Anchor bolt sorting and intelligent iron removal system for chute caving machine based on machine vision

Abstract

The present invention provides a machine vision-based anchor rod sorting and intelligent iron removal system for a chute ore-feeding machine, which relates to the field of image data processing and includes: an anchor rod sensing module, which is used to determine whether an anchor rod exists based on an ore-feeding machine image, magnetic field information, sound information and vibration information; a ore-feeding machine control module, which is used to control the ore-feeding machine to stop working after determining that an anchor rod exists; an intelligent iron removal module, which is used to determine anchor rod information based on the ore-feeding machine image, magnetic field information, sound information and vibration information, obtain a target area image, and determine whether an anchor rod exists based on the target area image, and after determining that an anchor rod exists, determine anchor rod sorting working parameters based on the target area image, and a collaborative robot performs anchor rod sorting according to the anchor rod sorting working parameters; the ore-feeding machine control module is also used to control the ore-feeding machine to continue working after determining that an anchor rod does not exist or the collaborative robot completes anchor rod sorting, and has the advantage of realizing automatic removal of anchor rods mixed in ore.

Description

Technical Field

[0001] The present invention relates to the field of image data processing, and in particular to a machine vision-based anchor rod sorting and intelligent iron removal system for a chute dredging machine. Background Art

[0002] An anchor is a rod-shaped structure that penetrates deep into the rock formation, usually made of metal or other high-strength materials. Its main function is to connect the unstable part of the rock formation with the stable part, and to enhance the overall stability of the rock formation by applying prestress. An anchor net is a mesh structure made of iron wire or steel wire, which is usually used in conjunction with an anchor. The main function of the anchor net is to tightly connect the rock formations between the anchors to form an integral support structure. This structure can effectively prevent the fall of small pieces of loose rock, while limiting the displacement and deformation of the rock formation. In the process of ore mining, with the increase of mining depth and mining intensity, the stability of the surrounding rock of the mine has become increasingly prominent. In order to ensure the safety of miners and the normal operation of the mine, effective support measures must be taken. As an economical, efficient and reliable support method, anchor nets are widely used in ore mining under various geological conditions.

[0003] In the process of mining, it is inevitable that a small number of anchor rods, anchor nets and other iron objects are mixed in the ore being transported due to various factors. If they are not picked out in time, they will cause the belt to tear and break after entering the transportation system with the ore, seriously affecting the normal operation of the transportation system and causing economic losses. The general method is for drivers and passengers to pick them out manually, which is inefficient and poses a safety risk to personnel.

[0004] Therefore, it is necessary to provide a machine vision-based anchor sorting and intelligent iron removal system for the chute ore placing machine to realize the automated removal of anchors mixed in the ore. Summary of the invention

[0005] The present invention provides a machine vision-based anchor sorting and intelligent iron removal system for a chute ore-feeding machine, comprising: an anchor sensing module, comprising an image acquisition unit, a magnetic induction unit, a sound acquisition unit, a vibration sensing unit and an anchor identification unit, wherein the image acquisition unit is used to acquire an image of the ore-feeding machine during the ore-feeding process, the magnetic induction unit, the sound acquisition unit and the vibration sensing unit are arranged on the ore-feeding machine, the magnetic induction unit is used to acquire magnetic field information during the ore-feeding process, the sound acquisition unit is used to acquire sound information during the ore-feeding process, the vibration sensing unit is used to acquire vibration information during the ore-feeding process, and the anchor identification unit is used to determine whether an anchor exists based on the ore-feeding machine image, magnetic field information, sound information and vibration information during the ore-feeding process; and a ore-feeding machine control module is used to control the ore-feeding machine to stop working after the anchor identification unit determines that an anchor exists. work; intelligent iron removal module, including a secondary determination unit, a parameter determination unit and at least one collaborative robot, wherein the secondary determination unit is used to determine the anchor information based on the ore-feeding machine image, magnetic field information, sound information and vibration information, and based on the anchor information, adjust the working parameters of the image acquisition unit to obtain the target area image, and the target area image, determine whether there is an anchor, the parameter determination unit is used to determine the anchor sorting working parameters based on the target area image after the secondary determination unit determines that there is an anchor, and the collaborative robot is used to perform anchor sorting according to the anchor sorting working parameters; the ore-feeding machine control module is also used to control the ore-feeding machine to continue working after the secondary determination unit determines that there is no anchor, and is also used to control the ore-feeding machine to continue working after the collaborative robot completes anchor sorting.

[0006] Furthermore, the magnetic induction unit includes a plurality of magnetic induction devices arranged at different positions; the sound collection unit includes a plurality of sound collection devices arranged at different positions; the vibration induction unit includes a plurality of vibration collection devices arranged at different positions; the anchor rod sensing module is also used to determine the setting positions of the plurality of magnetic induction devices, the plurality of sound collection devices and the plurality of vibration collection devices, specifically including: setting a plurality of magnetic induction devices at a plurality of first initial positions, setting a plurality of sound collection devices at a plurality of second initial positions, and setting a plurality of vibration collection devices at a plurality of third initial positions; setting a plurality of test scenes, obtaining the collection data of the plurality of magnetic induction devices, the plurality of sound collection devices and the plurality of vibration collection devices in each test scene; based on the plurality of magnetic induction devices, the plurality of sound collection devices and the plurality of vibration collection devices are set; The device, multiple sound collection devices and multiple vibration sensing devices collect data in each test scene to determine the effective parameters of magnetic induction at each first initial position, the effective parameters of sound induction at each second initial position and the effective parameters of vibration induction at each third initial position; based on the effective parameters of magnetic induction at each first initial position, the multiple first initial positions are screened to determine the setting positions of the multiple magnetic induction devices; based on the effective parameters of sound induction at each second initial position, the multiple second initial positions are screened to determine the setting positions of the multiple sound collection devices; based on the effective parameters of vibration induction at each third initial position, the multiple third initial positions are screened to determine the setting positions of the multiple vibration sensing devices.

[0007] Furthermore, the image acquisition unit includes a parameter determination device, a visible light irradiation device and an image acquisition device, wherein the parameter determination device is used to determine the optimal visible light parameters, the optimal projection posture and the optimal image acquisition posture; the visible light irradiation device is used to project visible light to the ore-dropping machine according to the optimal visible light parameters and the optimal projection posture during the ore-dropping process; and the image acquisition device is used to collect images of the ore-dropping machine in the optimal image acquisition posture during the ore-dropping process.

