill. Although in theory, the differential pressure can be used to represent the storage capacity, but the differential pressure at the entrance and exit of the ball mill is not only related to the storage capacity of the ball mill, but also affected by factors such as the steel ball loading and ventilation of the ball mill. Therefore, in actual use, the differential pressure method cannot accurately reflect the material level information of the ball mill. 2. The power method is a method to indirectly reflect the amount of material stored by detecting the change of the motor current when the ball mill is running. Since the motor current of the ball mill is mainly affected by the load of steel balls, the change of the grinding current between no-load and full-load is not large. And it's not a single-valued relationship, so the method doesn't have much precision. 3. The vibration measurement method determines the stock volume through the vibration frequency and amplitude during the operation of the steel ball coal mill, but this method has a large lag and is easily affected by the vibration of other equipment in the environment. In addition, because the sensor is directly connected to the vibrating ball mill The contact of the body makes the sensor easy to damage, the maintenance cost is high, and it is not conducive to the continuous operation of the system. 4. The sound measurement method is commonly known as the electric ear method. It uses the sound signal generated by the collision between the steel ball and the inner wall of the ball mill and the steel ball during the operation of the ball mill to reflect the material level of the ball mill. When the amount of coal stored is small, the impact between the steel balls and between the steel balls and the liner is relatively severe, and the sound is relatively loud, and the sound decreases with the increase of the amount of coal stored. Due to the non-contact measurement, this method has been widely used. However, this method also has defects, mainly in the following aspects: first, the noise of the adjacent ball mill will have a great impact on the measurement; The noise generated by ball collision is relatively low, while the background noise of the running part of the ball mill and the motor transmission mechanism is relatively high, resulting in low sensitivity of high material level measurement, which cannot accurately reflect the material level.

发明内容 Contents of the invention

本发明为了解决滚筒式球磨机料位的现有检测方法存在不能准确反映球磨机料位信息、检测精度低、成本高、测量灵敏度低等问题，提供了一种基于多传感器融合技术的滚筒式球磨机料位检测方法。In order to solve the problems that the existing detection method of the material level of the drum type ball mill cannot accurately reflect the information of the material level of the ball mill, the detection accuracy is low, the cost is high, and the measurement sensitivity is low, the invention provides a drum type ball mill material based on multi-sensor fusion technology. bit detection method.

本发明是采用如下技术方案实现的：基于多传感器融合技术的滚筒式球磨机料位检测方法，步骤如下：The present invention is realized by adopting the following technical scheme: a drum type ball mill material level detection method based on multi-sensor fusion technology, the steps are as follows:

(1)、同时应用三个音频传感器分别采集滚筒式球磨机滚筒处的噪声信号s₁、滚筒式球磨机本身背景噪声相对较高处(一般指机体运转部分及电机传动机构的位置)的噪声信号s₂、以及滚筒式球磨机所处环境的噪声信号s₃；(1) Simultaneously use three audio sensors to collect the noise signal _s at the drum of the drum ball mill respectively. ₂ , and the noise signal s ₃ of the environment where the roller ball mill is located;

(2)、对步骤(1)采集到的噪声信号s₁，s₂，s₃进行调理得信号u₁，u₂，u₃，将信号u₁，u₂，u₃模数转换得数字序列信号v₁，v₂，v₃，然后将数字序列信号v₁，v₂，v₃预白化处理得信号w₁，w₂，w₃，其中，对噪声信号s₁，s₂，s₃调理后得到的信号u₁，u₂，u₃要适合于进行模数转换采用的A/D转换器的输入量程范围；(2) Condition the noise signals s ₁ , s ₂ , and s ₃ collected in step (1) to obtain signals u ₁ , u ₂ , and u ₃ , and convert the signals u ₁ , u ₂ , and u ₃ into digital signals Sequence signals v ₁ , v ₂ , v ₃ , and then pre-whiten the digital sequence signals v ₁ , v ₂ , v ₃ to obtain signals w ₁ , w ₂ , w ₃ , where, for noise signals s ₁ , s ₂ , s ₃ The signals u ₁ , u ₂ , and u ₃ obtained after conditioning should be suitable for the input range of the A/D converter used for analog-to-digital conversion;

