Method for calculating the bulk material conveying rate or bulk material load of a vibrating machine

Abstract

In a method for calculating the bulk material conveying rate or the bulk material load of a vibrating conveyor machine, in which method: at at least two times with different load states, raw measured data from the vibrating conveyor machine are acquired by means of at least one acceleration sensor, speed sensor or travel sensor; and the raw measured data are then processed to give at least one vibration data feature selected from the list: amplitude, frequency and phase, it is provided that a feature data set consisting of at least one vibration data feature is created and stored, and on the basis of which a regression model is created. Then, on the basis of the created regression model and at least one current feature data set, the current actual load or bulk material conveying rate of the vibrating conveyor machine is determined and displayed.

Description

Technical Field

[0001] This invention relates to a method for calculating the bulk material feed rate or bulk material load of a vibrating machine. Background Technology

[0002] Vibrating machines or vibrating conveying machines (such as vibrating screens or vibrating conveyors) typically include: a movable vibrating body comprising a screening surface or a conveying surface; and a fixed support frame relative to which the movable vibrating body is mounted. Such vibrating machines are used for sorting and transporting bulk materials, for example, moving them from stockpiles or silos to locations where the bulk materials will be further processed.

[0003] To drive or move a vibrating body, a magnetic vibration actuator, or so-called an unbalanced exciter, is used. An unbalanced exciter has rotating unbalanced bodies or counterweights that transfer their accelerating force to the vibrating body, causing it to vibrate. An unbalanced exciter that causes the directional movement of the vibrating body is called a straightener.

[0004] Because many parameters must be considered, such as drive frequency, vibration amplitude, and vibration angle, adjusting and / or controlling the conveying capacity or unloading of vibrating machines is a difficult task.

[0005] Here, the measured weight of the conveyed quantity is valuable information for the customer, enabling them to individually determine and control or adjust the conveying capacity of the vibratory machine. In this way, the working potential of the vibratory machine can be fully utilized, and overload can be prevented.

[0006] For example, DE 103 01 143 A1 discloses an apparatus and method for adjusting the amount of bulk material on a conveying trough of a vibrating machine. The screening machine is supported relative to its support frame by four spring elements. Four weighing bars or load cells are arranged between the spring elements and the support frame as load detectors for determining the load on the conveying trough. These load cells are used to continuously determine the actual weight and current load of the vibrating machine and compare them with reference values.

[0007] Because of oscillating motion, where the vibrations may be within the resonant range of the vibrating machine, it is difficult to measure the weight of the bulk material being transported. Therefore, it is not possible to record the weight using pressure gauges or force sensors that employ strain gauges, which generate an electrical signal proportional to the weight based on their deformation.

[0008] Another method for controlling the unloading of a vibrating feeder is known from EP 1 188 695 A1. In this method, an accelerometer is used to measure the vertical acceleration of the vibrating feeder and, in addition, its drive frequency, so that conclusions about the actual conveying capacity can be drawn from these values.

[0009] However, using this method requires determining the relationship between vertical acceleration, drive frequency, and corresponding actual conveying capacity individually, either empirically or theoretically, for each vibrating machine, for example, in the form of a function.

[0010] Therefore, the purpose of this invention is to improve and simplify the determination of the bulk material feeding rate for vibrating machines. Summary of the Invention

[0011] This task is solved by the method and apparatus of the present invention.

[0012] The present invention provides a method for determining or calculating the bulk material feed rate or bulk material load of a vibrating machine, wherein the raw measurement data of the vibrating machine is first acquired using at least one acceleration sensor, velocity sensor or displacement sensor.

[0013] In the context of this invention, "vibrating machine" is understood to mean a vibrating conveying machine, such as a vibrating screen or a vibrating conveyor trough. In the case of a vibrating screen, the thickness of the bulk material conveyed on the screen liner decreases along the conveying direction due to continuous screening, while in the case of a vibrating conveyor, the material thickness remains substantially constant. Due to the change in material thickness, the position of the vibrating screen's center of gravity also changes. Although there are constant driving conditions, such as a fixed impact angle and a fixed stator frequency of the motor, the displacement of the center of gravity changes the distance to the impact axis, which in turn leads to a change in the "pitch motion." During the process of changing the material thickness and shifting the center of gravity, the amplitude and phase of the longitudinal acceleration also change accordingly. Furthermore, the mass or absolute load of the bulk material affects, for example, the "combined stroke" or maximum stroke of the vibrating machine. In addition, the load distribution of the bulk material on the loading or conveying surface affects the "longitudinal stroke" or stroke in the longitudinal direction of the vibrating machine. Conversely, the load distribution of the bulk material does not affect the lateral stroke of the vibrating machine.

[0014] Furthermore, there is a correlation between the load and the rotational speed of the rotor or unbalanced exciter because the increased inertial mass counteracts the centrifugal force of the exciter, thus leading to an increase in the moment of inertia at the drive shaft.