[0008] Furthermore, the parameter determination device determines the optimal visible light parameters, the optimal projection posture and the optimal image acquisition posture, including: obtaining a plurality of visible light parameters, wherein the visible light parameter parameters at least include wavelength range, light intensity and light source type; obtaining ore optical characteristics of a plurality of sample ores under each visible light parameter; obtaining the optical characteristics of a plurality of sample anchors under each visible light parameter; determining a plurality of sample pairs based on the plurality of sample ores and the plurality of sample anchors, wherein the sample pairs include a sample ore and a sample anchor; for any sample pair, based on the ore optical characteristics of the sample ore included in the sample pair under each visible light parameter and the optical characteristics of the sample anchor under each visible light parameter, calculating the optical characteristic difference value of the sample ore and the sample anchor included in the sample pair under each visible light parameter, and determining the target visible light parameters of the sample pair; obtaining the optical characteristics of the current test ore under multiple test parameters. Try out the optical characteristics of the ore under visible light parameters; obtain the optical characteristics of the current test anchor under multiple visible light parameters; calculate the similarity value of each sample pair based on the ore optical characteristics of the current test ore under multiple visible light parameters and the optical characteristics of the current test anchor under multiple visible light parameters, and determine similar sample pairs from the multiple sample pairs; determine the optimal visible light parameters based on the target visible light parameters of the similar sample pairs; obtain multiple groups of posture combinations, wherein the posture combination includes a projection posture and an image acquisition posture; place the current test ore and the current test anchor on the ore placing machine; for each group of posture combinations, obtain the test image corresponding to the posture combination; based on the test image corresponding to each group of posture combinations, select the optimal posture combination from the multiple groups of posture combinations, wherein the optimal posture combination includes the optimal projection posture and the optimal image acquisition posture.

[0009] Furthermore, the anchor identification unit is used to determine whether an anchor exists during the mining process based on the mining machine image, magnetic field information, sound information and vibration information, including: determining the possibility of the existence of a first anchor based on the mining machine image; determining the possibility of the existence of a second anchor based on the magnetic field information, sound information and vibration information; and determining whether an anchor exists based on the possibility of the existence of the first anchor and the possibility of the existence of the second anchor.

[0010] Further, based on the ore-dropping machine image, determining the possibility of the existence of the first anchor includes: determining the possibility of the existence of the first anchor based on the ore-dropping machine image through an anchor detection model.

[0011] Further, based on the magnetic field information, sound information and vibration information, determining the possibility of the existence of the second anchor includes: calculating the induction correlation coefficient of any two magnetic induction devices based on the collected data of the multiple magnetic induction devices in each test scene; clustering the multiple magnetic induction devices based on the induction correlation coefficient of any two magnetic induction devices to determine multiple magnetic induction device clusters; calculating the induction correlation coefficient of any two sound collection devices based on the collected data of the multiple sound collection devices in each test scene; clustering the multiple sound collection devices based on the induction correlation coefficient of any two sound collection devices to determine multiple sound collection device clusters; calculating the induction correlation coefficient of any two vibration induction devices based on the collected data of the multiple vibration induction devices in each test scene; clustering the multiple vibration induction devices based on the induction correlation coefficient of any two vibration induction devices to determine multiple vibration induction device clusters; for any magnetic induction device cluster and any sound collection device cluster, calculating the induction correlation coefficient of the magnetic induction device cluster and the sound collection device cluster based on the collected data of the multiple magnetic induction devices and the multiple sound collection devices in each test scene; for any magnetic induction device cluster and any a vibration sensing device cluster, and based on the collected data of multiple magnetic sensing devices and multiple vibration sensing devices in each test scene, calculate the induction correlation coefficient of the magnetic sensing device cluster and the vibration sensing device cluster; for any sound collection device cluster and any vibration sensing device cluster, calculate the induction correlation coefficient of the sound collection device cluster and the vibration sensing device cluster based on the collected data of multiple sound collection devices and multiple vibration sensing devices in each test scene; based on the induction correlation coefficient of any magnetic sensing device cluster and any sound collection device cluster, the induction correlation coefficient of any magnetic sensing device cluster and any vibration sensing device cluster, and the induction correlation coefficient of any sound collection device cluster and any vibration sensing device cluster, determine a plurality of cluster units, wherein each of the cluster units includes a magnetic sensing device cluster, a sound collection device cluster and a vibration sensing device cluster; for each cluster unit, based on the magnetic field information collected by the magnetic sensing device, the sound information collected by the sound collection device and the vibration information collected by the vibration sensing device included in the cluster unit, determine the anchor risk value corresponding to the cluster unit; based on the anchor risk value corresponding to each cluster unit, determine the possibility of the existence of the second anchor.

[0012] Furthermore, anchor information is determined based on the ore-dropping machine image, magnetic field information, sound information and vibration information, including: predicting a first anchor position and a first anchor size based on the ore-dropping machine image; predicting a second anchor position and a second anchor size based on the magnetic field information, sound information and vibration information; determining an anchor predicted position based on the first anchor position and the second anchor position; determining an anchor predicted size based on the first anchor size and the second anchor size, wherein the anchor information includes the anchor predicted position and the anchor predicted size.

[0013] Furthermore, based on the anchor rod information, the working parameters of the image acquisition unit are adjusted, including: based on the anchor rod information, determining the secondary projection posture and the secondary image acquisition posture; adjusting the posture of the visible light irradiation device according to the secondary projection posture; and adjusting the posture of the image acquisition unit according to the secondary image acquisition posture.

[0014] Furthermore, the parameter determination unit determines the anchor rod sorting working parameters based on the target area image, including: determining the anchor rod position and anchor rod size according to the target area image; determining the anchor rod sorting working parameters based on the anchor rod position and anchor rod size, wherein the anchor rod sorting working parameters at least include a working path.

[0015] Compared with the prior art, the machine vision-based chute ore-drawing machine bolt sorting and intelligent iron removal system provided by the present invention has at least the following beneficial effects:

[0016] 1. By real-time monitoring of images, magnetic fields, sounds and vibration information during the ore-feeding process, the system can promptly detect and identify the presence of anchors. Once an anchor is found, the ore-feeding machine stops working immediately, effectively avoiding safety accidents such as equipment damage, production interruption and even casualties caused by undetected anchors. The system can quickly and accurately identify anchors and automatically resume the work of the ore-feeding machine after confirming that there are no anchors. This immediate response mechanism reduces production delays caused by manual inspection or downtime waiting for confirmation, thereby improving overall production efficiency. The collaborative robot in the intelligent iron removal module performs anchor sorting work, reducing the need for manual intervention. This not only reduces labor costs, but also avoids equipment damage or personal injury that may be caused by improper manual operation, thereby reducing long-term maintenance costs. The system integrates a variety of sensors and intelligent control algorithms to achieve comprehensive monitoring and intelligent decision-making of the ore-feeding process. This highly automated and intelligent production method improves the controllability and stability of the production process, and provides strong support for the digital transformation and intelligent upgrading of mining enterprises.

[0017] 2. By setting up sensors at multiple initial positions and collecting data based on test scenarios, the performance of the sensors at each position can be evaluated. The best selected positions ensure that the sensors can capture the most accurate and relevant data, thereby improving the accuracy of anchor detection. Multi-position sensor layouts can provide more comprehensive information and reduce blind spots. This means that no matter from which direction or angle the anchor appears, it is likely to be detected by at least one sensor, thereby enhancing the ability to identify the anchor. By testing and screening to determine the sensor position, the effectiveness of each sensor can be maximized and redundant and unnecessary sensor configurations can be reduced. This helps reduce system complexity and cost while improving the efficiency and reliability of the overall system.

