An intelligent coal gangue sorting system

Abstract

The present application provides a coal gangue intelligent sorting system, including a transmission part, a data acquisition part, a blowing part and a host computer, wherein the transmission part includes a motor and a belt conveyor, and the belt conveyor is driven by the motor to transmit the coal gangue material; the data acquisition part is used to emit X-rays to the coal gangue material and collect transmission intensity data and transmission images; the blowing part is arranged at the end of the belt conveyor along the transmission direction, and includes a plurality of pneumatic valves, which are used to blow the coal gangue material to achieve the separation of coal and gangue; the host computer classifies the coal gangue material into coal gangue based on the transmission intensity data and the transmission image and determines the real-time blowing frequency of the blowing part. The coal gangue intelligent sorting system provided by the present application can significantly improve the proportion of coal sorted from the coal gangue material, and realize the efficient utilization of coal gangue.

Description

Technical Field

[0001] The present application belongs to the technical field of coal sorting, and specifically, relates to an intelligent coal gangue sorting system. Background Art

[0002] In the foreseeable future, coal will remain one of the most important non-renewable resources in my country and even in the world.

[0003] Coal preparation is the foundation of clean coal technology and the premise of deep processing of coal (water-coal slurry, coking, gasification, liquefaction) and clean and efficient utilization. Coal quality can be improved by coal preparation. Most of the mineral impurities in raw coal can be removed by washing and processing, which can reduce 65% of ash content and remove 65% of pyrite, reduce coal pollution to the atmosphere, and have significant environmental benefits. However, due to the limitations of resources, energy and technical conditions, the raw coal washed by power plants each year only accounts for 1 / 4 of the total raw coal consumed by power plants. Compared with developed countries, there is still a lot of room for improvement; in addition, traditional coal washing methods, such as manual method, heavy medium method, jigging method, etc., have large equipment and are prone to cause serious water pollution.

[0004] In addition to the above-mentioned problems in the raw coal washing process, achieving efficient sorting of coal gangue is also an urgent need to optimize the utilization rate of coal resources. Coal gangue is a solid waste with low carbon content and high ash content formed during coal mining and processing. It is one of the largest industrial solid wastes discharged in my country. So far, my country has accumulated a large amount of coal gangue. If it cannot be effectively utilized, it will not only waste the coal resources and other effective mineral components contained in it, making it impossible to improve the overall utilization rate of raw coal, but also seriously affect the utilization rate of land resources by large-scale stacking. Therefore, it is necessary to achieve efficient utilization of coal gangue resources through coal gangue sorting to achieve good economic and environmental benefits. .

[0005] However, since gangue itself is the product of raw coal after washing, the gangue content of gangue is increasing with the improvement of coal mining mechanization. Using traditional coal preparation methods such as manual method, heavy medium method, jigging method, etc. to sort it, not only is it difficult to classify and identify the gangue, but the sorting cost will also increase sharply with the increase of gangue content, and even exceed the benefits of sorting the gangue.

[0006] Therefore, it is necessary to provide a coal gangue sorting system that can accurately identify and sort coal gangue, is pollution-free and has a high benefit-cost ratio. Summary of the invention

[0007] In order to solve the problems existing in the above-mentioned prior art, the purpose of the present application is to provide a coal gangue intelligent sorting system, which can utilize the different transmission characteristics of coal and gangue under X-rays to achieve accurate, efficient and intelligent sorting of coal gangue.

[0008] The embodiments of the present application can be implemented through the following technical solutions:

[0009] A coal gangue intelligent sorting system comprises a conveying part, a data acquisition part, a blowing part and a host computer, wherein the conveying part comprises a motor and a belt conveyor, and the belt conveyor conveys coal gangue materials driven by the motor; the data acquisition part is used to emit X-rays to the coal gangue materials and collect transmission intensity data and transmission images; the blowing part is arranged at the end of the belt conveyor along the conveying direction, and comprises a plurality of pneumatic valves, which are used to blow the coal gangue materials to achieve separation of coal and gangue; the host computer classifies the coal gangue materials into coal gangue based on the transmission intensity data and the transmission image and determines the real-time blowing frequency of the blowing part.

[0010] Furthermore, the data acquisition unit includes an X-ray source and an X-ray detection device arranged opposite to each other on both sides of the conveying surface of the belt conveyor; the X-ray detection device is used to obtain the transmission intensity data and the transmission image, wherein the transmission intensity data includes high-energy transmission intensity data and low-energy transmission intensity data.

[0011] Furthermore, the host computer includes: a motor control unit, a data acquisition control unit, an image processing unit, a classification unit and a blowing control unit; the motor control unit is used to control the rotation speed of the motor; the data acquisition control unit is used to control the X-ray source to emit X-rays to the coal gangue material and receive the transmission intensity data and the transmission image; the image processing unit extracts the coal gangue target from the transmission image; the classification unit calculates the detection vector of the coal gangue target based on the transmission intensity data, and classifies the coal gangue target according to the detection vector; the blowing control unit controls the real-time blowing frequency of the blowing part based on the classification result of the coal gangue target.

[0012] Preferably, the transmission intensity data further includes equivalent transmission intensity data; and the transmission image is generated based on the equivalent transmission intensity data.

[0013] Furthermore, the image processing unit extracts the coal gangue target from the transmission image through the following steps:

[0014] S100. Denoising the transmission image by spatial smoothing;

[0015] S200. Generate a binary edge image based on an edge detection algorithm of multi-operator fusion;

[0016] S300. Determine the coal gangue target from the binary edge image based on a clustering algorithm.