(3)、分别对步骤(2)得到的信号w₁，w₂，w₃进行盲源分离处理得三个在统计上独立的信号x₁，x₂，x₃；(3), respectively performing blind source separation on the signals w ₁ , w ₂ , and w ₃ obtained in step (2) to obtain three statistically independent signals x ₁ , x ₂ , x ₃ ;

(4)、由于盲源分离处理的求解结果有不确定性，主要表现为盲源分离后信号矢量的排列位置的不确定、信号的幅值和初始相位的变化；因此需从信号x₁，x₂，x₃中甄别出代表滚筒式球磨机料位信息的信号；甄别方法如下：a、对与滚筒式球磨机滚筒处噪声信号s₁对应的经步骤2处理后得到的信号w₁，以及经步骤(3)得到的信号x₁，x₂，x₃进行频谱分析，按下式对信号w₁，x₁，x₂，x₃进行N点离散傅立叶变换，(4) Due to the uncertainty of the solution result of blind source separation, it is mainly manifested in the uncertainty of the arrangement position of the signal vector after blind source separation, the change of the amplitude and initial phase of the signal; therefore, it is necessary to start from the signal x ₁ , The signal representing the material level information of the roller ball mill is screened out from x ₂ and x ₃ ; the screening method is as follows: a. For the signal w ₁ obtained after processing in step 2 corresponding to the noise signal s ₁ at the drum of the drum ball mill, and Spectrum analysis is performed on the signals x ₁ , x ₂ , and x ₃ obtained in step (3), and N-point discrete Fourier transform is performed on the signals w ₁ , x ₁ , x ₂ , and x ₃ according to the following formula,

${F f}_{w w} ((n no)) = = {Σ Σ}_{k k = = 00}^{N N - - 11} {w w}_{11} ((k k)) {e e}^{- - j j 22 πnk πnk / / N N},, n no = = 0,1,2 0,1,2,, . . . . . .,, N N - - 11 - - - - - - ((11))$

${F f}_{i i} ((n no)) = = {Σ Σ}_{k k = = 00}^{N N - - 11} {x x}_{i i} ((k k)) {e e}^{- - j j 22 πnk πnk / / N N},, n no = = 0,1,2 0,1,2,, . . . . . .,, N N - - 11,, i i = = 1,2,3 1,2,3 - - - - - - ((22))$

b、按下式对信号w₁，x₁，x₂，x₃的离散傅立叶变换结果取模，并归一化处理，b. Take the modulus of the discrete Fourier transform results of the signals w ₁ , x ₁ , x ₂ , and x ₃ according to the following formula, and normalize them,

${A A}_{w w} ((n no)) = = \frac{| | {F f}_{w w} ((n no)) | |}{max max ((| | {F f}_{w w} ((n no)) | |))} - - - - - - ((33))$

${A A}_{i i} ((n no)) = = \frac{| | {F f}_{i i} ((n no)) | |}{max max (({| | F f}_{i i} ((n no)) | |))},, i i = = 1,2,3 1,2,3 - - - - - - ((44))$

c、按下式分别求出A_i(n)，i＝1，2，3与A_w(n)的欧氏距离D_i，c. Calculate the Euclidean distance D _i between A _i (n), i=1, 2, 3 and A _w (n) according to the following formula,

${D D.}_{i i} = = dis dis tan the tan ce ce (({A A}_{i i},, {A A}_{w w})) = = {Σ Σ}_{k k = = 00}^{N N - - 11} \sqrt{{| | | | {A A}_{i i,, k k} - - {A A}_{w w,, k k} | | | |}^{22}},, i i = = 1,2,3 1,2,3 - - - - - - ((55))$

两组信号的欧氏距离代表了两组信号的相似性，欧氏距离越小意味着两组信号越相似，因此与最小D_i对应的信号x_i即为代表滚筒式球磨机料位信息的信号，取与最小D_i对应的A_i(n)序列中的前半部分作为序列长度为

的输出信号z(n)，即The Euclidean distance of the two groups of signals represents the similarity of the two groups of signals, the smaller the Euclidean distance means the more similar the two groups of signals, so the signal x _i corresponding to the minimum D _i is the signal representing the material level information of the drum type ball mill , take the first half of the A _i (n) sequence corresponding to the minimum D _i as the sequence length