[0018] 3. By testing the optical characteristics of the ore and the anchor under various visible light parameters and calculating the difference between them, the optimal visible light parameters that can maximize the difference in optical characteristics between the ore and the anchor can be accurately found. This helps to distinguish the ore and the anchor more clearly during the image acquisition process, thereby improving the accuracy of anchor identification. By testing the test images under different pose combinations, the optimal projection pose and image acquisition pose can be screened out. These optimal poses can ensure that the clarity and contrast of the ore and the anchor in the image are optimal, thereby further improving the accuracy and reliability of image analysis. Since a variety of sample ores and sample anchors are used in the test process, and a variety of visible light parameters and pose combinations are considered, the determined optimal parameters and pose combinations have stronger adaptability and generalization capabilities. This means that the system can still maintain a high recognition accuracy and stability when facing ores and anchors of different types, shapes, and materials. By optimizing the visible light parameters and image acquisition poses, problems such as image blur and shadows caused by insufficient lighting and poor angles can be reduced, thereby reducing the false alarm rate and missed alarm rate of anchor identification. This is of great significance for improving the safety and efficiency of the ore-drawing process. The entire parameter determination process is highly automated, reducing the impact of manual intervention and subjective judgment. This helps to achieve automation and intelligence in bolt identification, improve production efficiency and reduce labor costs;

[0019] 4. By clustering the collected data of multiple sensors (magnetic induction device, sound collection device, vibration induction device), multiple cluster units can be formed. The sensors in each cluster unit are relatively close in spatial position, so their responses to the same anchor will be more consistent and sensitive. This helps to improve the detection sensitivity of the presence of anchors. By calculating the induction correlation coefficient between different sensor clusters and combining sensor clusters with strong correlation into cluster units, the fusion of multi-source data is realized. This data fusion method can comprehensively utilize the advantages of different sensors and improve the accuracy and reliability of the overall detection system. Since multiple sensor data and the correlation between them are considered, this method can more accurately identify the specific signal pattern caused by the anchor, thereby reducing the false alarm rate caused by environmental factors, equipment noise, etc. The cluster units formed by cluster analysis can roughly determine the area where the anchor may exist. Further combined with the sensor data in each cluster unit, a more accurate estimation of the anchor position can be achieved, which is of great significance for the subsequent anchor processing and maintenance. The sensor data is processed in real time and the possibility of the existence of the anchor is quickly calculated. This helps to achieve real-time monitoring and early warning of the ore discharge process and improve production safety and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0021] Figure 1 It is a module schematic diagram of a machine vision-based ore chute bolt sorting and intelligent iron removal system according to some embodiments of this specification;

[0022] Figure 2 It is a schematic diagram of a process for determining optimal visible light parameters, optimal projection posture and optimal image acquisition posture according to some embodiments of this specification;

[0023] Figure 3 It is a schematic diagram of a process for determining the possibility of the existence of a second anchor according to some embodiments of this specification. DETAILED DESCRIPTION

[0024] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0025] Figure 1 is a module schematic diagram of a machine vision-based chute ore-dropping machine bolt sorting and intelligent iron removal system according to some embodiments of this specification, such as Figure 1 As shown, the machine vision-based chute ore-feeding machine anchor rod sorting and intelligent iron removal system may include an anchor rod sensing module, an ore-feeding machine control module and an intelligent iron removal module.

[0026] The anchor rod sensing module may include an image acquisition unit, a magnetic induction unit, a sound acquisition unit, a vibration sensing unit and an anchor rod identification unit, wherein the image acquisition unit is used to acquire images of the ore-dropping machine during the ore-dropping process, the magnetic induction unit, the sound acquisition unit and the vibration sensing unit are arranged on the ore-dropping machine, the magnetic induction unit is used to acquire magnetic field information during the ore-dropping process, the sound acquisition unit is used to acquire sound information during the ore-dropping process, the vibration sensing unit is used to acquire vibration information during the ore-dropping process, and the anchor rod identification unit is used to determine whether an anchor rod exists during the ore-dropping process based on the ore-dropping machine image, magnetic field information, sound information and vibration information.

[0027] The magnetic induction unit includes multiple magnetic induction devices (e.g., Hall sensors, magnetoresistive sensors, fluxgate sensors, etc.) arranged at different positions. The magnetic induction device detects changes in the magnetic field by using the principle that the current in the conductor generates a voltage difference under the action of the magnetic field. When an iron object approaches the magnetic induction device, it changes the magnetic field distribution around it, thereby generating a voltage difference, which is then detected by the sensor.

[0028] The sound collection unit includes multiple sound collection devices (e.g., microphones, etc.) set at different locations. The sound characteristics of ore are affected by multiple factors such as its composition, structure, density, etc., and are usually manifested as more complex sound signals containing multiple frequency components and noise. As a metal product, the sound characteristics of anchor rods are relatively simple, mainly manifested as the sound produced by metal impact or vibration, with specific frequency and waveform characteristics. Therefore, sound detection can be used to assist in determining whether the anchor rod exists.

[0029] The vibration sensing unit includes multiple vibration sensing devices (e.g., vibration sensors, etc.) arranged at different positions. Vibration information detection is a non-destructive detection method based on physical properties. It uses the difference in vibration characteristics between the anchor and the ore to identify the presence of the anchor. When the anchor is embedded in the ore, its vibration characteristics (such as vibration frequency, amplitude, etc.) will be affected by factors such as the material, shape, size, and embedding method of the anchor, resulting in differences in vibration characteristics with the surrounding ore, thereby affecting the vibration of the ore dropper.

[0030] Preferably, the anchor rod sensing module is also used to determine the setting positions of multiple magnetic sensing devices, multiple sound collection devices and multiple vibration sensing devices, specifically including:

[0031] A plurality of magnetic induction devices are arranged at a plurality of first initial positions, a plurality of sound collection devices are arranged at a plurality of second initial positions, and a plurality of vibration induction devices are arranged at a plurality of third initial positions;

[0032] Setting multiple test scenarios, obtaining data collected by multiple magnetic induction devices, multiple sound collection devices, and multiple vibration induction devices in each test scenario;

[0033] Determine the effective magnetic induction parameters of each first initial position, the effective sound induction parameters of each second initial position, and the effective vibration induction parameters of each third initial position based on the collected data of the multiple magnetic induction devices, the multiple sound collection devices, and the multiple vibration induction devices in each test scene;

[0034] Based on the effective magnetic induction parameter of each first initial position, multiple first initial positions are screened to determine the setting positions of multiple magnetic induction devices. For example, the first initial position whose effective magnetic induction parameter is greater than the effective magnetic induction parameter threshold can be used as the setting position of the magnetic induction device.