[0017] Preferably, the spatial smoothing process is to filter the transmission image using a median filter template.

[0018] Furthermore, the multi-operator fusion edge detection algorithm uses the following fractional differential operator to perform edge detection on the transmission image:

[0019]

[0020] Among them, f(x,y) is the edge value at the image (x,y) coordinate, f0 is the RL fractional differential operator, w0 is the weight of f0, f l , l∈[1,…,L] are L different integer-order differential operators, w l f l The weight of

[0021] Further, step S300 includes the following steps:

[0022] S310. Importing the binary edge image;

[0023] S320. Randomly select a plurality of edge points from the binary edge image as starting centroids;

[0024] S330. When the cluster allocation of any edge point changes, execute steps S340 and S350 in a loop, otherwise exit the loop, wherein:

[0025] S340. Perform the following steps in sequence for each edge point in the edge image:

[0026] S341. For each centroid, respectively calculate the connectivity between centroids, the distance between the centroid and the cluster center, and the distance between the centroid and the edge point;

[0027] S342. based on the calculation results, delete the centroids connected to other centroids and the centroids surrounded by other clusters;

[0028] S343. Assign edge points to the clusters closest to them;

[0029] S350. For each cluster, find the cluster mean and update it as the centroid;

[0030] S360. Determine the coal gangue target based on the clustering results.

[0031] Preferably, after determining the coal gangue target, the image processing unit further optimizes the sampling points of the coal gangue target, and the sampling point optimization further includes the following steps:

[0032] S410. Divide each coal gangue target into multiple sampling rings with equal spacing from the cluster center to the cluster edge;

[0033] S420. Delete the outermost sampling ring of the coal gangue target;

[0034] S430. Selecting a sampling ring for gangue classification based on the area of ​​the remaining portion of the gangue target;

[0035] S440. Traverse each sampling ring for gangue classification;

[0036] S450. Determine the coal gangue target for coal gangue classification based on the traversal results.

[0037] Furthermore, the classification unit classifies the coal gangue targets through the following steps:

[0038] A100. Determine the energy detection vector of the coal gangue target based on the transmission intensity data Where m is the number of sampling points in the coal gangue target, is the low energy transmission intensity of the i-th sampling point, R i is the classification feature of the i-th sampling point determined by the following formula:

[0039]

[0040] in, are the initial values ​​of high energy transmission intensity and low energy transmission intensity, is the high energy transmission intensity of the i-th sampling point;

[0041] A200. Calculate the classification correlation coefficient r of the coal gangue target based on the energy detection vector j ,

[0042]

[0043] Among them, c is coal, g is gangue, for The mean value of I i,j To R i Substitute the coal / gangue fitting curve The estimated value obtained is For I i,j The mean of

[0044] A300. Based on the classification correlation coefficient, the coal gangue target is classified through a preset target recognition strategy.

[0045] Furthermore, the target identification strategy is one of the following strategies:

[0046] Simple classification strategy, if r c >r g , then the target of coal gangue is coal, otherwise the target of coal gangue is gangue;

[0047] Reduce the strategy of containing gangue, if r c > the first threshold, the gangue target is coal, otherwise the gangue target is gangue;

[0048] Reducing coal content strategy, if r g > the second threshold, the gangue target is gangue, otherwise the gangue target is coal.

[0049] Furthermore, the blowing control unit determines the real-time blowing frequency of the blowing part through the following steps:

[0050] B100. Establish a gangue injection queuing model (X, Y, Z, A, B, C), where X is the gangue arrival time distribution, Y is the injection time distribution, Z is the number of pneumatic valves, A is the maximum queue number allowed by the optimized queuing system, B is the total number of gangue, and C is the queuing rule;

[0051] B200. Determine the injection queue index based on the gangue injection queue model (L s ,L q ,W s ,W q ,P n ,λ,μ,ρ), where L s is the average number of gangue arriving, L q is the average queue length, W s is the average waiting time, W q is the average waiting time of the queue, P n is the probability of stably injecting n pieces of gangue, λ is the average number of gangue arriving per unit time, μ is the number of gangue injected by a single pneumatic valve each time, and ρ = λ / μ is the real-time blowing frequency of the blowing unit;

[0052] B300. Substitute the real-time classification results of the gangue materials arriving at the end of the conveyor belt into the above-mentioned injection queue index, and determine the optimal value of ρ based on the following optimization objectives:

[0053]

[0054] Among them, z is the overall benefit function of injection, C s is the cost coefficient of the pneumatic valve for injection, and G is the profit of injecting one piece of gangue.

[0055] The embodiment of the present application provides a coal gangue intelligent sorting system which has at least the following beneficial effects:

[0056] (1) The coal gangue intelligent sorting system provided by the present application can quickly realize the intelligent identification and sorting of the target based on the different X-ray transmission characteristics of coal and gangue, and effectively solve the problems of various coal sorting methods in the prior art, such as complex process flow, poor recognition rate, low sorting efficiency and easy water pollution;

[0057] (2) The coal gangue intelligent sorting system provided by the present application, whose image processing unit respectively uses a multi-operator fusion edge detection algorithm to perform edge detection of coal gangue targets, can effectively improve the target positioning and clustering accuracy, and at the same time optimize the sampling points according to the size range of the coal gangue targets, can further eliminate the sampling points that affect the classification accuracy, and improve the classification accuracy;

[0058] (3) The coal gangue intelligent sorting system provided by the present application adopts a preset classification and identification strategy to classify and identify coal and gangue. By setting different discrimination conditions, it can meet different coal gangue sorting needs, greatly increasing the application scope of the system;