The output signal z(n) of

z(n)＝A_i(n)， $n = 0,1,2, . . ., \frac{N}{2} - 1 - - - (6);$ z(n)=A _i (n), $no = 0,1,2, . . ., \frac{N}{2} - 1 - - - (6);$

(5)、根据步骤(4)的输出信号z(n)，按下式计算滚筒式球磨机内料位L的值，(5), according to the output signal z (n) of step (4), calculate the value of material level L in the drum type ball mill according to the following formula,

$L L = = \frac{{Σ Σ}_{n no = = 00}^{{N N}_{c c}} z z ((n no))}{{Σ Σ}_{n no = = 00}^{\frac{N N}{22} - - 11} z z ((n no))} - - - - - - ((77)),,$

其中，式(7)中N_c值的确定方法如下：Wherein, the determination method of _Nc value in formula (7) is as follows:

在滚筒式球磨机料位较高的情况下，选择两种料位工况，按步骤(1)-(4)取得与两种料位工况对应的输出信号z_h(n)、z_l(n)；然后令N_c从1到

依次取值，根据与两种料位工况对应的输出信号z_h(n)、z_l(n)，按照步骤(5)中的式(7)分别对两种料位工况进行料位计算，得到与各N_c对应的料位值L_h(k)、L_l(k)，其中，

k = 1,2, \cdot \cdot \cdot, \frac{N}{2} - 1;

L_h(k)-L_l(k)结果的绝对值最大时所对应的k值即为要确定的N_c值。In the case of a high material level in the roller ball mill, select two material level working conditions, and follow steps (1)-(4) to obtain the output signals z _h (n), z _l ( n); then let N _c range from 1 to

Values are taken in turn, according to the output signals z _h (n) and z _l (n) corresponding to the two material level working conditions, according to the formula (7) in step (5), the material level Calculate and obtain the material level value L _h (k), L _l (k) corresponding to each N _c , wherein,

k = 1,2, \cdot \cdot \cdot, \frac{N}{2} - 1;

The value of k corresponding to the maximum absolute value of the result of L _h (k)-L _l (k) is the value of N _c to be determined.

与现有技术相比，本发明采用多个音频传感器分别获取滚筒式球磨机滚筒的噪声信号、滚筒式球磨机本身的背景噪声信号和滚筒式球磨机所处现场环境的噪声信号；并利用盲源分离技术对所采集的噪声信号进行处理，应用频谱分析法对信号进行甄别，将代表球磨机料位的音频信号从噪声信号分离、确定，然后采用频谱比值法对滚筒式球磨机内的料位进行计算，准确度高，精度高，灵敏度高。其中，盲源分离技术是近几年才发展起来的一种信息处理方法，通过对采集的混合信号的特性分析，能够将混合信号中的源信号分离出来，而不需要源信号和传输信道的过多信息，已经广泛应用于音频信号处理和图像信号处理等领域。目前，经验丰富的运行人员在复杂的现场环境中可以通过现场声音判断出球磨机内部的料位，主要的原因在于人耳的特性和人类听觉系统对声音信号的处理能力，而本发明所述方法恰是模仿了人耳的特性和人类听觉系统对声音信号的处理能力来实现对滚筒式球磨机料位的判断。Compared with the prior art, the present invention adopts a plurality of audio sensors to separately acquire the noise signal of the roller mill drum, the background noise signal of the roller mill itself and the noise signal of the field environment where the roller mill is located; and utilizes blind source separation technology Process the collected noise signal, use the spectrum analysis method to screen the signal, separate and determine the audio signal representing the material level of the ball mill from the noise signal, and then use the spectrum ratio method to calculate the material level in the roller ball mill, accurately High accuracy, high precision, high sensitivity. Among them, the blind source separation technology is an information processing method developed in recent years. By analyzing the characteristics of the collected mixed signal, the source signal in the mixed signal can be separated without the need for source signals and transmission channels. Too much information has been widely used in the fields of audio signal processing and image signal processing. At present, experienced operating personnel can judge the material level inside the ball mill through on-site sound in a complex on-site environment. The main reason is that the characteristics of the human ear and the processing ability of the human auditory system to sound signals. It just imitates the characteristics of the human ear and the ability of the human auditory system to process sound signals to realize the judgment of the material level of the roller ball mill.