[0035] Based on the sound sensing effective parameter of each second initial position, multiple second initial positions are screened to determine the setting positions of multiple sound collection devices. For example, the second initial position whose sound sensing effective parameter is greater than the sound sensing effective parameter threshold can be used as the setting position of the sound collection device;

[0036] Based on the vibration sensing effective parameter of each third initial position, multiple third initial positions are screened to determine the setting positions of multiple vibration sensing devices. For example, the third initial position where the vibration sensing effective parameter is greater than the vibration sensing effective parameter threshold can be used as the setting position of the vibration sensing device.

[0037] Specifically, the multiple first initial positions may be positions on the ore-dropping machine that are easy to sense changes in magnetic field distribution, and the multiple first initial positions may be determined by artificial or machine learning models. The multiple second initial positions may be positions on the ore-dropping machine that are easy to sense changes in sound, and the multiple second initial positions may be determined by artificial or machine learning models. The multiple third initial positions may be positions on the ore-dropping machine that are easy to sense changes in vibration, and the multiple third initial positions may be determined by artificial or machine learning models.

[0038] The test scenario may be a scenario simulating ore in a ore-dropping machine being embedded in an anchor rod for ore dropping. In different test scenarios, at least one of the material, shape, size, position and embedding method of the anchor rod may be different.

[0039] It is possible to first obtain magnetic field distribution information of multiple first initial positions, sound information of multiple second initial positions and vibration information of multiple third initial positions in a scenario where the anchor is not embedded for ore dropping.

[0040] For each test scenario, it is possible to determine whether the first initial position detects an anchor based on the magnetic field distribution information of the first initial position in the test scenario and the magnetic field distribution information of the first initial position in the scenario where the anchor is not embedded and the ore is being released. For example, if the difference between the magnetic field distribution information of the first initial position in the test scenario and the magnetic field distribution information of the first initial position in the scenario where the anchor is not embedded and the ore is being released is large, it is determined that the first initial position detects an anchor. It is possible to determine whether the second initial position detects an anchor based on the sound information of the second initial position in the test scenario and the sound information of the second initial position in the scenario where the anchor is not embedded and the ore is being released. For example, if the difference between the sound information of the second initial position in the test scenario and the sound information of the second initial position in the scenario where the anchor is not embedded and the ore is being released is large, it is determined that the second initial position detects an anchor. Whether the third initial position detects an anchor rod can be determined based on the vibration information of the third initial position in the test scenario and the vibration information of the third initial position in the scenario where the anchor rod is not embedded and the mining is being released. For example, if the difference between the vibration information of the third initial position in the test scenario and the vibration information of the third initial position in the scenario where the anchor rod is not embedded and the mining is being released is large, then it is determined that the third initial position detects an anchor rod.

[0041] The effective magnetic induction parameters at the first initial position can be calculated according to the following formula:

[0042]

[0043] in, is the effective parameter of magnetic induction at the i-th first initial position, The number of test scenes where anchors were detected for the i-th first initial position, is the total number of test scenarios.

[0044] The calculation method of the sound induction effective parameters of the second initial position and the vibration induction effective parameters of the third initial position is similar to the calculation method of the magnetic induction effective parameters of the first initial position, and will not be repeated here.

[0045] The image acquisition unit includes a parameter determination device, a visible light irradiation device and an image acquisition device, wherein the parameter determination device is used to determine the optimal visible light parameters, the optimal projection posture and the optimal image acquisition posture; the visible light irradiation device is used to project visible light to the ore-dropping machine according to the optimal visible light parameters and the optimal projection posture during the ore-dropping process; and the image acquisition device is used to collect images of the ore-dropping machine in the optimal image acquisition posture during the ore-dropping process.

[0046] Figure 2is a schematic diagram of a process for determining optimal visible light parameters, optimal projection posture, and optimal image acquisition posture according to some embodiments of this specification, such as Figure 2 As shown, as a preferred embodiment, the parameter determination device determines the optimal visible light parameter, the optimal projection posture and the optimal image acquisition posture, including:

[0047] Acquire multiple visible light parameters, wherein the visible light parameters at least include wavelength range, light intensity and light source type;

[0048] Obtain the optical characteristics of various sample ores under each visible light parameter;

[0049] Obtain the optical characteristics of various sample anchors under each visible light parameter;

[0050] Based on a plurality of sample ores and a plurality of sample bolts, a plurality of sample pairs are determined, wherein the sample pairs include a sample ore and a sample bolt;

[0051] For any sample pair, based on the ore optical characteristics of the sample ore included in the sample pair under each visible light parameter and the bolt optical characteristics of the sample bolt under each visible light parameter, the optical characteristic difference value of the sample ore and the sample bolt included in the sample pair under each visible light parameter is calculated to determine the target visible light parameter of the sample pair;

[0052] Obtaining the optical characteristics of the current test ore under various test visible light parameters;

[0053] Obtaining the optical characteristics of the current test anchor under various test visible light parameters;

[0054] Based on the ore optical characteristics of the current test ore under multiple test visible light parameters and the anchor optical characteristics of the current test anchor under multiple test visible light parameters, a similarity value of each sample pair is calculated, and similar sample pairs are determined from multiple sample pairs;

[0055] Determining optimal visible light parameters based on target visible light parameters of similar sample pairs;

[0056] Acquire multiple groups of posture combinations, wherein the posture combinations include projection postures and image acquisition postures;

[0057] Placing the current test ore and the current test anchor on the ore draw machine;

[0058] For each pose combination, obtain the test image corresponding to the pose combination;

[0059] Based on the test images corresponding to each group of posture combinations, the optimal posture combination is selected from multiple groups of posture combinations, wherein the optimal posture combination includes the optimal projection posture and the optimal image acquisition posture.

[0060] Specifically, the sample ore color image of the sample ore under the visible light parameter can be collected, and based on the sample ore color image of the sample ore under the visible light parameter, the ore optical characteristics of the sample ore under the visible light parameter can be determined, wherein the ore optical characteristics can include color characteristics, brightness characteristics, etc. The color characteristics of the sample ore under the visible light parameter can be calculated based on the RGB value of each pixel in the sample ore color image of the sample ore under the visible light parameter, wherein the color characteristics include the RGB value mean and the RGB value standard deviation, etc. The sample ore color image of the sample ore under the visible light parameter is converted into a sample ore grayscale image, and the brightness characteristics of the sample ore under the visible light parameter are calculated based on the grayscale value of each pixel in the sample ore grayscale image, wherein the brightness characteristics can include the frequency of occurrence of each grayscale level, etc.

[0061] The optical characteristics of the sample anchor rod under the visible light parameters can be determined by referring to the above method, which will not be described in detail here.

[0062] The optical characteristic difference value of the sample ore and sample anchor included in the sample pair under each visible light parameter can be calculated according to the following formula:

[0063]

[0064] in, is the optical characteristic difference value of the sample ore and sample anchor included in the i-th sample pair under the j-th visible light parameter, and is the preset weight, and are greater than 0, , is the e-th color feature of the sample ore included in the i-th sample pair under the j-th visible light parameter, is the e-th color feature of the sample anchor included in the i-th sample pair under the j-th visible light parameter, is the total number of color features, is the frequency of occurrence of the kth gray level of the sample ore included in the i-th sample pair under the j-th visible light parameter, is the frequency of occurrence of the kth gray level of the sample anchor included in the i-th sample pair under the j-th visible light parameter, is the total number of gray levels.

