Device and method for measuring elastic recovery coefficient of wet particles based on CNN (Convolutional Neural Network)

Abstract

The invention provides a wet particle elastic recovery coefficient measuring device and method based on a CNN, and belongs to the technical field of wet particle elastic coefficient measurement. The problem that in the prior art, when an empirical formula obtained through manually processed data is adopted, large deviation often exists, prediction on the particle collision behavior in the solid-containing system is not accurate, and therefore the macroscopic fluidization characteristic is affected is solved. The particle emitter is located obliquely above the water tank, the water tank is installed on the fixed horizontal plate, the vertical height gauge is installed at one end of the fixed horizontal plate, the high-speed camera is placed on the left side of the water tank, the first CCD camera is placed above the water tank, and the second CCD camera is placed on the right side of the water tank. The high-speed camera, the first CCD camera and the second CCD camera transmit acquired image information to the image preprocessing unit, the image preprocessing unit transmits processed information to the CNN convolution processing unit, and the CNN is a convolutional neural network. The device is mainly used for measuring the wet particle elastic coefficient.

Description

Technical Field

[0001] The invention belongs to the technical field of wet particle elastic coefficient measurement, and in particular relates to a wet particle elastic recovery coefficient measurement device and a measurement method based on CNN. Background Art

[0002] Processes such as fluidization, pneumatic conveying and mixing are characterized by strong particle-particle and particle-wall interactions. In coating, granulation and agglomeration processes, liquids are usually involved in the form of liquid layers or droplets, resulting in partial or complete wetting of the particle surface. Wet particles behave completely differently from dry particles, resulting in very complex dynamics, for example in fluidized beds. Therefore, knowledge of the collision dynamics of wet solids is crucial for an accurate description of such processes. To characterize the collision dynamics, the elastic recovery coefficient is widely used, which is defined as the ratio of the rebound velocity to the impact velocity and represents the energy dissipation during the particle collision.

[0003] Traditional particle elastic recovery coefficient measurement devices are usually used to measure the elastic recovery coefficient of dry particles when they are released vertically or at an angle. For example, dry particles are released vertically from above the surface to make them rebound vertically, and the process is recorded by a high-speed camera; dry particles are released from an angle above, and the rotational collision process is recorded by two high-speed cameras. This method is difficult to study wet particles, mainly because wet particles undergo a wetting process during collision, which makes it difficult to control the angle of incidence, etc., and their collision process is also more complicated. Traditional elastic recovery coefficient measurement devices rely solely on two high-speed cameras to capture the entire process accurately. In addition, traditional devices mostly manually process experimental data after the experiment, so human factors are easily introduced during data processing, resulting in increased errors in the calculation of the elastic recovery coefficient. These problems often lead to large deviations when researchers use the empirical formulas obtained from manually processed data, further leading to inaccurate predictions of particle collision behavior in solid-containing systems, thereby affecting the macroscopic fluidization characteristics. Summary of the invention

[0004] In view of this, the present invention aims to propose a measuring device and method for the elastic recovery coefficient of wet particles based on CNN, so as to solve the problem that the prior art often has a large deviation when using the empirical formula obtained by manually processed data, which further leads to inaccurate prediction of the collision behavior of particles in solid-containing systems, thereby affecting the macroscopic fluidization characteristics.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A device for measuring elastic recovery coefficient of wet particles based on CNN includes a particle emitter, a water tank, a fixed horizontal plate, a vertical height meter, a high-speed camera, a first CCD camera, a second CCD camera, an image preprocessing unit and a CNN convolution processing unit, wherein the particle emitter is located obliquely above the water tank, the water tank is installed on the fixed horizontal plate, a vertical height meter is installed at one end of the fixed horizontal plate, the high-speed camera is placed on the left side of the water tank, the first CCD camera is placed above the water tank, and the second CCD camera is placed on the right side of the water tank, the high-speed camera, the first CCD camera and the second CCD camera all transmit the collected image information to the image preprocessing unit, and the image preprocessing unit transmits the processed information to the CNN convolution processing unit, wherein CNN is a convolutional neural network.

[0007] Furthermore, a laser pointer is installed at the emission port of the particle emitter.

[0008] Furthermore, the water tank is arranged on a fixed horizontal plate through a shock-absorbing plate.

[0009] Furthermore, the nozzle of the particle launcher is a circular hole, the nozzle size range is 0.001-0.005m, the outlet shooting speed range is 0.01-10m / s, and the rotation angular velocity range is 0-1rad / s.

[0010] Furthermore, the inclination angle of the fixed horizontal plate is less than ±0.1°, and the height difference within the fixed horizontal plate is less than ±0.02 mm.

