An intelligent ore identification method and device combining laser radar with vibration signal

Abstract

The present invention discloses an artificial intelligence ore identification method and device based on laser imaging, which relates to the technical fields of photoelectric imaging, image processing and target detection. The method and device can obtain the volume characteristics of the ore by using the distance echo point cloud collected by the laser radar, and use the volume characteristics to make up for the defects of identifying the ore by a single vibration signal. The present invention includes the following steps: the ore falls from the discharge port and falls freely onto the vibration signal collection board, the collision generates a vibration signal, the vibration signal is collected, and the vibration signal is demodulated to obtain the impact characteristic parameter. After hitting the vibration signal collection board, the ore falls onto the uniformly moving conveyor belt; the ore is scanned by the laser radar to obtain the distance echo point cloud data of the ore, and the distance echo point cloud data is processed to obtain the volume parameter of the ore. After the impact characteristic parameter and the volume parameter are jointly processed, a joint characteristic parameter is obtained, and the joint characteristic parameter is used to identify the type of ore.

Description

Technical Field

[0001] The present invention relates to the technical fields of photoelectric imaging, image processing and target detection, and in particular to an intelligent ore recognition method and device combining laser radar with vibration signals. Background Art

[0002] In the industrial field, especially in the field of conveyor-based target detection and recognition, object segmentation, positioning, and recognition are essential, but the implementation of these steps is usually based on a clear image without background noise interference. For ordinary CCD, when the background is more complex, using algorithms to remove image background noise cannot achieve an ideal effect, making object segmentation and positioning more difficult, and ultimately leading to a lower recognition rate.

[0003] The applicability and stability of existing ore identification methods are poor, and the requirements for the cleanliness of the ore surface are high. For ores with greatly different physical properties, such as coal and gangue, when their volumes are similar, the vibration signals generated by the falling impact will be quite different. In recent years, some domestic scholars have carried out research on ore identification based on vibration signals, but due to the lack of relevant theories and the limitations of the production site, the engineering progress of ore identification based on vibration signals has been severely restricted.

[0004] Therefore, there is an urgent need for an ore identification solution that can make up for the defects of identifying ore with a single vibration signal. Summary of the invention

[0005] In view of this, the present invention provides an intelligent ore identification method and device combining laser radar with vibration signals, which can obtain the volume characteristics of the ore using the distance echo point cloud collected by the laser radar, and use the volume characteristics to make up for the defects of identifying ore with a single vibration signal, which has engineering significance.

[0006] To achieve the above object, the technical solution of the present invention comprises the following steps:

[0007] The ore falls from the discharge port and falls freely onto the vibration signal collection board, where it impacts and generates a vibration signal, which is then collected and demodulated to obtain impact characteristic parameters.

[0008] The ore falls onto the uniformly moving conveyor belt after hitting the vibration signal collection plate.

[0009] The ore is scanned by laser radar to obtain the distance echo point cloud data of the ore, and the distance echo point cloud data is processed to obtain the volume parameters of the ore.

[0010] The impact characteristic parameters and volume parameters are jointly processed to obtain the joint characteristic parameters, and the neural network model is used to identify the ore types.

[0011] Furthermore, the vibration signal is collected and demodulated to obtain the impact characteristic parameters, specifically:

[0012] The dynamic equation of the process in which the ore falls from the discharge port and freely lands on the vibration signal acquisition board is:

[0013]

[0014] Where M is the structural mass matrix of the vibration signal acquisition board; C is the structural damping matrix of the vibration signal acquisition board; K is the structural stiffness matrix of the vibration signal acquisition board; x(t) is the overall node displacement of the vibration signal acquisition board, and is the speed of the vibration signal acquisition board, is the acceleration of the vibration signal acquisition board, t is the time variable; F is the load matrix;

[0015] During the collision process, the initial collision time is t1, and when the collision time Δt→0, the collision impulse I is:

[0016]

[0017] The impact characteristic parameter is the ore velocity increment Δv2 of the ore and the vibration signal collection plate at the moment of collision and separation;

[0018] Δv2=M -1 I+(1+e)v1

[0019] Wherein v1 is the instantaneous velocity of the ore before the collision between the ore and the vibration signal collection board; e is the recovery coefficient of the vibration signal collection board.

