A coal quality detection method and system based on image data processing
Through real-time monitoring of dust and coal surface characteristics and dynamically adjusting the detection rate, the missed and mis-checking problems of coal quality inspection in complex environments are solved, the detection accuracy and efficiency are improved, and environmental pollution and resource waste are reduced.
Patent Information
- Application Number
- CN202411249085.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-06
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-09-06
AI Technical Summary
Existing coal quality detection technology based on image data processing may lead to missed inspection or missed inspection in complex environments, affecting detection accuracy and efficiency, and poor coal entering the combustion process will lead to environmental pollution and waste of resources.
By monitoring the dynamic changes of dust in real time, obtaining dust particle diameter information and coal surface characteristics, generating dust particle diameter exceeding index and coal surface color difference index, inputting pre-trained machine learning model, evaluating image acquisition environment, and dynamically adjusting detection rate to ensure detection accuracy and efficiency.
Effectively reduce missed inspections and mis-checks, improve the accuracy and efficiency of the detection system in complex environments, reduce inferior coal entering the combustion process, reduce environmental pollution, and improve resource utilization efficiency.
Smart Images

Figure CN118758936B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of coal quality detection, and in particular to a coal quality detection method and system based on image data processing. Background Art
[0002] Coal quality detection based on image data processing is a method that uses image analysis technology to evaluate and monitor coal quality. Traditional coal quality detection usually relies on physical and chemical analysis, such as coal calorific value, sulfur content, ash content and other indicators. Although these methods are accurate, they are time-consuming and require professional equipment and operators. Image data processing technology can quickly and non-contactly evaluate the quality of coal by analyzing the appearance characteristics of coal (such as color, texture, particle size, etc.). The image processing system uses a camera device to obtain coal images, and then processes and analyzes the images through computer vision and machine learning algorithms to extract characteristic data related to coal quality.
[0003] Specifically, this method includes several key steps. First, use a high-definition camera or industrial camera to capture images of coal samples to ensure that the images have sufficient resolution and clarity. Then, image processing software preprocesses the images, such as denoising, contrast enhancement, and segmentation of coal particles. Next, feature extraction algorithms (such as edge detection, texture analysis, etc.) are used to extract feature information from coal images. Finally, these features are classified and evaluated through a trained machine learning model to obtain coal quality indicators, such as coal ash, sulfur content, and calorific value. Coal quality detection based on image data processing can not only significantly improve detection efficiency and accuracy, but also realize online real-time monitoring, which helps to optimize quality control during coal production and use.
[0004] When testing coal quality based on image data processing, the existing technology usually places the coal on a conveyor belt and uses mobile detection to test the coal quality. This method can monitor a large number of coal samples in real time and is suitable for large-scale, continuous production quality control. The camera is fixed above the conveyor belt, and as the coal moves, the camera equipment will continuously collect image data. The advantages of this method are high detection efficiency, the ability to cover a wider range of samples, and the ability to achieve online real-time monitoring, but it requires higher image processing and analysis technology.
[0005] The prior art has the following deficiencies:
[0006] When testing coal quality based on image data processing, the existing technology usually sets an optimal detection rate (the detection rate is the rate at which the conveyor belt conveys coal) based on historical experience and previous detection data to ensure accurate detection of coal while improving detection efficiency. However, when the inspection process is greatly affected, if this optimal transmission rate is still maintained, the detection system may not be able to capture certain subtle quality problems in time, which may lead to inaccurate coal quality detection, resulting in missed detection or wrong detection. Missed detection may cause coal with high ash, high moisture or high sulfur content to enter the combustion process. These low-quality coals will reduce combustion efficiency, increase fuel consumption, and produce more smoke and harmful gases, polluting the environment. Wrong detection may cause qualified coal to be mistakenly regarded as unqualified and discarded or downgraded, resulting in a huge waste of resources and affecting the economic benefits and resource utilization efficiency of the enterprise.
[0007] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention
[0008] The purpose of the present invention is to provide a coal quality detection method and system based on image data processing, which obtains dust particle diameter information and surface characteristics of coal samples by real-time monitoring of the dynamic changes of dust in the data collection area, generates a dust particle diameter excess index and a coal surface color difference index, inputs the dust particle diameter excess index and the coal surface color difference index into a pre-trained machine learning model, evaluates the impact of image acquisition on the environment, and divides the environment into high-impact and low-impact environments according to the evaluation results. The optimal detection rate is maintained in the low-impact environment, and the detection rate is reduced in the high-impact environment to ensure detection accuracy and efficiency, reduce the entry of inferior coal into the combustion process, improve economic benefits and resource utilization, so as to solve the problems in the above-mentioned background technology.
[0009] In order to achieve the above object, the present invention provides the following technical solution: a coal quality detection method based on image data processing, comprising the following steps:
[0010] According to historical experience and previous detection data, an optimal detection rate is set, that is, the rate at which the conveyor belt conveys coal. At the optimal detection rate, an image acquisition device is used to collect images of coal samples in real time for coal detection;
[0011] The conical area between the image acquisition device and the acquisition area is defined as the data acquisition influence area. In the data acquisition influence area, the dynamic changes of dust are monitored in real time to analyze the impact of dust on image acquisition accuracy.