本发明所述方法合理，能有效地降低环境噪声、本身背景噪声和相邻球磨机噪声的干扰，对滚筒式球磨机料位进行准确可靠的检测。The method of the invention is reasonable, can effectively reduce the interference of environmental noise, its own background noise and the noise of adjacent ball mills, and can accurately and reliably detect the material level of the drum type ball mill.

附图说明 Description of drawings

图1为本发明所述方法步骤1中音频传感器的指向布置示意图；Fig. 1 is a schematic diagram of the pointing arrangement of the audio sensor in the method step 1 of the present invention;

图2为本发明所述方法的流程图；Fig. 2 is a flowchart of the method of the present invention;

图中：1、2、3-音频传感器；4-滚筒式球磨机；5-电机传动机构；6-落料口。In the figure: 1, 2, 3- audio sensor; 4- roller ball mill; 5- motor transmission mechanism; 6- blanking port.

具体实施方式 Detailed ways

如图1、2所示，基于多传感器融合技术的滚筒式球磨机料位检测方法，步骤如下：As shown in Figures 1 and 2, the method of detecting the material level of a roller ball mill based on multi-sensor fusion technology, the steps are as follows:

(4)、从信号x₁，x₂，x₃中甄别出代表滚筒式球磨机料位信息的信号；甄别方法如下：a、对与滚筒式球磨机滚筒处噪声信号s₁对应的经步骤2处理后得到的信号w₁，以及经步骤(3)得到的信号x₁，x₂，x₃进行频谱分析，按下式对信号w₁，x₁，x₂，x₃进行N点离散傅立叶变换，(4) From the signals x ₁ , x ₂ , and x ₃ , identify the signal representing the material level information of the drum type ball mill; the screening method is as follows: a. The signal corresponding to the noise signal s ₁ at the drum of the drum type ball mill is processed in step 2 Spectrum analysis is performed on the obtained signal w ₁ and the signals x ₁ , x ₂ , and x ₃ obtained in step (3), and N-point discrete Fourier transform is performed on the signals w ₁ , x ₁ , x ₂ , and x ₃ according to the following formula ,

与最小D_i对应的信号x_i即为代表滚筒式球磨机料位信息的信号，取与最小D_i对应的A_i(n)序列中的前半部分作为序列长度为

的输出信号z(n)，即The signal x _i corresponding to the minimum D _i is the signal representing the material level information of the roller ball mill, and the first half of the A _i (n) sequence corresponding to the minimum D _i is taken as the sequence length is

The output signal z(n) of

k = 1,2, \cdot \cdot \cdot, \frac{N}{2} - 1;

k = 1,2, &Center Dot; \cdot \cdot, \frac{N}{2} - 1;

具体实施时，滚筒式球磨机4滚筒处噪音信号s₁的采集一般通过指向滚筒式球磨机4落料口6位置的音频传感器1实现；滚筒式球磨机本身背景噪声相对较高处的噪音信号s₂的采集通过指向滚筒式球磨机体背景噪声相对较高的电机传动机构5的音频传感器2实现；滚筒式球磨机所处环境音频信号s₃的采集通过指向与音频传感器1相反的音频传感器3实现。During specific implementation, the collection of the noise signal s ₁ at the drum of the drum mill 4 is generally realized by the audio sensor 1 pointing to the position of the feeding opening 6 of the drum mill 4; the background noise of the drum mill itself is relatively _high . Acquisition is achieved by pointing to the audio sensor 2 of the motor transmission mechanism 5 with relatively high background noise of the roller mill body; the acquisition of the audio signal s ₃ of the environment where the roller mill is located is realized by pointing to the audio sensor 3 opposite to the audio sensor 1.