[0020] Furthermore, the volume characteristic parameter is the volume integral of the range echo point cloud data of the ore in three-dimensional space.

[0021] Furthermore, the vibration signal collection plate is a metal plate

[0022] Furthermore, the joint characteristic parameter is the ratio of the impact characteristic parameter to the volume characteristic parameter.

[0023] Furthermore, the joint characteristic parameters are used to identify the type of ore using a neural network model, specifically: the joint characteristic parameters are identified using a convolutional neural network model.

[0024] Another embodiment of the present invention further provides an intelligent ore identification device combining laser radar with vibration signals, comprising a conveyor belt, a vibration signal acquisition module, a laser radar module and a comprehensive processing module;

[0025] A discharge port is arranged at one end of the conveyor belt, and a vibration signal acquisition module is arranged below the discharge port; a laser radar module is arranged above the middle of the conveyor belt; and the conveyor belt moves at a constant speed.

[0026] The vibration signal acquisition module includes a vibration signal acquisition board and a vibration signal sensor. The ore falls from the discharge port and freely falls onto the vibration signal acquisition board to generate a vibration signal. The vibration signal sensor generated on the vibration signal acquisition board collects the vibration signal and transmits the vibration signal to the comprehensive processing module.

[0027] After hitting the vibration signal acquisition board, the ore falls onto the conveyor belt and runs smoothly through the laser radar module. The laser radar module scans and obtains the distance echo point cloud data of the ore, and the distance echo point cloud is transmitted to the comprehensive processing module.

[0028] The comprehensive processing module demodulates the vibration signal to obtain the impact characteristic parameters.

[0029] The comprehensive processing module processes the distance echo point cloud data of the ore to obtain the volume parameters of the ore.

[0030] The comprehensive processing module jointly processes the impact characteristic parameters and the volume parameters to obtain the joint characteristic parameters, and uses the neural network model to identify the ore type based on the joint characteristic parameters.

[0031] Furthermore, the vibration signal collection plate is a metal plate.

[0032] Beneficial effects:

[0033] 1. The artificial intelligence ore recognition method and device based on laser imaging provided in the embodiment of the present invention can greatly improve the recognition accuracy of the ore by using the vibration signal of the falling ore combined with the laser radar to identify the ore. The volume recognition accuracy of the laser radar can reach the millimeter level. The accuracy of the vibration signal sensor in the embodiment of the present invention can reach 20mv / g, and the density physical properties of the ore can be sensitively obtained through multi-mode characteristic parameters. The present invention can well solve the problem of affecting the recognition rate due to the high water content and dust covering the ore surface. The recognition rate of ores with covered surfaces, such as coal and gangue, reaches 97%. It does not rely on ambient light during operation, has strong anti-vibration ability, and has explosion-proof characteristics.

[0034] 2. The artificial intelligence ore identification method and device based on laser imaging provided by the present invention have strong environmental adaptability, do not require a light source and have no radiation hazards.

[0035] 3. The artificial intelligence ore identification method and device based on laser imaging provided by the present invention use a convolutional neural network algorithm to identify ores on a conveyor belt in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A flow chart of an intelligent ore identification method using a laser radar combined with a vibration signal provided in an embodiment of the present invention;

[0037] Figure 2 A structural diagram of an intelligent ore identification device combining laser radar with vibration signals provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0039] The present invention provides an intelligent ore identification method combining laser radar with vibration signals, such as Figure 1 As shown, the method comprises the following steps:

[0040] S1. The ore falls from the discharge port and falls freely onto the vibration signal collection board. The vibration signal collection board is a metal plate, specifically a steel plate.

[0041] S2. The ore collides with the vibration signal collection board to generate a vibration signal, and the vibration signal is collected. When the ore hits the vibration signal collection board, the corresponding information of time and impulse can be obtained through the vibration signal sensor, and the impact characteristic parameters can be obtained through demodulation. The vibration signal is demodulated to obtain the impact characteristic parameters.