[0059] (4) The coal gangue intelligent sorting system provided in the present application uses a queuing model to establish an injection queuing index and introduces the cost of abandoned gangue and the benefit of injected gangue to construct an injection cost function. The real-time injection frequency of the pneumatic valve is determined based on the real-time classification result of the coal gangue target. Through the above method, the injection scheme with the best comprehensive performance can be obtained based on comprehensive consideration of various factors affecting the injection effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of equipment assembly of a coal gangue intelligent sorting system according to an embodiment of the present application;

[0061] Figure 2 is a schematic diagram of a transmission image according to an embodiment of the present application;

[0062] Figure 3 A schematic diagram of the system framework and data transmission of a host computer according to an embodiment of the present application;

[0063] Figure 4 According to the median filtering template of the embodiment of the present application Figure 2 The result of denoising the transmission image;

[0064] Figure 5 A transmission image according to an embodiment of the present application and a plurality of coal gangue targets extracted therefrom;

[0065] Figure 6 It is a schematic diagram of dividing a coal gangue target into multiple sampling rings according to an embodiment of the present application;

[0066] Figure 7 This is the measured energy detection vector distribution result of coal and gangue. DETAILED DESCRIPTION

[0067] Hereinafter, the present application will be further described based on preferred embodiments with reference to the accompanying drawings.

[0068] In addition, various components in the drawings are enlarged or reduced in size for ease of understanding, but this practice is not intended to limit the scope of protection of the present application.

[0069] Words importing the singular also include the plural and vice versa.

[0070] In the description of the embodiments of the present application, it should be noted that if the terms "upper", "lower", "inside", "outside", etc. indicate an orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship in which the products of the embodiments of the present application are usually placed when in use, it is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present application. In addition, in the description of the present application, in order to distinguish different units, the words first, second, etc. are used in this specification, but these are not limited by the order of manufacture, nor can they be understood as indicating or implying relative importance, and their names may be different in the detailed description and claims of the present application.

[0071] The vocabulary in this specification is used to illustrate the embodiments of the present application, but is not intended to limit the present application. It should also be noted that, unless otherwise clearly specified and limited, the terms "disposed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, a direct connection, an indirect connection through an intermediate medium, or a connection between the two elements. For those skilled in the art, the specific meanings of the above terms in this application can be specifically understood.

[0072] Figure 1 The following is a schematic diagram of the assembly of a coal gangue intelligent sorting system according to some embodiments of the present application. Figure 1 As shown, the coal gangue intelligent sorting system provided in the present application includes a conveying part, a data acquisition part, a blowing part and a host computer, wherein the conveying part includes a motor and a belt conveyor, and the belt conveyor conveys the coal gangue material driven by the motor; the data acquisition part is used to emit X-rays to the coal gangue material and collect transmission intensity data and transmission images; the blowing part is arranged at the end of the belt conveyor along the conveying direction, and includes a plurality of pneumatic valves, which are used to blow the coal gangue material to achieve separation of coal and gangue; the host computer classifies the coal gangue material into coal gangue based on the transmission intensity data and the transmission image and determines the real-time blowing frequency of the blowing part.

[0073] In some embodiments, the data acquisition unit further includes an X-ray source and an X-ray detection device arranged opposite to each other on both sides of the conveying surface of the belt conveyor, wherein the X-ray source is connected to the host computer via a data line (such as an RSC232 data line), and emits X-rays to the conveying surface of the belt conveyor under the control of the host computer; the X-ray detection device further includes an X-ray detector and a data acquisition card connected thereto, the X-ray detector converts the penetrating X-rays into electrical signals through the photoelectric effect, generates transmission intensity data after being collected by the data acquisition card, and sends the transmission intensity data to the host computer via a data line (such as a network cable based on the TCP communication protocol).

[0074] In some embodiments, the host computer adjusts the speed of the motor and the switch of the pneumatic valve respectively through a programmable logic controller (PLC) to achieve control of the conveying speed of the conveying part and the real-time blowing frequency of the blowing part.

[0075] Table 1 below gives the equipment parameters of a coal gangue intelligent sorting system in a specific embodiment:

[0076] Table 1 Equipment parameters of coal gangue intelligent sorting system

[0077]

[0078] The X-ray detector includes 20 detection cards, each of which can collect transmission intensity data of 64 sampling points, and the width of each sampling point is 1.5 mm. Each sampling of the 20 detection cards can form 1280×1 transmission intensity data, wherein, preferably, the transmission intensity data includes high-energy transmission intensity data and low-energy transmission intensity data. In the above embodiment, the transmission intensity data can be mapped from an analog signal to an 8-bit digital signal (0 to 255) according to its intensity range.

[0079] In addition, by continuously sampling at a certain time interval and then splicing the multiple sampling results, a transmission image can be formed. For example, in some embodiments, after every 320 samplings, a transmission image of 1280×320 pixels is formed, wherein the gray value of each pixel reflects the transmission intensity of the corresponding sampling point. The above-mentioned technology of performing analog-to-digital conversion based on signal intensity data and generating grayscale images is known to those skilled in the art.

[0080] In the actual X-ray transmission and detection process, due to the complex composition of the material, its mass, density, and thickness distribution range is large, and the projection image constructed by simply using high-energy transmission intensity data or low-energy transmission intensity data may not accurately reflect the attenuation characteristics of the X-ray transmission process. For example, when the thickness of some coal gangue materials is large, the high-energy part of the X-ray may not be able to penetrate both the coal and the gangue; when the thickness of some coal gangue materials is small, after the low-energy part of the X-ray penetrates the coal and the gangue, there may be little difference in intensity attenuation, resulting in a small degree of differentiation of the intensity data.