[0065] Based on the optical characteristic difference values ​​of the sample ore and the sample anchor included in the sample pair under each visible light parameter, the visible light parameter with the largest optical characteristic difference value is taken as the target visible light parameter of the sample pair.

[0066] The current test ore may be the ore currently being drawn, and the current test anchor may be an anchor that may be mixed in the ore currently being drawn. It is understandable that before a batch of ore is drawn, a portion of the ore may be extracted from the batch as the current test ore. The anchors that may be mixed in the ore currently being drawn may be predicted by artificial or machine learning models, and the anchors that may be mixed in the ore currently being drawn predicted from multiple sample anchors may be used as the current test anchors.

[0067] The multiple test visible light parameters may be visible light parameters sampled from the multiple visible light parameters. For example, the multiple visible light parameters include visible light parameter 1, visible light parameter 2, visible light parameter 3, visible light parameter 4, and visible light parameter 5, and visible light parameter 2 and visible light parameter 3 may be extracted as test visible light parameters.

[0068] The optical characteristics of the ore under various test visible light parameters and the optical characteristics of the anchor under various test visible light parameters can be determined by referring to the above method, which will not be repeated here.

[0069] The similarity value of a sample pair can be calculated according to the following formula:

[0070]

[0071] in, is the similarity value of the g-th sample pair, is the similarity value of the visible light parameter for the g-th sample pair in the h-th test, is the total number of visible light parameters tested, is the preset parameter, greater than 0, , , , , and is the preset weight, , , , , and are greater than 0, , , , is the difference value of the ore optical characteristics between the ore optical characteristics of the sample ore of the g-th sample pair under the h-th test visible light parameter and the ore optical characteristics of the current test ore under the h-th test visible light parameter, is the difference value of the anchor optical characteristics between the anchor optical characteristics of the sample anchor of the g-th sample pair under the h-th test visible light parameter and the anchor optical characteristics of the current test anchor under the h-th test visible light parameter, is the e-th color feature of the sample ore of the g-th sample pair under the h-th test visible light parameter, is the eth color characteristic of the current test ore under the hth test visible light parameter, is the frequency of occurrence of the kth gray level of the sample ore of the gth sample pair under the hth test visible light parameter, is the frequency of the kth gray level of the current test ore under the hth test visible light parameter, is the e-th color feature of the sample anchor of the g-th sample pair under the h-th test visible light parameter, is the e-th color feature of the current test anchor under the h-th test visible light parameter, is the frequency of occurrence of the kth gray level of the sample anchor of the gth sample pair under the hth test visible light parameter, is the frequency of occurrence of the kth gray level of the current test anchor under the hth test visible light parameter.

[0072] Similar sample pairs whose similarity is greater than the similarity value threshold are regarded as similar sample pairs.

[0073] Based on the test image corresponding to each group of posture combinations, the score of each group of posture combinations in multiple screening indicators can be determined, the scores of multiple screening indicators are weighted and summed, and the posture combination with the highest score after the weighted sum is taken as the optimal posture combination, wherein the multiple screening indicators may include an image clarity index, a target recognition rate index, a feature integrity index, a lighting uniformity index and a background interference index. The image clarity index can be used to evaluate the clarity of the anchor in the test image, the target recognition rate index can be used to evaluate the recognition accuracy of the ore and the anchor in the image, the feature integrity index can be used to evaluate the degree of retention of the key features of the anchor in the test image (such as shape, texture, size, etc.), the lighting uniformity index can be used to evaluate the uniformity of lighting in the test image, the proportion of shadows or reflections, and the background interference index can be used to evaluate the interference degree of the background in the test image on the target recognition.

[0074] Preferably, the bolt identification unit is used to determine whether there is a bolt based on the ore-dropping machine image, magnetic field information, sound information and vibration information during the ore-dropping process, including:

[0075] Based on the ore-drawing machine image, determining the possibility of the first anchor bolt existing;

[0076] Determining the possibility of the second anchor bolt existing based on the magnetic field information, the sound information and the vibration information;

[0077] Based on the first anchor existence possibility and the second anchor existence possibility, it is determined whether an anchor exists.

[0078] For example, a weighted sum of the first anchor rod existence possibility and the second anchor rod existence possibility may be obtained to obtain a comprehensive anchor rod existence possibility. When the comprehensive anchor rod existence possibility is greater than a comprehensive anchor rod existence possibility threshold, it is determined that an anchor rod exists.

[0079] Preferably, based on the ore-drawing machine image, determining the possibility of the first anchor bolt existence includes:

[0080] Through the anchor detection model, based on the ore-dropping machine image, the possibility of the existence of the first anchor is determined, wherein the anchor detection model can be a convolutional neural network model.

[0081] Figure 3 is a schematic diagram of a process for determining the possibility of the existence of a second anchor according to some embodiments of this specification, such as Figure 3 As shown, preferably, determining the possibility of the existence of the second anchor based on the magnetic field information, the sound information and the vibration information includes:

[0082] Based on the collected data of the multiple magnetic induction devices in each test scene, calculating the induction correlation coefficient of any two magnetic induction devices;

[0083] Based on the induction correlation coefficients of any two magnetic induction devices, clustering the multiple magnetic induction devices to determine multiple magnetic induction device clusters;

[0084] Based on the collected data of the multiple sound collection devices in each test scene, calculating the induction correlation coefficient of any two sound collection devices;

[0085] Based on the induction correlation coefficient of any two sound collection devices, clustering the multiple sound collection devices to determine multiple sound collection device clusters;

[0086] Based on the collected data of the multiple vibration sensing devices in each test scene, calculating the sensing correlation coefficient of any two vibration sensing devices;

[0087] Based on the induction correlation coefficients of any two vibration sensing devices, clustering the multiple vibration sensing devices to determine multiple vibration sensing device clusters;

[0088] For any magnetic induction device cluster and any sound collection device cluster, based on the collected data of multiple magnetic induction devices and multiple sound collection devices in each test scene, calculate the induction correlation coefficient of the magnetic induction device cluster and the sound collection device cluster;

[0089] For any magnetic induction device cluster and any vibration induction device cluster, based on the collected data of multiple magnetic induction devices and multiple vibration induction devices in each test scene, calculate the induction correlation coefficient of the magnetic induction device cluster and the vibration induction device cluster;

[0090] For any sound collection device cluster and any vibration sensing device cluster, based on the collected data of multiple sound collection devices and multiple vibration sensing devices in each test scene, calculate the induction correlation coefficient of the sound collection device cluster and the vibration sensing device cluster;

[0091] Determine a plurality of cluster units based on an induction correlation coefficient between any magnetic induction device cluster and any sound collection device cluster, an induction correlation coefficient between any magnetic induction device cluster and any vibration induction device cluster, and an induction correlation coefficient between any sound collection device cluster and any vibration induction device cluster, wherein each cluster unit includes a magnetic induction device cluster, a sound collection device cluster, and a vibration induction device cluster;

[0092] For each cluster unit, based on the magnetic field information collected by the magnetic induction device included in the cluster unit, the sound information collected by the sound collection device, and the vibration information collected by the vibration induction device, the anchor risk value corresponding to the cluster unit is determined;

[0093] Based on the anchor risk value corresponding to each cluster unit, the possibility of the second anchor existence is determined.