[0011] Furthermore, the lens plane of the high-speed camera is perpendicular to the ground, the inclination angle is less than ±0.1°, the resolution is 4K, and the frame rate is 8000FPS. The lens plane of the first CCD camera is horizontal to the ground, and the inclination angle is less than ±0.1°.

[0012] Furthermore, the lens plane of the first CCD camera is kept horizontal to the ground, and the inclination angle is less than ±0.1°.

[0013] Furthermore, the resolution range of the image preprocessing unit is 480P-4K, and the image batch parallel processing capability is 30 images per second.

[0014] Furthermore, the CNN convolution processing unit has 2 convolution layers, a convolution kernel size ≥ 3x3, a real-time processing response time of less than 100ms, an inference speed ≥ 10FPS / s, and a calculation accuracy ≥ 98%.

[0015] A device for measuring elastic recovery coefficient of wet particles based on CNN, comprising the following steps:

[0016] Step 1: The particles are positioned by the particle launcher and launched into the water tank to achieve elastic collision and rebound;

[0017] Step 2: Capture the collision and rebound process of wet particles from different angles and distances through a high-speed camera, a first CCD camera, and a second CCD camera to collect image information;

[0018] Step 3: The collected image information is transmitted to the image preprocessing unit, and the image is preprocessed and integrated by the image preprocessing unit to correct and restore the information of the complete process of wet particles;

[0019] Step 4: The information of the complete process is passed to the CNN convolution processing unit to extract the image features of the collision process and calculate the elastic recovery coefficient of the wet particles.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] 1. The wet particle elastic recovery coefficient measuring device based on convolutional neural network of the present invention controls the particle incident speed and angle through a particle emitter and a laser indicator, thereby achieving accurate particle incidence, reducing initial experimental errors of particles, and effectively increasing the diversity of wet particle elastic recovery coefficient measurement data.

[0022] 2. The image collection device of the present invention has multi-angle, multi-distance images and macro-integration capabilities, so it can effectively and accurately capture the collision and rebound process of wet particles, which is conducive to analyzing the collision process from multiple angles and enhancing the understanding of the elastic recovery coefficient of wet particles and the accuracy of the calculation method.

[0023] 3. The present invention introduces a convolutional neural network to perform real-time batch processing of experimental data, which improves the data processing speed, effectively reduces human errors, and enhances the reliability of the wet particle elastic recovery coefficient measurement device.

[0024] 4. The convolutional neural network processing unit has the ability to automatically extract key features, can quickly locate the particle collision area without human intervention, extract rebound features, and further improve the calculation accuracy and efficiency of the elastic recovery coefficient of wet particles. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0026] Figure 1 This is a schematic diagram of a device for measuring elastic recovery coefficient of wet particles based on CNN described in the present invention.

[0027] 1-particle launcher, 2-laser pointer, 3-water tank, 4-shock-absorbing plate, 5-fixed horizontal plate, 6-vertical height meter, 7-high-speed camera, 8-first CCD camera, 9-second CCD camera, 10-image preprocessing unit, 11-CNN convolution processing unit. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely explain the technical solutions in the embodiments of the present invention. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict, and the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0029] It should be noted that the descriptions of the present invention regarding directions such as "left", "right", "left side", "right side", "upper", "lower", "top", "bottom", etc. are all defined based on the relationship between the orientations or positions shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the structure described must be constructed and operated in a specific orientation, and therefore, cannot be understood as a limitation on the present invention. In the description of the present invention, the meaning of "multiple" is more than two, unless otherwise clearly and specifically defined.

[0030] In the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0031] See attached Figure 1 This embodiment is described as a device for measuring the elastic recovery coefficient of wet particles based on CNN, including a particle emitter 1, a water tank 3, a fixed horizontal plate 5, a vertical height meter 6, a high-speed camera 7, a first CCD camera 8, a second CCD camera 9, an image preprocessing unit 10 and a CNN convolution processing unit 11. The particle emitter 1 is located obliquely above the water tank 3, the water tank 3 is installed on the fixed horizontal plate 5, and a vertical height meter 6 is installed at one end of the fixed horizontal plate 5. The high-speed camera 7 is placed on the left side of the water tank 3, the first CCD camera 8 is placed above the water tank 3, and the second CCD camera 9 is placed on the right side of the water tank 3. The high-speed camera 7, the first CCD camera 8 and the second CCD camera 9 all transmit the collected image information to the image preprocessing unit 10, and the image preprocessing unit 10 transmits the processed information to the CNN convolution processing unit 11. A laser indicator 2 is installed at the emission port of the particle emitter 1, and the water tank 3 is arranged on the fixed horizontal plate 5 through a shock-absorbing plate 4.