[0081] To this end, in some preferred embodiments of the present application, the transmission intensity data also includes equivalent transmission intensity data; the transmission image is generated based on the equivalent transmission intensity data. Specifically, the X-ray detection device is based on the high energy transmission intensity I at each sampling point. H and low energy transmission intensity I L Processing is performed to obtain the equivalent transmission intensity I corresponding to the sampling point e , through the equivalent transmission intensity I e , a unified transmission attenuation model for different gangue materials can be established:

[0082]

[0083] Among them, I e0 is the initial value of the equivalent transmission intensity, T is the thickness of the gangue material, u e The equivalent transmission model can effectively characterize the comprehensive attenuation characteristics of the high-energy and low-energy transmission of X-rays by gangue materials of different masses, densities, and thicknesses, thereby greatly increasing the discrimination of the generated transmission image for different materials, which is beneficial to edge detection and clustering processing in the subsequent target extraction process. Figure 2 A schematic diagram showing transmission images of coal gangue materials at multiple locations generated according to a preferred embodiment of the present application is shown.

[0084] Figure 3 It further shows Figure 1 The system framework of the middle and upper computer and the schematic diagram of data transmission between it and the transmission part, data acquisition part and blowing part, as shown in Figure 3 As shown, in an embodiment of the present application, the host computer includes a motor control unit, a data acquisition control unit, an image processing unit, a classification unit and a blowing control unit.

[0085] Specifically, the motor control unit is used to control the rotation speed of the motor; the data acquisition control unit is used to control the X-ray source to emit X-rays to the gangue material and receive transmission intensity data and transmission images; the image processing unit extracts the gangue target from the transmission image; the classification unit calculates the detection vector of the gangue target based on the transmission intensity data, and classifies the gangue target according to the detection vector; the blowing control unit controls the real-time blowing frequency of the blowing unit based on the classification result of the gangue target. The connection between the above-mentioned units and the transmission unit, the data acquisition unit and the blowing unit has been introduced in the previous text. The specific implementation methods of the image processing unit, the classification unit and the blowing control unit in the host computer are described in detail below through the accompanying drawings and preferred embodiments.

[0086] In some embodiments of the present application, the image processing unit extracts the coal gangue target from the transmission image through the following steps:

[0087] S100. Denoising the transmission image by spatial smoothing;

[0088] S200. Generate a binary edge image based on an edge detection algorithm of multi-operator fusion;

[0089] S300. Determine the coal gangue target from the binary edge image based on a clustering algorithm.

[0090] Specifically, the image processing unit first performs noise reduction processing on the transmission image in step S100, such as Figure 2 As shown in the figure, the original transmission image obtained by the X-ray detection device contains a lot of noise. There are many reasons for the noise, such as the dust, moisture and debris in the actual working conditions covering the ray tube, Gaussian noise generated by the cross-current of the electronic circuit, sharp noise generated by the analog-to-electric conversion of the detection card, etc. The existence of the above noise causes serious interference to the subsequent target extraction and classification, so it is necessary to first denoise the transmission image.

[0091] Through the analysis of the noise existing in the transmission image, it is found that the above noise is frequency-independent noise, so it can be processed by spatial smoothing. The essence of filtering based on spatial smoothing is blurring, which blurs the image details while reducing noise. Excessive smoothing has a certain impact on subsequent target positioning. Through experimental comparison of commonly used spatial smoothing filtering algorithms, it is found that median filtering can better overcome the image detail blurring caused by linear filters (mean filters), and has an obvious effect on filtering out pulse interference and image scanning noise. Therefore, in the preferred embodiment of the present application, median filtering is selected to reduce the noise of the transmission image. Figure 4 In some preferred embodiments, a 5×5 median filter template is used. Figure 2 The result of denoising the transmission image.

[0092] After the transmission image is denoised, the image processing unit performs edge detection on the transmission image in step S200; and then extracts the coal gangue target through a clustering algorithm in step S300.

[0093] Target recognition and extraction is to divide the image into material-centered sub-graphs according to the distribution of materials from the perspective of changes in image attributes, so as to support subsequent target recognition. Significant changes in image attributes are manifested as discontinuities in image attributes, such as discontinuities in grayscale, texture orientation, and material attributes, which reflect changes in the distribution of materials in the image. Since the image processing unit in the embodiment of the present application needs to quickly and accurately identify and extract gangue targets from continuously transmitted gangue materials, its real-time requirements have abandoned most of the image target positioning algorithms. In addition, in order to ensure the accuracy of gangue target recognition, the image filtering and denoising operation is weak, and there is still a lot of noise remaining, which also has a certain impact on the subsequent edge processing operations. To this end, the present application eliminates most of the irrelevant attribute information through a multi-operator fusion edge detection algorithm, and retains the structural attribute information of important materials while ensuring the real-time performance of the algorithm, providing a guarantee for subsequent clustering and target extraction.

[0094] Specifically, in a preferred embodiment of the present application, the edge detection algorithm of multi-operator fusion uses the following fractional differential operator to perform edge detection on the transmission image:

[0095]

[0096] Among them, f(x,y) is the edge value at the image (x,y) coordinate, f0 is the RL fractional differential operator, w0 is the weight of f0, f l , l∈[1,…,L] are L different integer-order differential operators, w l f l The weight of

[0097] Specifically, the RL fractional-order differential operator f0 can be discretized from the Riemann-Liouville fractional-order differential:

[0098]

[0099]

[0100] in, is an RL fractional discrete operator, f(t) is a continuously differentiable function near t; α and t are the upper and lower limits of the integration of the differential operator, n is the smallest integer greater than α, and is the differential order of the RL fractional differential operator, α>0 and n-1≤α≤n.