Claims

1. A method for detecting material level of a roller ball mill based on multi-sensor fusion technology, characterized in that the steps are as follows:

(1) Simultaneously use three audio sensors to collect the noise signal s ₁ at the drum of the drum mill, the noise signal s ₂ of the drum mill itself where the background noise is relatively high, and the noise signal s of the environment where the drum mill is located ₃ ;

(2) Condition the noise signals s ₁ , s ₂ , and s ₃ collected in step (1) to obtain signals u ₁ , u ₂ , and u ₃ , and convert the signals u ₁ , u ₂ , and u ₃ into digital signals Sequence signals v ₁ , v ₂ , v ₃ , and then pre-whiten the digital sequence signals v ₁ , v ₂ , v ₃ to obtain signals w ₁ , w ₂ , w ₃ , where, for noise signals s ₁ , s ₂ , s ₃ The signals u ₁ , u ₂ , and u ₃ obtained after conditioning should be suitable for the input range of the A/D converter used for analog-to-digital conversion;

(3), respectively performing blind source separation on the signals w ₁ , w ₂ , and w ₃ obtained in step (2) to obtain three statistically independent signals x ₁ , x ₂ , x ₃ ;

(4) From the signals x ₁ , x ₂ , and x ₃ , identify the signal representing the material level information of the drum type ball mill; the screening method is as follows: a. The signal corresponding to the noise signal s ₁ at the drum of the drum type ball mill is processed in step 2 Spectrum analysis is performed on the obtained signal w ₁ and the signals x ₁ , x ₂ , and x ₃ obtained in step (3), and N-point discrete Fourier transform is performed on the signals w ₁ , w ₁ , w ₂ , and w ₃ according to the following formula ,

f_{w} (no) = Σ_{k = 0}^{N - 1} w_{1} (k) e^{- j 2 πnk / N},

n=0, 1, 2, . . . , N-1 (1)

f_{i} (no) = Σ_{k = 0}^{N - 1} x_{i} (k) e^{- j 2 πnk / N},

n=0, 1, 2, . . . , N-1, i=1, 2, 3 (2)

b. Take the modulus of the discrete Fourier transform results of the signals w ₁ , w ₁ , w ₂ , and w ₃ according to the following formula, and normalize them,

{A A}_{w w} ((n no)) = = \frac{| | {F f}_{w w} ((n no)) | |}{max max ((| | {F f}_{w w} ((n no)) | |))} - - - - - - ((33))

{A A}_{i i} ((n no)) = = \frac{| | {F f}_{i i} ((n no)) | |}{max max ((| | {F f}_{i i} ((n no)) | |))},, i i = = 1,2,3 1,2,3 - - - - - - ((44))

c. Calculate the Euclidean distance D _i between A _i (n), i=1, 2, 3 and A _w (n) according to the following formula,

{D D.}_{i i} = = dis dis tan the tan ce ce (({A A}_{i i},, {A A}_{w w})) = = {Σ Σ}_{k k = = 00}^{N N - - 11} \sqrt{{| | | | {A A}_{i i,, k k} - - {A A}_{w w,, k k} | | | |}^{22}},, i i = = 1,2,3 1,2,3 - - - - - - ((55))

The signal x _i corresponding to the minimum D _i is the signal representing the material level information of the roller ball mill, and the first half of the A _i (n) sequence corresponding to the minimum D _i is taken as the sequence length is The output signal z(n) of

z(n)=A _i (n),

no = 0,1,2, . . ., \frac{N}{2} - 1 - - - (6);

(5), according to the output signal z (n) of step (4), calculate the value of material level L in the drum type ball mill according to the following formula,

L L = = \frac{{Σ Σ}_{n no = = 00}^{{N N}_{c c}} z z ((n no))}{{Σ Σ}_{n no = = 00}^{\frac{N N}{22} - - 11} z z ((n no))} - - - - - - ((77)),,

Wherein, the determination method of _Nc value in formula (7) is as follows:

In the case of a high material level in the roller ball mill, select two material level working conditions, and follow steps (1)-(4) to obtain the output signals z _h (n), z _l ( n); then let N _c range from 1 to

Values are taken in turn, according to the output signals z _h (n) and z _l (n) corresponding to the two material level working conditions, according to the formula (7) in step (5), the material level Calculate and obtain the material level values L _h (k) and L _l (k) corresponding to each N _c , wherein, k=1, 2, ..., The value of k corresponding to the maximum absolute value of the result of L _h (k)-L _l (k) is the value of N _c to be determined.