[0094] Specifically, for each test scenario, the Pearson correlation coefficient of any two magnetic induction devices in the test scenario can be calculated based on the collected data of any two magnetic induction devices in the test scenario. The Pearson correlation coefficients of any two magnetic induction devices in each test scenario are weighted and summed to obtain the induction correlation coefficients of the two magnetic induction devices.

[0095] A clustering algorithm (eg, K-means clustering algorithm, hierarchical clustering algorithm, etc.) may be used to cluster multiple magnetic induction devices based on the induction correlation coefficients of any two magnetic induction devices to determine multiple magnetic induction device clusters.

[0096] The inductive correlation coefficient of any two sound collection devices may be calculated, multiple sound collection device clusters may be determined, the inductive correlation coefficient of any two vibration sensing devices may be calculated, and multiple vibration sensing device clusters may be determined, and will not be described in detail here.

[0097] For each test scenario, the Pearson correlation coefficient of any magnetic induction device and any sound collection device in the test scenario can be calculated based on the collected data of any magnetic induction device and any sound collection device in the test scenario. The Pearson correlation coefficient of any magnetic induction device and any sound collection device in each test scenario is weighted and summed to obtain the induction correlation coefficient of the magnetic induction device and the sound collection device.

[0098] For any magnetic induction device cluster and any sound collection device cluster, the average of the induction correlation coefficients of any magnetic induction device included in the magnetic induction device cluster and any sound collection device included in the sound collection device cluster can be calculated as the induction correlation coefficient of the magnetic induction device cluster and the sound collection device cluster.

[0099] The induction correlation coefficients of the magnetic induction device cluster and the vibration induction device cluster and the induction correlation coefficients of the sound collection device cluster and the vibration induction device cluster can be calculated by referring to the above method, which will not be repeated here.

[0100] When there is a magnetic induction device cluster, a sound collection device cluster and a vibration induction device cluster, and the induction correlation coefficient between the magnetic induction device cluster and the sound collection device cluster is greater than the induction correlation coefficient threshold, the induction correlation coefficient between the vibration induction device cluster and the sound collection device cluster is greater than the induction correlation coefficient threshold, and the induction correlation coefficient between the vibration induction device cluster and the magnetic induction device cluster is greater than the induction correlation coefficient threshold, then the magnetic induction device cluster, the sound collection device cluster and the vibration induction device cluster can be divided into one cluster unit.

[0101] For each cluster unit, a unit risk prediction model corresponding to the cluster unit can be established. The unit risk prediction model corresponding to the cluster unit can be used to determine the anchor risk value corresponding to the cluster unit based on the magnetic field information collected by the magnetic induction device included in the cluster unit, the sound information collected by the sound collection device, and the vibration information collected by the vibration sensing device. The unit risk prediction model can be a long short-term memory network model.

[0102] Based on the anchor risk value corresponding to each cluster unit, the maximum anchor risk value can be taken as the possibility of the second anchor existence.

[0103] The ore-dropping machine control module can be used to control the ore-dropping machine to stop working after the anchor rod identification unit determines that there is an anchor rod.

[0104] The intelligent iron removal module may include a secondary judgment unit, a parameter determination unit and at least one collaborative robot, wherein the secondary judgment unit is used to determine the anchor information based on the ore-dropping machine image, magnetic field information, sound information and vibration information, and based on the anchor information, adjust the working parameters of the image acquisition unit to obtain the target area image, and the target area image to determine whether there is an anchor. The parameter determination unit is used to determine the anchor sorting working parameters based on the target area image after the secondary judgment unit determines that there is an anchor, and the collaborative robot is used to perform anchor sorting work according to the anchor sorting working parameters.

[0105] Preferably, the secondary determination unit determines the anchor information based on the ore-dropping machine image, magnetic field information, sound information and vibration information, including:

[0106] Based on the ore drawer image, predict the first anchor position and the first anchor size;

[0107] Predicting the position of the second anchor and the size of the second anchor based on the magnetic field information, the sound information and the vibration information;

[0108] Determining a predicted anchor position based on the first anchor position and the second anchor position;

[0109] Based on the first anchor rod size and the second anchor rod size, the predicted anchor rod size is determined, wherein the anchor rod information includes the predicted anchor rod position and the predicted anchor rod size.

[0110] Specifically, the first anchor position and the first anchor size can be predicted based on the ore-dropping machine image by a first prediction model, wherein the first prediction model can be a convolutional neural network model. As an example only, first, the ore-dropping machine image is preprocessed, including denoising, contrast enhancement, edge detection, etc., so as to better identify the anchor in the ore-dropping machine image. The features of the anchor, such as shape, color, texture, etc., are extracted from the preprocessed ore-dropping machine image using image processing techniques (such as template matching, shape recognition, machine learning algorithms, etc.). Position prediction: Based on the extracted features, an image recognition algorithm (such as a convolutional neural network CNN) is used to predict the position of the anchor in the ore-dropping machine image. Once the position of the anchor is determined, the actual size of the anchor can be estimated by measuring the size of the bounding box.

[0111] The cluster unit whose anchor risk value is greater than the anchor risk value threshold can be used as the target cluster unit.

[0112] For each cluster unit, a second prediction model corresponding to the cluster unit may be established, wherein the second prediction model may be a long short-term memory network model.

[0113] For each target cluster unit, the second anchor position and the second anchor size can be predicted by the second prediction model corresponding to the target cluster unit based on the magnetic field information collected by each magnetic induction device included in the target cluster unit, the sound information collected by the sound collection device, and the vibration information collected by the vibration sensing device.

[0114] The first anchor position and the second anchor position are fused by weighted average, Kalman filter, particle filter and other methods to determine the predicted anchor position. The first anchor size and the second anchor size are fused by weighted average, Kalman filter, particle filter and other methods to determine the predicted anchor size.

[0115] Preferably, the parameter determination unit adjusts the working parameters of the image acquisition unit based on the anchor information, including:

[0116] Based on the anchor information, determine the secondary projection posture and secondary image acquisition posture;

[0117] According to the secondary projection posture, adjusting the posture of the visible light irradiation device;

[0118] According to the secondary image acquisition posture, the posture of the image acquisition unit is adjusted.

[0119] Specifically, the secondary projection pose and the secondary image acquisition pose can be determined based on the anchor information through a pose generation model, wherein the pose generation model can be a generative adversarial network model.

[0120] Preferably, the parameter determination unit determines the anchor sorting working parameters based on the target area image, including:

[0121] Determine the anchor position and anchor size based on the target area image;

[0122] Based on the anchor position and anchor size, anchor sorting working parameters are determined, wherein the anchor sorting working parameters at least include a working path.