[0101] Specifically, the integer-order differential operator f l It can be one or more of the Prewitt operator, the Laplacian operator, and the Canny operator.

[0102] The use of common first-order differential operators, such as Robert operator, Prewitt operator, Sobel operator, Canny operator, etc. for image edge detection is well known to those skilled in the art. However, there are certain defects in using the above operators alone for edge detection of coal gangue transmission images. For example, the Robert operator is more accurate in positioning, but it is more sensitive to noise because it does not include smoothing. Both the Prewitt operator and the Sobel operator are differential operators with mean filtering properties, while the former is an average filter and the latter is a weighted average filter, and the detected image edge may be greater than 2 pixels. Both have a good detection effect on grayscale gradient and low-noise images, but for X-ray images with mixed complex noises, the processing effect is not ideal. The Canny operator has a high resistance to noise at the edge, but it is easy to misjudge isolated and clumping noise points as boundaries. Therefore, the use of the above first-order differential operators alone cannot well extract the edge of coal gangue targets.

[0103] Based on the above reasons, this application selects the Robert operator with accurate positioning, as well as the Prewitt operator and Sobel operator with high noise resistance from the integer-order differential operators, and further integrates them with the RL fractional-order differential operator to achieve accurate material edge detection under complex noise.

[0104] Furthermore, after acquiring the binary edge image, the image processing unit further extracts the coal gangue target through the following steps:

[0105] S310. Importing the binary edge image;

[0106] S320. Randomly select a plurality of edge points from the binary edge image as starting centroids;

[0107] S330. When the cluster allocation of any edge point changes, execute steps S340 and S350 in a loop, otherwise exit the loop, wherein:

[0108] S340. Perform the following steps in sequence for each edge point in the edge image:

[0109] S341. For each centroid, respectively calculate the connectivity between centroids, the distance between the centroid and the cluster center, and the distance between the centroid and the edge point;

[0110] S342. based on the calculation results, delete the centroids connected to other centroids and the centroids surrounded by other clusters;

[0111] S343. Assign edge points to the clusters closest to them;

[0112] S350. For each cluster, find the cluster mean and update it as the centroid;

[0113] S360. Determine the coal gangue target based on the clustering results.

[0114] Through the above steps S310 to S360, different gangue targets can be extracted from the gangue material. Figure 5 A transmission image and a plurality of extracted gangue targets in a specific embodiment are shown, wherein each gangue target is surrounded by a corresponding area through an edge line, and the sampling points in the area can be used for further gangue classification.

[0115] However, due to the large size distribution span of different gangue targets in the gangue material, especially when X-rays are irradiated with larger materials, the transmission intensity data may be distorted. By statistically analyzing the shapes of gangue materials of various sizes, it is found that the material is wedge-shaped from the center to the edge, that is, the center is thick and the edge is thin, and the center thickness increases with the increase of size. The above characteristics make the transmission intensity data of materials of different sizes and thicknesses present different usability. For example, for some large-sized materials with high center thickness, the area with excessive thickness may not be penetrated, so that the transmission intensity data of this part of the area cannot truly reflect the attenuation characteristics of X-rays; the X-ray penetration rate of its edge area is high, but correspondingly, its transmission intensity is also highly affected by noise; in addition, the larger material contains too many sampling points in the image, which makes the classification calculation time too long, seriously affecting the real-time performance of the sorting system, and further affecting the subsequent blowing. Therefore, it is necessary to further optimize the sampling points of the gangue target extracted in step S300.

[0116] Specifically, in some preferred embodiments of the present application, the optimization of the target sampling point of coal gangue is performed by the following steps:

[0117] S410. Divide each coal gangue target into multiple sampling rings with equal spacing from the cluster center to the cluster edge;

[0118] S420. Delete the outermost sampling ring of the coal gangue target;

[0119] S430. Selecting a sampling ring for gangue classification based on the area of ​​the remaining portion of the gangue target;

[0120] S440. Traverse each sampling ring for gangue classification;

[0121] S450. Determine the coal gangue target for coal gangue classification based on the traversal results.

[0122] In the above steps, the number of sampling rings can be set according to the size distribution range of the gangue target and the resolution of the gangue target image (for example, for the case where most material sizes are distributed in [10 mm, 300 mm], the number of divided sampling rings can be set to 10 to 20). After the gangue target is divided into multiple sampling rings, the edge part is first removed, and then the sampling ring for subsequent gangue classification is selected according to the area of ​​the remaining part, and then the sampling points that are not penetrated by the X-ray (i.e., the sampling points with a transmission intensity of 0) are removed, and finally an optimized gangue target for subsequent gangue classification is obtained.

[0123] Figure 6 A schematic diagram of dividing sampling rings for a specific coal gangue target is shown. The coal gangue target is divided into 10 sampling rings with equal spacing from the cluster center to the cluster edge.

[0124] Table 2 below shows a sampling ring selection strategy under a partitioning scheme with a variable number of sampling rings.

[0125] Table 2 Sampling ring selection strategy (division scheme with variable number of sampling rings)

[0126]

[0127] After the image processing unit extracts the coal gangue targets through the above steps, the classification unit can classify each coal gangue target into coal and gangue. In the embodiment of the present application, the classification of coal gangue targets is based on the different attenuation characteristics of coal and gangue for X-ray transmission.