[0015] Therefore, the applicant has determined that there is no fixed algorithm for determining the bulk material feed rate or bulk material load of a vibrating machine. Instead, a learning-based AI algorithm is needed to take into account the individual characteristics of the vibrating machine that affect its vibrational behavior. In particular, the mass of the vibrating machine, its geometry, the screen liner, and / or the specific characteristics of the bulk material should be considered.

[0016] To account for the individual vibration behavior of the vibrating machine, raw measurement data is recorded in the method according to the invention—acceleration, velocity, or displacement, depending on the selected sensor, for at least two different loading conditions of the vibrating machine. For example, measurements at 0% loading may be involved when there is no bulk material on the vibrating machine, while measurements at 100% loading may be involved when the vibrating machine reaches its rated load. These raw measurement data are processed during calculation by an electronic evaluation unit into at least one characteristic selected from the following list: processed amplitude, frequency, and phase.

[0017] Subsequently, a so-called feature dataset is created from the features obtained in this way. These feature datasets may consist of only one or more features selected from the following list: amplitude, frequency, and / or phase. These feature datasets, or vibration signals, are stored and thus can be used for subsequent evaluation.

[0018] Importantly, the feature dataset generated from vibrating machine A is used only for evaluation of the same vibrating machine A.

[0019] The applicant's research indicates that the raw measurement data, or corresponding feature datasets or vibration signals, interact with the load of the vibrating machine. Therefore, in the method according to the invention, those indicators are obtained or filtered out from the feature datasets or vibration signals that are highly correlated with the reference signal of the bulk material conveying rate or bulk material load.

[0020] Then, a classification or regression model is created and / or trained based on the stored feature datasets, each of which shows a correlation between load and vibration behavior. In the classification or regression model, the metrics obtained from the feature datasets are assumed to be input variables, and the corresponding load information (e.g., 0% and 100% load) or reference signals are assumed to be response variables.

[0021] If only 0 to 100% load information is used as the response variable, then the regression model can estimate / calculate the absolute load at least in percentage terms. Furthermore, this information is already valuable for detecting overload or underload in vibrating machinery.

[0022] According to the invention, a reference signal or reference load signal can serve as a response variable. This reference signal can, for example, be a force measurement signal or motor current signal generated from an upstream, alternative, or indirect measurement process of the bulk material feed rate or bulk material load. For example, the reference load signal can be obtained by prior weighing of the bulk material using a funnel scale or weighing device. Alternatively, characteristics of the accelerometer, velocity sensor, or displacement sensor itself can be used as the reference signal.

[0023] By means of the method according to the invention, a multivariate classification model or regression model is generated, which takes into account the linear and / or nonlinear forms and includes, if necessary, an index of the feature dataset containing coefficients.

[0024] Based on regression models, it is possible to predict the value of one measurement, given the large common variance between two variables and the knowledge of one measurement or reference value.

[0025] Subsequently, the classification or regression model obtained in this way can be implemented in the electronic evaluation device in such a way that the actual load of the vibrating machine can be determined and / or displayed based on the currently measured raw measurement data or vibration data.

[0026] One embodiment of this method proposes repeating the step "Create a classification or regression model" after a time period Δt, following wear on the vibrating conveyor, maintenance measures, and / or other systemic changes such as changes in load, machine components, drive characteristics, or material properties. This ensures that the created regression model can continuously adapt to the changed boundary conditions. Typically, changes in the vibration behavior of a vibrating machine or vibrating conveyor caused by wear or other altered conditions are not immediately apparent. By repeating the measurements and training the created regression model (e.g., by adjusting the input variables), it can be ensured that the determination of the bulk material feed rate or bulk material load is correct despite the systemic changes.

[0027] An AI algorithm can be used to build a regression model that is weakly adaptive to suit the characteristics of individual machines. The regression model, in mathematical equation form, itself represents a working algorithm. An advantageous embodiment of the method according to the invention further specifies that only input variables highly correlated with reference signals or reference load signals of the bulk material feed rate or bulk material load are used to create the regression model. Here, correlation calculations are used to determine which input variables are suitable for use in the regression model.

[0028] According to one embodiment of the method of the present invention, the regression model is defined as C1 X1+C2 X2+C3 X2^2...CN The form Xn^n uses variables where the input variable X is considered a linear or non-linear factor based on the feature dataset, and / or uses coefficients C1, C2, ...; CN ≥ 0 to consider the input variable X.

[0029] To validate the method according to the invention, it is advantageous to examine the regression model using a feature dataset that was not used in the creation of the model. The historical dataset or feature dataset is thus divided into a training dataset and a validation dataset.

[0030] To determine and display the bulk material conveying volume or bulk material load of a vibratory conveyor, the present invention also provides an apparatus suitable for acquiring raw measurement data of the vibratory conveyor using at least one acceleration sensor, velocity sensor, or displacement sensor. Furthermore, this apparatus provides an electronic evaluation unit by which the raw measurement data is converted into at least one feature consisting of direction-dependent vibration measurement variables selected from the following list: processed amplitude, frequency, and phase.