[0123] Specifically, the anchor position and anchor size can be determined according to the target area image through a target recognition model, wherein the target recognition model can be a convolutional neural network model.

[0124] A working path may be generated based on anchor positions and anchor sizes through a path generation model, wherein the path generation model may be a generative adversarial network model.

[0125] The ore-feeding machine control module is also used to control the ore-feeding machine to continue working after the secondary determination unit determines that there is no anchor bolt, and is also used to control the ore-feeding machine to continue working after the collaborative robot completes anchor bolt sorting.

[0126] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. The machine vision-based chute ore-drawing machine bolt sorting and intelligent iron removal system is characterized by: include: An anchor sensing module, comprising an image acquisition unit, a magnetic induction unit, a sound acquisition unit, a vibration sensing unit and an anchor identification unit, wherein the image acquisition unit is used to acquire an image of a ore-dropping machine during ore-dropping, the magnetic induction unit, the sound acquisition unit and the vibration sensing unit are arranged on the ore-dropping machine, the magnetic induction unit is used to acquire magnetic field information during ore-dropping, the sound acquisition unit is used to acquire sound information during ore-dropping, the vibration sensing unit is used to acquire vibration information during ore-dropping, and the anchor identification unit is used to determine whether an anchor exists during ore-dropping based on the image of the ore-dropping machine, the magnetic field information, the sound information and the vibration information; A ore-dropping machine control module, used to control the ore-dropping machine to stop working after the anchor rod identification unit determines that there is an anchor rod; An intelligent iron removal module, comprising a secondary determination unit, a parameter determination unit and at least one collaborative robot, wherein the secondary determination unit is used to determine the anchor information based on the ore-dropping machine image, magnetic field information, sound information and vibration information, and based on the anchor information, adjust the working parameters of the image acquisition unit to obtain the target area image, and determine whether there is an anchor according to the target area image; the parameter determination unit is used to determine the anchor sorting working parameters based on the target area image after the secondary determination unit determines that there is an anchor, and the collaborative robot is used to perform anchor sorting according to the anchor sorting working parameters; The ore-dropping machine control module is also used to control the ore-dropping machine to continue working after the secondary determination unit determines that there is no anchor, and is also used to control the ore-dropping machine to continue working after the collaborative robot completes the anchor sorting; The image acquisition unit includes a parameter determination device, a visible light irradiation device and an image acquisition device, wherein the parameter determination device is used to determine the optimal visible light parameters, the optimal projection posture and the optimal image acquisition posture, the visible light irradiation device is used to project visible light to the ore-dropping machine according to the optimal visible light parameters and the optimal projection posture during the ore-dropping process, and the image acquisition device is used to collect images of the ore-dropping machine in the optimal image acquisition posture during the ore-dropping process; The parameter determination device determines the optimal visible light parameter, the optimal projection posture and the optimal image acquisition posture, including: Acquire multiple visible light parameters, wherein the visible light parameters at least include wavelength range, light intensity and light source type; Obtain the optical characteristics of various sample ores under each visible light parameter; Obtain the optical characteristics of various sample anchors under each visible light parameter; Based on the plurality of sample ores and the plurality of sample bolts, a plurality of sample pairs are determined, wherein the sample pairs include one sample ore and one sample bolt; For any sample pair, based on the ore optical characteristics of the sample ore and the bolt optical characteristics of the sample bolt included in the sample pair under each visible light parameter, calculate the optical characteristic difference value of the sample ore and the sample bolt included in the sample pair under each visible light parameter, and determine the target visible light parameter of the sample pair; Obtaining the optical characteristics of the current test ore under various test visible light parameters; Obtaining the optical characteristics of the current test anchor under various test visible light parameters; Based on the ore optical characteristics of the current test ore under multiple test visible light parameters and the anchor optical characteristics of the current test anchor under multiple test visible light parameters, calculate the similarity value of each of the sample pairs, and determine similar sample pairs from the multiple sample pairs; Determining the optimal visible light parameter based on the target visible light parameter of the similar sample pair; Acquire multiple groups of posture combinations, wherein the posture combinations include projection postures and image acquisition postures; Placing the current test ore and the current test anchor on the ore drawer; For each group of posture combinations, obtaining a test image corresponding to the posture combination; Based on the test images corresponding to each group of the posture combinations, an optimal posture combination is selected from the multiple groups of posture combinations, wherein the optimal posture combination includes an optimal projection posture and an optimal image acquisition posture; Specifically, by collecting a sample ore color image of the sample ore under visible light parameters, based on the sample ore color image of the sample ore under the visible light parameters, determining the ore optical characteristics of the sample ore under the visible light parameters, wherein the ore optical characteristics include color characteristics and brightness characteristics, based on the RGB value of each pixel in the sample ore color image of the sample ore under the visible light parameters, calculating the color characteristics of the sample ore under the visible light parameters, wherein the color characteristics include the RGB value mean and the RGB value standard deviation, converting the sample ore color image of the sample ore under the visible light parameters into a sample ore grayscale image, and based on the grayscale value of each pixel in the sample ore grayscale image, calculating the brightness characteristics of the sample ore under the visible light parameters, wherein the brightness characteristics include the frequency of occurrence of each grayscale level; The optical characteristic difference value of the sample ore and sample anchor included in the sample pair under each visible light parameter is calculated according to the following formula: in, is the optical characteristic difference value of the sample ore and sample anchor included in the i-th sample pair under the j-th visible light parameter, and is the preset weight, and are greater than 0, , is the e-th color feature of the sample ore included in the i-th sample pair under the j-th visible light parameter, is the e-th color feature of the sample anchor included in the i-th sample pair under the j-th visible light parameter, is the total number of color features, is the frequency of occurrence of the kth gray level of the sample ore included in the i-th sample pair under the j-th visible light parameter, is the frequency of occurrence of the kth gray level of the sample anchor included in the i-th sample pair under the j-th visible light parameter, is the total number of gray levels; Based on the optical characteristic difference values ​​of the sample ore and the sample anchor included in the sample pair under each visible light parameter, the visible light parameter with the largest optical characteristic difference value is used as the target visible light parameter of the sample pair; The similarity value of the sample pair is calculated according to the following formula: in, is the similarity value of the g-th sample pair, is the similarity value of the visible light parameter for the h-th test for the g-th sample pair, is the total number of visible light parameters tested, is the preset parameter, greater than 0, , , , , and is the preset weight, , , , , and are greater than 0, , , , is the difference value of the ore optical characteristics between the ore optical characteristics of the sample ore of the g-th sample pair under the h-th test visible light parameter and the ore optical characteristics of the current test ore under the h-th test visible light parameter, is the difference value of the anchor optical characteristics between the anchor optical characteristics of the sample anchor of the g-th sample pair under the h-th test visible light parameter and the anchor optical characteristics of the current test anchor under the h-th test visible light parameter, is the e-th color feature of the sample ore of the g-th sample pair under the h-th test visible light parameter, is the eth color characteristic of the current test ore under the hth test visible light parameter, is the frequency of occurrence of the kth gray level of the sample ore of the gth sample pair under the hth test visible light parameter, is the frequency of the kth gray level of the current test ore under the hth test visible light parameter, is the e-th color feature of the sample anchor of the g-th sample pair under the h-th test visible light parameter, is the e-th color feature of the current test anchor under the h-th test visible light parameter, is the frequency of occurrence of the kth gray level of the sample anchor of the gth sample pair under the hth test visible light parameter, is the frequency of occurrence of the kth gray level of the current test anchor under the hth test visible light parameter; The similar sample pairs whose similarity is greater than the similarity value threshold are regarded as similar sample pairs; Based on the test images corresponding to each group of posture combinations, the scores of each group of posture combinations in multiple screening indicators are determined, and the scores of multiple screening indicators are weighted and summed, and the posture combination with the highest score after the weighted sum is taken as the optimal posture combination, wherein the multiple screening indicators include image clarity index, target recognition rate index, feature integrity index, illumination uniformity index and background interference index; The anchor rod identification unit is used to determine whether an anchor rod exists based on the image, magnetic field information, sound information and vibration information of the ore-dropping machine during the ore-dropping process, including: Based on the ore drawer image, determining the possibility of the first anchor bolt existing; Determining the possibility of the existence of the second anchor based on the magnetic field information, the sound information and the vibration information; Based on the possibility that the first anchor exists and the possibility that the second anchor exists, determining whether an anchor exists; Wherein, based on the ore-drawing machine image, determining the possibility of the first anchor bolt existence includes: Determining the possibility of the first anchor bolt existing based on the ore-drawing machine image by means of an anchor bolt detection model; Determining the possibility of the second anchor rod existing based on the magnetic field information, the sound information and the vibration information includes: Based on the collected data of the multiple magnetic induction devices in each test scene, calculating the induction correlation coefficient of any two magnetic induction devices; Based on the induction correlation coefficients of any two magnetic induction devices, clustering the multiple magnetic induction devices to determine multiple magnetic induction device clusters; Based on the collected data of the multiple sound collection devices in each test scene, calculating the induction correlation coefficient of any two sound collection devices; Based on the induction correlation coefficient of any two sound collection devices, clustering the multiple sound collection devices to determine multiple sound collection device clusters; Based on the collected data of the multiple vibration sensing devices in each test scene, calculating the sensing correlation coefficient of any two vibration sensing devices; Based on the induction correlation coefficients of any two vibration sensing devices, clustering the multiple vibration sensing devices to determine multiple vibration sensing device clusters; For any magnetic induction device cluster and any sound collection device cluster, based on the collected data of multiple magnetic induction devices and multiple sound collection devices in each test scene, calculate the induction correlation coefficient of the magnetic induction device cluster and the sound collection device cluster; For any magnetic induction device cluster and any vibration induction device cluster, based on the collected data of multiple magnetic induction devices and multiple vibration induction devices in each test scene, calculate the induction correlation coefficient of the magnetic induction device cluster and the vibration induction device cluster; For any sound collection device cluster and any vibration sensing device cluster, based on the collected data of multiple sound collection devices and multiple vibration sensing devices in each test scene, calculating the induction correlation coefficient of the sound collection device cluster and the vibration sensing device cluster; Determine a plurality of cluster units based on an induction correlation coefficient between any magnetic induction device cluster and any sound collection device cluster, an induction correlation coefficient between any magnetic induction device cluster and any vibration induction device cluster, and an induction correlation coefficient between any sound collection device cluster and any vibration induction device cluster, wherein each cluster unit includes a magnetic induction device cluster, a sound collection device cluster, and a vibration induction device cluster; For each cluster unit, based on the magnetic field information collected by the magnetic induction device included in the cluster unit, the sound information collected by the sound collection device, and the vibration information collected by the vibration induction device, determine the anchor risk value corresponding to the cluster unit; Based on the anchor risk value corresponding to each of the cluster units, the possibility of the second anchor existing is determined.