[0128] Statistics show that the true density of coal is about 1.3-1.8, and the true density of gangue is about 1.8-2.3. The density of coal and gangue is very different, which makes coal and gangue present different characteristics of X-ray penetration attenuation, thus determining that X-rays can be used to identify and sort coal and gangue. In the embodiment of the present application, the classification unit classifies coal and gangue targets through the following steps:

[0129] A100. Determine the energy detection vector of the coal gangue target based on the transmission intensity data Where m is the number of sampling points in the coal gangue target, is the low energy transmission intensity of the i-th sampling point, R i is the classification feature of the i-th sampling point determined by the following formula:

[0130]

[0131] in, are the initial values ​​of high energy transmission intensity and low energy transmission intensity, is the high energy transmission intensity of the i-th sampling point;

[0132] A200. Calculate the classification correlation coefficient r of the coal gangue target based on the energy detection vector j ,

[0133]

[0134] Among them, c is coal, g is gangue, for The mean value of I i,j To R i Substitute the coal / gangue fitting curve The estimated value obtained is For I i,j The mean of

[0135] A300. Based on the classification correlation coefficient, the coal gangue target is classified through a preset target recognition strategy.

[0136] Specifically, step A100 is used to construct the energy detection vector of each sampling point of each coal gangue target, where: High-energy X-rays and low-energy X-rays can be detected and acquired in the absence of coal gangue materials.

[0137] For coal and gangue with different transmission attenuation characteristics, their energy detection vectors show different R~I L Corresponding relationship, Figure 7 The energy detection vector distribution results of coal and gangue obtained by multiple field tests are shown. Figure 7 It can be seen that the energy detection vector of coal and gangue is between R and I L Obvious clustering features are shown in the feature space, and the separation degree of the clustered areas is good. Therefore, the energy detection vector constructed in the present application can be effectively applied to the classification and identification of coal and gangue.

[0138] In order to facilitate the classification of coal gangue, y = a·e bx The function relationship is used to fit the measured multiple energy detection vectors of coal and gangue to obtain the fitting curve of coal Fitting curve with gangue Then, the classification correlation coefficient r of each gangue target corresponding to coal and gangue is calculated through step A200. c With r g .

[0139] After obtaining the r of each gangue target c With r g After that, the coal gangue target can be classified by using the preset target recognition strategy in step A300. In some preferred embodiments of the present application, the preset target recognition strategy includes:

[0140] Simple classification strategy, if r c >r g , then the target of coal gangue is coal, otherwise the target of coal gangue is gangue;

[0141] Reduce the strategy of containing gangue, if r c > the first threshold, the gangue target is coal, otherwise the gangue target is gangue;

[0142] Reducing coal content strategy, if r g > the second threshold, the gangue target is gangue, otherwise the gangue target is coal.

[0143] The above different target recognition strategies correspond to different sorting requirements. Among them, the simple classification strategy simply c With r g The coal gangue targets are classified by comparing their sizes, which can be used in general coal gangue sorting occasions; the strategy of reducing the gangue content can improve the coal content of the coal gangue targets identified as coal by identifying the coal gangue targets less than or equal to the first threshold as gangue, which is suitable for occasions where high-quality coal needs to be extracted; the strategy of reducing the coal content can improve the gangue content of the coal gangue targets identified as gangue by identifying the targets greater than the second threshold as gangue, which is suitable for occasions where effective mineral components in materials with high gangue content are identified and sorted. In the specific implementation process, the first threshold and the second threshold can be determined based on the sampling and detection data of different batches of coal gangue materials, combined with actual sorting needs.

[0144] After the classification unit classifies the various coal gangue targets continuously transported on the conveyor belt, the injection control unit can adjust the real-time injection frequency of multiple pneumatic valves set at the end of the conveyor belt according to the classification results and the different coal gangue targets arriving at the injection part along the conveying direction of the conveyor belt.

[0145] In some optional embodiments, the real-time blowing frequency of multiple pneumatic valves can be determined according to the ratio of coal to gangue in the classification results. For example, when the coal ratio (quantity ratio or area ratio) is less than a certain value of the gangue material reaching the end of the conveyor belt, the pneumatic valve is started for blowing so that the material at this time is blown away. Otherwise, the blowing is stopped so that the material falls directly, thereby realizing coal gangue sorting.

[0146] In the actual blowing control process, since a single pneumatic valve cannot achieve continuous blowing of the continuously arriving gangue, it is generally necessary to consist of multiple pneumatic valves to form a blowing unit, and the blowing control unit controls them to blow in turns. However, there is a delay in the mechanical action of the pneumatic valve, which makes the control of the pneumatic valve a key point that restricts the blowing performance of the system, that is, if the real-time blowing frequency of multiple pneumatic valves is set unreasonably, it will lead to the inability to blow the gangue in time or the pneumatic valve will be idle for a long time. In order to avoid blowing small and mistakenly blowing big and blowing big with small, it is necessary to optimize the blowing frequency of the pneumatic valve according to the gangue classification results of the materials arriving at the end of the conveyor belt.

[0147] The essence of injection optimization is an optimization scheduling problem for the targeted queue injection of gangue, that is, to find the least pneumatic valves that can meet the injection demand in the queue, so that the cost and index loss of injection are minimized under the conditions of injection, and the efficiency of the injection system is the highest. The randomness of the order and time of gangue arrival can be described by probability, and the optimization solution can be found through queuing theory.