[0032] The particles are positioned by the particle launcher 1 and launched into the built-in water tank 3 of the shock-absorbing plate 4 placed on the fixed horizontal plate 5, and then the particles contact the water in the tank and realize elastic collision and rebound. This process is recorded by the high-speed camera 7, the first CCD camera 8 and the second CCD camera 9. Among them, the high-speed camera 7 placed on the horizontal side records the process of the rotating particles contacting and colliding with the liquid on the horizontal plane and rebounding, and obtains the particle rebound height according to the vertical height meter 6. The first CCD camera 8 placed on the vertical top records the process of the rotating particles contacting and colliding with the liquid on the vertical plane and rebounding. The second CCD camera 9 placed slightly further on the right uses a wide-angle lens to record the entire process from particle launch, particle contact with liquid to particle collision and rebound. The image preprocessing unit 10 receives the image from the high-speed camera, preprocesses and integrates the image to correct and further restore the information of the complete process of wet particle collision, and transmits the information to the CNN convolution processing unit 11 to extract the image features of the collision process and calculate the elastic recovery coefficient of the wet particles.

[0033] The wet particle elastic recovery coefficient measuring device based on convolutional neural network controls the particle incident speed and angle through the particle emitter 1 and the laser indicator 2 to achieve accurate particle incidence, reduce the initial experimental error of the particles, and effectively increase the diversity of the wet particle elastic recovery coefficient measurement data. The high-speed camera 7, the first CCD camera 8 and the second CCD camera 9 have multi-angle and multi-distance images, as well as macro integration capabilities, and can effectively and accurately capture the wet particle collision and rebound process, which is conducive to analyzing the collision process from multiple angles, enhancing the understanding of the wet particle elastic recovery coefficient and the accuracy of the calculation method. The introduction of convolutional neural network for real-time batch processing of experimental data improves the data processing speed, effectively reduces human errors, and enhances the reliability of the wet particle elastic recovery coefficient measuring device. In addition, the convolutional neural network processing unit has the ability to automatically extract key features, can quickly locate the particle collision area without human intervention, extract rebound features, and further improve the calculation accuracy and efficiency of the wet particle elastic recovery coefficient.

[0034] Furthermore, the incident angle and position of the particles are determined by the laser pointer 2, and the particle emitter 1 further controls the incident speed and the rotational angular velocity and emits the particles.

[0035] Furthermore, the particles are wetted by contact with the water in the water tank 3 and are buffered by the shock absorbing plate 4 , and rebound on the fixed horizontal plate 5 , and the rebound height is read by the vertical altimeter 6 .

[0036] Furthermore, the nozzle of the particle launcher is a circular hole, the nozzle size is adjustable in the range of 0.001 to 0.005m, the outlet shooting speed is adjustable in the range of 0.01 to 10m / s, and the rotation angular velocity is adjustable in the range of 0 to 1rad / s, which can ensure that the device can realize the measurement of the elastic recovery coefficient under various conditions.

[0037] Furthermore, the image preprocessing unit 10 supports a resolution range from 480p to 4K, and has an image batch parallel processing capability of 30 images per second; the CNN convolution processing unit 11 can achieve an inference speed of 10FPS / s or above, and a calculation accuracy of not less than 98%.

[0038] Furthermore, the inclination angle of the fixed horizontal plate 5 should be controlled within ±0.1°, and the height difference within the plate should be controlled within ±0.02mm; the lens plane of the high-speed camera 17 should be kept perpendicular to the ground, and the maximum allowable inclination angle should be less than ±0.1°; the lens plane of the first CCD camera 8 should be kept horizontal to the ground, and the maximum allowable inclination angle should be less than ±0.1°, which can ensure that the image collection system can effectively and comprehensively collect information on the collision and rebound process of wet particles and ensure the reliability of the data obtained.

[0039] Furthermore, the image preprocessing unit 10 supports a resolution range from 480p to 4K, and the image batch parallel processing capability reaches 30 images per second. The CNN convolution processing unit 11 can achieve an inference speed of 10FPS / s or above, which can ensure that the measuring device fully utilizes the advantages of artificial intelligence methods to achieve efficient, fast and accurate measurement of the elastic recovery coefficient of wet particles.