2. The machine vision-based chute dredging machine bolt sorting and intelligent iron removal system according to claim 1 is characterized in that: The magnetic induction unit includes a plurality of magnetic induction devices arranged at different positions; The sound collection unit includes a plurality of sound collection devices arranged at different positions; The vibration sensing unit includes a plurality of vibration sensing devices arranged at different positions; The anchor rod sensing module is also used to determine the setting positions of multiple magnetic sensing devices, multiple sound collection devices and multiple vibration sensing devices, specifically including: A plurality of magnetic induction devices are arranged at a plurality of first initial positions, a plurality of sound collection devices are arranged at a plurality of second initial positions, and a plurality of vibration induction devices are arranged at a plurality of third initial positions; Setting multiple test scenarios, obtaining data collected by multiple magnetic induction devices, multiple sound collection devices, and multiple vibration induction devices in each test scenario; Determine the effective magnetic induction parameters of each first initial position, the effective sound induction parameters of each second initial position, and the effective vibration induction parameters of each third initial position based on the collected data of the multiple magnetic induction devices, the multiple sound collection devices, and the multiple vibration induction devices in each test scene; Based on the effective magnetic induction parameters of each first initial position, the plurality of first initial positions are screened to determine the setting positions of the plurality of magnetic induction devices; Based on the effective sound sensing parameter of each second initial position, the plurality of second initial positions are screened to determine the setting positions of the plurality of sound collecting devices; Based on the effective vibration sensing parameters of each third initial position, the plurality of third initial positions are screened to determine the setting positions of the plurality of vibration sensing devices.

3. The machine vision-based chute dredging machine bolt sorting and intelligent iron removal system according to claim 1 or 2, characterized in that: The secondary determination unit determines the anchor information based on the ore-dropping machine image, magnetic field information, sound information and vibration information, including: Based on the ore drawer image, predicting the first anchor position and the first anchor size; Based on the magnetic field information, sound information and vibration information, predict the second anchor position and the second anchor size; Determining a predicted anchor position based on the first anchor position and the second anchor position; Based on the first anchor rod size and the second anchor rod size, a predicted anchor rod size is determined, wherein the anchor rod information includes the predicted anchor rod position and the predicted anchor rod size.

4. The machine vision-based chute dredging machine bolt sorting and intelligent iron removal system according to claim 1 is characterized in that: The parameter determination unit adjusts the working parameters of the image acquisition unit based on the anchor information, including: Based on the anchor information, determining a secondary projection posture and a secondary image acquisition posture; According to the secondary projection posture, adjusting the posture of the visible light irradiation device; According to the secondary image acquisition posture; adjusting the posture of the image acquisition unit.

5. The machine vision-based chute dredging machine bolt sorting and intelligent iron removal system according to claim 1 or 2, characterized in that: The parameter determination unit determines the anchor bolt sorting working parameters based on the target area image, including: Determining the anchor position and anchor size according to the target area image; Based on the anchor position and anchor size, the anchor sorting working parameters are determined, wherein the anchor sorting working parameters at least include a working path.

Images