[0148] Specifically, in some preferred embodiments of the present application, the blowing control unit determines the real-time blowing frequency of the blowing part through the following steps:

[0149] B100. Establish a gangue injection queuing model (X, Y, Z, A, B, C), where X is the gangue arrival time distribution, Y is the injection time distribution, Z is the number of pneumatic valves, A is the maximum queue number allowed by the optimized queuing system, B is the total number of gangue, and C is the queuing rule;

[0150] B200. Determine the injection queue index based on the gangue injection queue model (L s ,L q ,W s ,W q ,P n ,λ,μ,ρ), where L s is the average number of gangue arriving, L q is the average queue length, W s is the average waiting time, W q is the average waiting time of the queue, P n is the probability of stably injecting n pieces of gangue, λ is the average number of gangue arriving per unit time, μ is the number of gangue injected by a single pneumatic valve each time, and ρ = λ / μ is the real-time blowing frequency of the blowing unit;

[0151] B300. Substitute the real-time classification results of the gangue materials arriving at the end of the conveyor belt into the above-mentioned injection queue index, and determine the optimal value of ρ based on the following optimization objectives:

[0152]

[0153] Among them, z is the overall benefit function of injection, C s is the cost coefficient of the pneumatic valve for injection, and G is the profit of injecting one piece of gangue.

[0154] Specifically, the gangue arriving with the conveyor belt can be regarded as infinite, there is no correlation between the gangues, their arrival times are independent of each other and the arrival rules are random; further, within the time Δt, the probability of a piece of gangue arriving is λΔt; the probability of no gangue arriving is 1-λΔt; Δt is regarded as infinitesimal, and the number of gangue arriving at the same time is less than or equal to 1; the arrival of gangue obeys λ.

[0155] Based on the above settings, a queuing model (X, Y, Z, A, B, C) is established, where X is the time interval distribution of gangue arrival, Y is the blowing time distribution, Z is the number of pneumatic valves, A is the maximum queue number allowed by the optimized queuing system, B is the total number of gangue, and C is the queuing rule. In actual sorting, different queuing rules can be set according to different sorting goals, such as: first come first blow, large gangue priority, gangue not adjacent to coal priority, etc.

[0156] Furthermore, the injection queue index (L s ,L q ,W s ,W q ,P n ,λ,μ,ρ), where L s is the average number of gangue arriving, L q is the average queue length, W s is the average waiting time, W q is the average waiting time of the queue, P n is the probability of stably blowing n pieces of gangue, λ is the average number of gangue arriving per unit time, μ is the number of gangues blown by a single pneumatic valve each time, and ρ=λ / μ is the real-time blowing frequency of the blowing part.

[0157] The following relationship exists between the above-mentioned injection queue indicators: L s =λW s ; L q =λW q ; L s =L q +λ / μ;W s =W q +1 / μ. When the gangue falls from the conveyor belt, the corresponding pneumatic valve is determined. Multiple gangues correspond to the same pneumatic valve, and the first-come-first-served blows. The valve action time is relatively independent, and the first-come-first-served blows and the valve action time are relatively independent. It obeys the negative exponential distribution of μ.

[0158] Furthermore, in the actual sorting process, the real-time classification results of the gangue materials arriving at the end of the conveyor belt are substituted into the above-mentioned injection queue index to calculate the probability of stable injection of the system: P n =(1-ρ)ρ n , Optimize queuing indicators: L s =λ / (μ-λ), L q =ρλ / (μ-λ), W s =1 / (μ-λ),W q =ρ / (μ-λ) and optimization parameters: effective arrival rate λ e =λ(1-P n ), effective blowing intensity Average number of gangue arriving The average queue length L q =L s -(1-P0), average waiting time for spraying Average waiting time in queue

[0159] Finally, we introduce parameters Cs and G, where C s is the cost coefficient of the pneumatic valve for injection, and G is the income of injecting a piece of gangue. Through the above queuing indicators and parameters, we can find the maximum total injection income z, that is, ρ that satisfies the following formula:

[0160]

[0161] In some specific embodiments, the total injection benefit function z can be expressed in an iterative form as follows:

[0162]

[0163] Among them, z (k) 、z (k+1) The total benefits of injection are calculated for the kth and k+1th iterations respectively. By continuously changing ρ and iteratively calculating, the total benefits of injection are iteratively gradually approaching the maximum value. When the difference between the calculation results of two adjacent iterations is less than the preset threshold, the iterative calculation can be stopped, and ρ at this time is taken as the optimal value.

[0164] The above-mentioned spraying scheme with the largest overall spraying benefit is the spraying optimization queue result. The time complexity of the queue optimization algorithm is O(nlogn). Due to the limitations of belt width and belt speed, the queue optimization time is less than 100ms, which meets the real-time spraying requirements. When the belt width and speed become larger, the queue optimization queue indicators can be simplified to further reduce the queue optimization time.