[0042] The dynamic equation of the ore during the collision process can be specifically expressed by the following formula:

[0043]

[0044] Where M is the structural mass matrix of the vibration signal acquisition board; C is the structural damping matrix of the vibration signal acquisition board; K is the structural stiffness matrix of the vibration signal acquisition board; x(t) is the overall node displacement of the vibration signal acquisition board, and is the speed of the vibration signal acquisition board, is the acceleration of the vibration signal acquisition board, t is the time variable; F is the load matrix;

[0045] During the collision process, the initial collision time is t1, and when the collision time Δt→0, the collision impulse I is:

[0046]

[0047] The impact characteristic parameter is the ore velocity increment Δv2 of the ore and the vibration signal collection plate at the moment of collision and separation;

[0048] Δv2=M -1 I+(1+e)v1

[0049] Wherein v1 is the instantaneous velocity of the ore before the collision between the ore and the vibration signal collection board; e is the recovery coefficient of the vibration signal collection board.

[0050] S3. The ore then falls onto a uniformly moving conveyor belt, and the ore 6 is scanned using a laser radar.

[0051] S4. The laser radar emits laser signals to the ore on the conveyor belt. The laser signals will be reflected on the surface of the ore. The laser radar has a laser return signal receiving unit, which receives the laser return signal emitted by its own laser signal transmitting unit and performs data conversion to obtain the distance echo point cloud of the ore. The distance echo point cloud data is processed to obtain the volume parameter of the ore. The volume feature parameter is the volume integral of the point cloud in three-dimensional space.

[0052] S5. After jointly processing the impact characteristic parameters and the volume parameters, a joint characteristic parameter is obtained; and the joint characteristic vector is input into a training recognition model in a convolutional neural network.

[0053] In the present invention, a convolutional neural network model can be used to identify the information after joint processing, specifically:

[0054] The existing ore impact characteristic parameters and volume parameters are jointly processed to construct a convolutional neural network model, and the convolutional neural network model is trained and tested using the existing ore impact characteristic parameters and volume information after the joint processing to obtain a trained convolutional neural network model. Among them, the existing ore impact characteristic parameters and volume parameters are collected in advance, and the corresponding ore types are also known.

[0055] The ore texture and shape information are processed jointly, and the combined information is input into the trained convolutional neural network model. The trained convolutional neural network model can output the ore recognition result.

[0056] S6. Using a neural network model to identify the type of ore based on the joint characteristic parameters, and outputting the ore identification result. In the present invention, a convolutional neural network model can be used to identify the characteristic parameters, specifically:

[0057] The artificial intelligence ore identification method based on laser imaging of the present invention can greatly improve the accuracy of ore identification by using the vibration signal of the falling ore combined with the laser radar to identify the ore. The volume recognition accuracy of the laser radar can reach the millimeter level, and the accuracy of the vibration signal sensor can reach 20mv / g, and the density physical properties of the ore can be sensitively obtained through multi-mode characteristic parameters. The present invention can well solve the problem of affecting the recognition rate due to the high water content and dust covering on the surface of the ore. It does not rely on ambient light during the working process, has strong anti-vibration ability, and has explosion-proof characteristics. Using the convolutional neural network model, the identification of ore impact characteristic parameters and volume parameters is more accurate. The operation process is simple, the applicability is good, and the ore on the conveyor belt can be identified in real time.

[0058] like Figure 2 As shown, another embodiment of the present invention further provides an intelligent ore identification device combining laser radar with vibration signals, the device comprising: a conveyor belt 4, a vibration signal acquisition module 3, a laser radar module 2 and a comprehensive processing module 5.

[0059] A discharge port 1 is arranged at one end of the conveyor belt 4 , and a vibration signal acquisition module 3 is arranged below the discharge port 1 ; a laser radar module 2 is arranged above the middle of the conveyor belt 4 .

[0060] The vibration signal acquisition module 3 includes a vibration signal acquisition board and a vibration signal sensor. The ore 6 falls from the discharge port and freely falls onto the vibration signal acquisition board to generate a vibration signal. The vibration signal sensor generated on the vibration signal acquisition board collects the vibration signal and transmits the vibration signal to the comprehensive processing module 5.

[0061] After hitting the vibration signal collection board, the ore falls onto the conveyor belt 4 and runs smoothly through the laser radar module 2. The laser radar module 2 scans and obtains the distance echo point cloud data of the ore, and the distance echo point cloud is transmitted to the comprehensive processing module 5.

[0062] The comprehensive processing module 5 demodulates the vibration signal to obtain the impact characteristic parameters.