[0040] A device for measuring elastic recovery coefficient of wet particles based on CNN, comprising the following steps:

[0041] Step 1: The particles are positioned by the particle launcher 1 and launched into the water tank 3 to achieve elastic collision and rebound;

[0042] Step 2: The collision and rebound process of wet particles is captured from different angles and distances by a high-speed camera 7, a first CCD camera 8 and a second CCD camera 9 to collect image information;

[0043] Step 3: The collected image information is transmitted to the image preprocessing unit 10, and the image is preprocessed and integrated by the image preprocessing unit 10 to correct and restore the information of the complete process of wet particles;

[0044] Step 4: The information of the complete process is transmitted to the CNN convolution processing unit 11 to extract the image features of the collision process and calculate the elastic recovery coefficient of the wet particles.

[0045] The embodiments of the present invention disclosed above are only used to help illustrate the present invention. The embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. According to the content of this specification, many modifications and changes can be made. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well.

Claims

1. A device for measuring elastic recovery coefficient of wet particles based on CNN, characterized in that: The invention comprises a particle emitter (1), a water tank (3), a fixed horizontal plate (5), a vertical height meter (6), a high-speed camera (7), a first CCD camera (8), a second CCD camera (9), an image preprocessing unit (10) and a CNN convolution processing unit (11), wherein the particle emitter (1) is located obliquely above the water tank (3), the water tank (3) is mounted on the fixed horizontal plate (5), one end of the fixed horizontal plate (5) is mounted with a vertical height meter (6), the high-speed camera (7) is placed on the left side of the water tank (3), the first CCD camera (8) is placed above the water tank (3), the second CCD camera (9) is placed on the right side of the water tank (3), the high-speed camera (7), the first CCD camera (8) and the second CCD camera (9) all transmit the collected image information to the image preprocessing unit (10), and the image preprocessing unit (10) transmits the processed information to the CNN convolution processing unit (11).

2. The device for measuring elastic recovery coefficient of wet particles based on CNN according to claim 1, characterized in that: A laser pointer (2) is installed at the emission port of the particle emitter (1).

3. The device for measuring elastic recovery coefficient of wet particles based on CNN according to claim 1, characterized in that: The water tank (3) is arranged on a fixed horizontal plate (5) via a shock absorbing plate (4).

4. The device for measuring elastic recovery coefficient of wet particles based on CNN according to claim 1, characterized in that: The orifice of the particle emitter (1) is a circular hole, the orifice size range is 0.001-0.005m, the outlet shooting speed range is 0.01-10m / s, and the rotation angular speed range is 0-1rad / s.

5. The device for measuring elastic recovery coefficient of wet particles based on CNN according to claim 1, characterized in that: The inclination angle of the fixed horizontal plate (5) is less than ±0.1°, and the height difference within the fixed horizontal plate (5) is less than ±0.02 mm.

6. The device for measuring elastic recovery coefficient of wet particles based on CNN according to claim 1, characterized in that: The lens plane of the high-speed camera (7) is kept vertical to the ground, with an inclination angle of less than ±0.1°, a resolution of 4K, and a frame rate of 8000 FPS. The lens plane of the first CCD camera (8) is kept horizontal to the ground, with an inclination angle of less than ±0.1°.

7. The device for measuring elastic recovery coefficient of wet particles based on CNN according to claim 1, characterized in that: The lens plane of the first CCD camera (8) is kept horizontal to the ground, with an inclination angle of less than ±0.1°.

8. The device for measuring elastic recovery coefficient of wet particles based on CNN according to claim 1, characterized in that: The image preprocessing unit (10) has a resolution range of 480P-4K and an image batch parallel processing capability of 30 images per second.

9. The device for measuring elastic recovery coefficient of wet particles based on CNN according to claim 1, characterized in that: The CNN convolution processing unit (11) has two convolution layers, a convolution kernel size ≥ 3x3, a real-time processing response time of less than 100ms, an inference speed ≥ 10FPS / s, and a calculation accuracy ≥ 98%.

10. A device for measuring elastic recovery coefficient of wet particles based on CNN according to any one of claims 1 to 9, characterized in that: It includes the following steps: Step 1: particles are positioned by a particle launcher (1) and then launched into a water tank (3) to achieve elastic collision and rebound; Step 2: The collision and rebound process of the wet particles is captured from different angles and distances by a high-speed camera (7), a first CCD camera (8) and a second CCD camera (9), and image information is collected; Step 3: The collected image information is transmitted to the image preprocessing unit (10), and the image is preprocessed and integrated by the image preprocessing unit (10), so as to correct and restore the information of the complete process of wet particles; Step 4: The information of the complete process is transmitted to the CNN convolution processing unit (11) to extract the image features during the collision process and calculate the elastic recovery coefficient of the wet particles.

Images