[0165] The above is a detailed introduction to the specific implementation methods of the present application. For those skilled in the art, several improvements and modifications may be made to the present application without departing from the principles of the present application. These improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A coal gangue intelligent sorting system, comprising a transmission unit, a data acquisition unit, a blowing unit and a host computer, characterized in that: The conveying part includes a motor and a belt conveyor, and the belt conveyor conveys the coal gangue material under the drive of the motor; The data acquisition unit is used to emit X-rays to the gangue material and collect transmission intensity data and transmission images, including an X-ray source and an X-ray detection device arranged opposite to each other on both sides of the conveying surface of the belt conveyor; the X-ray detection device is used to obtain the transmission intensity data and the transmission image, wherein the transmission intensity data includes high-energy transmission intensity data and low-energy transmission intensity data; The blowing part is arranged at the end of the belt conveyor along the conveying direction, and includes a plurality of pneumatic valves, which are used to blow the coal gangue material to achieve separation of coal and gangue; The host computer classifies the gangue material into gangue based on the transmission intensity data and the transmission image and determines the real-time blowing frequency of the blowing part, and includes a data acquisition control unit, an image processing unit, and a classification unit. The data acquisition control unit is used to control the X-ray source to emit X-rays to the gangue material and receive the transmission intensity data and the transmission image collected by the X-ray detection device; the image processing unit extracts the gangue target from the transmission image; the classification unit calculates the detection vector of the gangue target based on the transmission intensity data, and classifies the gangue target according to the detection vector; The image processing unit extracts the coal gangue target from the transmission image through the following steps: S100. Denoising the transmission image by spatial smoothing; S200. Generate a binary edge image based on an edge detection algorithm of multi-operator fusion; S300. Determine the gangue target from the binary edge image based on a clustering algorithm; After determining the gangue target, the image processing unit further optimizes the sampling points of the gangue target, and the sampling point optimization further includes the following steps: S410. Divide each coal gangue target into multiple sampling rings with equal spacing from the cluster center to the cluster edge; S420. Delete the outermost sampling ring of the coal gangue target; S430. Selecting a sampling ring for gangue classification based on the area of ​​the remaining portion of the gangue target; S440. Traverse each sampling ring for gangue classification; S450. Determine the coal gangue target for coal gangue classification based on the traversal results.

2. The intelligent coal gangue sorting system according to claim 1 is characterized in that: The host computer also includes: Motor control unit and spray control unit; The motor control unit is used to control the rotation speed of the motor; The injection control unit controls the real-time injection frequency of the injection part based on the classification result of the coal gangue target.

3. The intelligent coal gangue sorting system according to claim 1 is characterized in that: The transmission intensity data also includes equivalent transmission intensity data; The transmission image is generated based on the equivalent transmission intensity data.

4. The coal gangue intelligent sorting system according to claim 1 is characterized in that: The spatial smoothing process is to filter the transmission image using a median filter template.

5. The coal gangue intelligent sorting system according to claim 1 is characterized in that: The multi-operator fusion edge detection algorithm uses the following fractional differential operator to perform edge detection on the transmission image: Among them, f(x,y) is the edge value at the image (x,y) coordinate, f0 is the RL fractional differential operator, w0 is the weight of f0, f l , l∈[1,…,L] are L different integer-order differential operators, w l f l The weight of 6. The intelligent coal gangue sorting system according to claim 1 is characterized in that: Step S300 further includes the following steps: S310. Importing the binary edge image; S320. Randomly select a plurality of edge points from the binary edge image as starting centroids; S330. When the cluster allocation of any edge point changes, execute steps S340 and S350 in a loop, otherwise exit the loop, wherein: S340. Perform the following steps in sequence for each edge point in the edge image: S341. For each centroid, respectively calculate the connectivity between centroids, the distance between the centroid and the cluster center, and the distance between the centroid and the edge point; S342. based on the calculation results, delete the centroids connected to other centroids and the centroids surrounded by other clusters; S343. Assign edge points to the clusters closest to them; S350. For each cluster, find the cluster mean and update it as the centroid; S360. Determine the coal gangue target based on the clustering results.

7. The intelligent coal gangue sorting system according to claim 1 is characterized in that: The classification unit classifies the coal gangue targets through the following steps: A100. Determine the energy detection vector of the coal gangue target based on the transmission intensity data Where m is the number of sampling points in the coal gangue target, is the low energy transmission intensity of the i-th sampling point, R i is the classification feature of the i-th sampling point determined by the following formula: in, are the initial values ​​of high energy transmission intensity and low energy transmission intensity, is the high energy transmission intensity of the i-th sampling point; A200. Calculate the classification correlation coefficient r of the coal gangue target based on the energy detection vector j , Among them, c is coal, g is gangue, for The mean value of I i,j To R i Substitute the coal / gangue fitting curve The estimated value obtained, I i,j For I i,j The mean of A300: Based on the classification correlation coefficient, the coal gangue target is classified by a preset target recognition strategy.

8. The intelligent coal gangue sorting system according to claim 7 is characterized in that: The target identification strategy is one of the following strategies: Simple classification strategy, if r c >r g , then the target of coal gangue is coal, otherwise the target of coal gangue is gangue; Reduce the strategy of containing gangue, if r c > the first threshold, the gangue target is coal, otherwise the gangue target is gangue; Reducing coal content strategy, if r g > the second threshold, the gangue target is gangue, otherwise the gangue target is coal.

9. The intelligent coal gangue sorting system according to claim 2, characterized in that: The spray control unit determines the real-time spray frequency of the spray part by the following steps: B100. Establish a gangue injection queuing model (X, Y, Z, A, B, C), where X is the gangue arrival time distribution, Y is the injection time distribution, Z is the number of pneumatic valves, A is the maximum queue number allowed by the optimized queuing system, B is the total number of gangue, and C is the queuing rule; B200. Determine the injection queue index based on the gangue injection queue model (L s ,L q ,W s ,W q ,P n ,λ,μ,ρ), where L s is the average number of gangue arriving, L q is the average queue length, W s is the average waiting time, W q is the average waiting time of the queue, P n is the probability of stably injecting n pieces of gangue, λ is the average number of gangue arriving per unit time, μ is the number of gangue injected by a single pneumatic valve each time, and ρ = λ / μ is the real-time blowing frequency of the blowing unit; B300. Substitute the real-time classification results of the gangue materials arriving at the end of the conveyor belt into the above-mentioned injection queue index, and determine the optimal value of ρ based on the following optimization objectives: Among them, z is the overall benefit function of injection, C s is the cost coefficient of the pneumatic valve for blowing, and G is the profit of blowing one piece of gangue.

Images