[0063] The comprehensive processing module 5 processes the distance echo point cloud data of the ore to obtain the volume parameters of the ore.

[0064] The comprehensive processing module 5 obtains the joint characteristic parameters after jointly processing the impact characteristic parameters and the volume parameters, and uses the neural network model to identify the ore type based on the joint characteristic parameters.

[0065] In summary, the above are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent ore identification method combining laser radar with vibration signals, It is characterized in that The steps include: The ore falls from the discharge port and falls freely onto the vibration signal collection board, generating a vibration signal by impact. The vibration signal is collected and demodulated to obtain the impact characteristic parameter. The ore falls onto the uniformly moving conveyor belt after hitting the vibration signal collection plate; Scanning the ore using a laser radar to obtain distance echo point cloud data of the ore, and processing the distance echo point cloud data to obtain volume parameters of the ore; The impact characteristic parameters and the volume parameters are jointly processed to obtain joint characteristic parameters, and the joint characteristic parameters are used to identify the type of ore using a neural network model; The collecting of the vibration signal and demodulating the vibration signal to obtain the impact characteristic parameter are specifically as follows: The dynamic equation of the process in which the ore falls from the discharge port and freely lands on the vibration signal acquisition board is: Where M is the structural mass matrix of the vibration signal acquisition board; C is the structural damping matrix of the vibration signal acquisition board; K is the structural stiffness matrix of the vibration signal acquisition board; x(t) is the overall node displacement of the vibration signal acquisition board, and is the speed of the vibration signal acquisition board, is the acceleration of the vibration signal acquisition board, t is the time variable; F is the load matrix; During the collision process, the initial collision time is t 1 , when the collision time Δt→0, the collision impulse I is: The impact characteristic parameter is the ore velocity increment Δv of the ore and the vibration signal collection plate at the moment of collision and separation. 2 ; Δv 2 =M -1 I+(1+e)v 1 where v 1 is the instantaneous velocity of the ore before the collision between the ore and the vibration signal collection board; e is the recovery coefficient of the vibration signal collection board.

2. The method according to claim 1, It is characterized in that The volume parameter is the volume integral of the range echo point cloud data of the ore (6) in three-dimensional space.

3. The method according to claim 1, It is characterized in that The vibration signal collection plate is a metal plate.

4. The method according to any one of claims 1 to 3, It is characterized in that The joint characteristic parameter is the ratio of the impact characteristic parameter to the volume parameter.

5. The method according to claim 4, It is characterized in that The joint characteristic parameters use a neural network model to identify the type of ore, specifically: The joint feature parameters are identified using a convolutional neural network model.

6. An intelligent ore identification device based on the laser radar combined with vibration signals according to the method of claim 1, It is characterized in that It comprises a conveyor belt (4), a vibration signal acquisition module (3), a laser radar module (2) and a comprehensive processing module (5); A discharge port (1) is arranged at one end of the conveyor belt (4), and the vibration signal acquisition module (3) is arranged below the discharge port (1); the laser radar module (2) is arranged above the middle of the conveyor belt (4); and the conveyor belt (4) moves at a constant speed; The vibration signal acquisition module (3) comprises a vibration signal acquisition board and a vibration signal sensor. The ore (6) falls from the discharge port and freely falls onto the vibration signal acquisition board to generate a vibration signal. The vibration signal sensor generated on the vibration signal acquisition board collects the vibration signal and transmits the vibration signal to the comprehensive processing module (5). After hitting the vibration signal collection board, the ore falls onto the conveyor belt (4) and smoothly passes through the laser radar module (2). The laser radar module (2) scans and obtains the distance echo point cloud data of the ore, and the distance echo point cloud is transmitted to the comprehensive processing module (5); The comprehensive processing module (5) demodulates the vibration signal to obtain impact characteristic parameters; The comprehensive processing module (5) processes the distance echo point cloud data of the ore to obtain the volume parameter of the ore; The comprehensive processing module (5) jointly processes the impact characteristic parameters and the volume parameters to obtain joint characteristic parameters, and uses a neural network model to identify the type of ore based on the joint characteristic parameters.

7. The device according to claim 6, It is characterized in that The vibration signal collection plate is a metal plate.

Images