[0031] Furthermore, the evaluation device is used to create a feature set consisting of at least one feature, and subsequently to create a regression model based on these feature datasets. The device also has a screen or display to show the bulk material load or bulk material conveying rate of the vibratory conveyor based on the created model.

[0032] As mentioned above, determining the bulk material load or conveying rate of a vibrating machine or vibrating conveyor is challenging because both the amount of bulk material on the vibrating machine and the characteristics of the vibrating machine can vary continuously. However, the bulk material load or conveying rate represents valuable information. Besides serving as a performance indicator of the vibrating machine, it allows for conclusions about potential machine overloads that affect the machine's lifespan. Conversely, a consistently low bulk material feed rate may indicate inefficient system utilization. Therefore, the method according to the invention offers advantages over known methods by taking into account the actual vibration behavior of the vibrating machine. Consequently, on the one hand, the underlying regression model can be continuously adjusted, and on the other hand, faults in the vibration behavior can be detected if adjustments to the regression model or algorithm become necessary. Attached Figure Description

[0033] The method according to the invention is explained in more detail below with the aid of flowcharts, and other features and advantages of the invention are disclosed.

[0034] Figure 1 A schematic diagram illustrating the operation of the method according to the present invention is shown. Detailed Implementation

[0035] Figure 1 A method for calculating the bulk material feed rate of a vibrating machine 1, in the form of a vibrating screen, according to the present invention is schematically illustrated. At least one sensor 12 is attached to the vibrating machine 1 and is data-connected to the computing unit of an evaluation device 2. This data connection is shown as a dashed line in the figure and can be implemented via a radio or wired connection, or via a permanent or temporary connection. Measurement data supplied by the sensor 12 is processed and stored in the computing unit to form a feature dataset 13. A regression model 6 is formed from the feature dataset 13 (which serves as input variables) and a reference signal 7 derived from an upstream or independent measurement process of the bulk material load. The regression model 6 based on the feature dataset 13 is validated and trained using a feature dataset 9 that is not used to create the model.

[0036] The validated regression model 8 is then transmitted to software 10 and to evaluation device 2 to display the calculation of bulk material load.

Claims

1. A method for calculating the feeding rate or loading capacity of bulk materials in a vibrating conveyor, characterized in that, The method includes the following steps: a) Record raw measurement data of the vibrating conveyor at at least two time points with different loading states using at least one acceleration sensor, velocity sensor, or displacement sensor. b) Process the raw measurement data into at least one vibration data feature selected from the following list: amplitude, frequency, phase. c) Create and store a feature dataset consisting of at least one vibration data feature. d) Using the stored feature dataset, a regression model is created, wherein, to train the regression model, model-based bulk material loading values ​​are matched with a reference signal or reference load signal for the bulk material delivery amount or bulk material loading amount, and e) Based on the created regression model and at least one current feature dataset, determine and display the current actual load of the vibrating conveyor.

2. The method according to claim 1, characterized in that, Step d) of creating the regression model is repeated after wear occurs on the vibrating conveyor, after maintenance measures and / or after other system changes, which are changes in load, machine components, drive characteristics or material properties.

3. The method according to any one of claims 1 or 2, characterized in that, Step a) to obtain the raw measurement data shall be performed at least at 0% bulk material loading and 100% rated load bulk material loading.

4. The method according to claim 1, characterized in that, The reference load signal, or the reference signal, is a force measurement signal or motor current signal generated from an upstream, alternative, or indirect measurement process of the bulk material feed rate or bulk material load.

5. The method according to any one of claims 1 or 4, characterized in that, To train the regression model, only predictor variables are used, whose model-based bulk material loading values ​​are highly correlated with the bulk material delivery volume or the reference load signal of the bulk material loading.

6. The method according to claim 5, characterized in that, The multivariate regression method is used during the training phase of the regression model.

7. The method according to claim 6, characterized in that, With C1 X1+C2 X2+C3 X2^2.... CN The regression model is created in the form Xn^n = bulk loading quantity, where the predictor variable X is considered as a linear or nonlinear factor, and / or the predictor variable X is considered by using coefficients C1, C2, ..., CN ≥ 0.

8. The method according to any one of claims 1-2, characterized in that, The regression model is validated using a feature dataset that has not yet been used to train the regression model.

9. An apparatus for determining the bulk material conveying rate or bulk material load of a vibrating conveyor, the apparatus comprising: - At least one acceleration sensor, velocity sensor, or path sensor is arranged to acquire raw measurement data of the vibrating conveyor. - Electronic evaluation unit, the electronic evaluation unit is used for - The raw measurement data is processed into at least one feature, which consists of directional vibrations measured from the following list: Amplitude, frequency, and phase - Create a feature dataset consisting of at least one of the features. - A regression model is created using the stored feature dataset, wherein, to train the regression model, model-based bulk material loading values ​​are matched with a reference signal or reference load signal for the bulk material delivery amount or bulk material loading amount. and a screen or display that shows the model-based bulk material load value or the model-based bulk material feed rate of the vibrating conveyor.

Images