[0012] Process the collected coal images, extract the surface characteristic information of the coal samples, and understand the surface conditions of different coal samples;
[0013] Conduct a comprehensive analysis of the image acquisition environment during coal detection and evaluate the impact of the image acquisition environment on image detection results;
[0014] Based on the analysis results, the image acquisition environment is divided into high-impact environment and low-impact environment;
[0015] In low-impact environments, maintain the optimal detection rate and continue to conduct coal testing. For high-impact environments, adjust the actual detection rate during coal testing based on the optimal detection rate and reduce the actual detection rate to ensure that in complex environments, the detection system can still accurately capture subtle quality problems in coal.
[0016] Preferably, the conical area between the image acquisition device and the acquisition area is defined as the data acquisition influence area, and the specific steps are as follows:
[0017] Determine the installation position and angle of the image acquisition device;
[0018] After determining the position and angle of the image acquisition device, measure the vertical distance from the camera lens to the conveyor belt surface;
[0019] The camera's field of view is calculated based on the camera's viewing angle and the measured vertical distance. The field of view is a cone-shaped area.
[0020] The calculated cone area is marked in the actual production environment, and the data collection influence area is delineated on both sides and the top of the conveyor belt using software tools.
[0021] Preferably, the dynamic changes of dust are monitored in real time, which includes obtaining the dust particle diameter information in the data acquisition influence area, analyzing the dust particle diameter information within the detection time window, generating a dust particle diameter excess index, and analyzing the impact of dust on image acquisition accuracy through the dust particle diameter excess index.
[0022] Preferably, the steps of obtaining dust particle diameter information in the data collection influence area, analyzing the dust particle diameter information within the detection time window and generating a dust particle diameter exceeding standard index are as follows:
[0023] Within the detection time window, the laser particle size analyzer is used to collect data in real time to collect the dust particle diameter data in the affected area, and the collected dust particle diameter data are used to establish a data set, and the data set is calibrated as D ,but ,in Indicates l The diameter of a dust particle,n Indicates the total number of dust particles obtained within the detection time window;
[0024] The collected dust particle diameter data is divided into several intervals at equal intervals, and the number of particles in each interval is counted. k intervals, each of which has a range of ,in, , Indicates The upper limit of the interval;
[0025] Calculate the proportion of the number of particles in each interval to the total number of particles. u The number of particles in the interval is , then u The proportion of particles in each interval is: , where Indicates u The proportion of particles in each interval;
[0026] In order to reflect the impact of large particles on detection, a weight is assigned to each interval. The calculation expression of the weight is: , where Indicates u The weight of each interval;
[0027] Calculate the weighted particle quantity ratio in each interval. The calculation expression is: , where Indicates u The weighted particle quantity ratio of each interval;
[0028] The dust particle diameter excess index is calculated by comprehensively considering the weighted particle quantity ratios in all intervals. The calculation expression is: , where Indicates the dust particle diameter exceeding the standard index.
[0029] Preferably, surface characteristic information of the coal sample is extracted, wherein the surface characteristic information of the coal sample includes the surface color of the coal. After analyzing and processing the surface color of the coal within the detection time window, a coal surface color difference index is generated, and the surface conditions of different coal samples can be understood through the coal surface color difference index.
[0030] Preferably, within the detection time window, the steps of analyzing and processing the coal surface color to generate the coal surface color difference index are as follows:
[0031] Preprocessing the images acquired by the image acquisition device within the detection time window;
[0032] Convert the pre-processed image from RGB color space to CIE Lab* color space for more accurate color analysis;
[0033] The image is processed in blocks. For each small block, the local color difference value is calculated based on the high-order color difference, combined with the color gradient and edge information. The calculation expression is: , where i , j represents adjacent pixels in the image, , , Indicates that within time t i The Lab* value of pixels, where Indicates that within time t i The brightness of a pixel, Indicates that within time t i The chromaticity value of a pixel, Indicates that within time t i The chromaticity value of a pixel, , , Indicates that within time t j The Lab* value of pixels, where Indicates that within time t j The brightness of a pixel, Indicates that within time t j The chromaticity value of a pixel, Indicates that within time t j The chromaticity value of a pixel, and are the horizontal color gradient and vertical color gradient within time t, and Respectively represent the horizontal pixel spacing and vertical pixel spacing, indicating the distance between adjacent pixels, Represents the local color difference value within time t, indicating the color difference between adjacent pixels;
[0034] Combining all local color difference values, the coal surface color difference index is calculated, and the calculation expression is: , Indicates that in the detection time window Internally generated coal surface color difference index, and Respectively represent the start and end time of the detection time window, N and M are the number of rows and columns of the image, respectively. Represents the color weighting coefficient within time t, which is used to adjust the weights of different color differences.
[0035] Preferably, within the detection time window, after obtaining the dust particle diameter excess index generated by analyzing the dust particle diameter information during coal detection and the coal surface color difference index generated by analyzing and processing the coal surface color, the dust particle diameter excess index and the coal surface color difference index are input into a pre-trained machine learning model, and an influence coefficient is generated through the machine learning model. The influence coefficient is used to evaluate the impact of the image acquisition environment on the image detection results.
[0036] Preferably, the influence coefficient generated by comprehensive analysis of the image acquisition environment during coal detection in the detection time window is compared with a preset reference threshold of the influence coefficient to divide the image acquisition environment. The specific steps are as follows:
[0037] If the influence coefficient is greater than or equal to the influence coefficient reference threshold, the image acquisition environment is classified as a high-impact environment;
[0038] If the impact coefficient is less than the impact coefficient reference threshold, the image acquisition environment is classified as a low-impact environment.
[0039] Preferably, the specific steps for regulating the actual detection rate during coal detection are as follows:
[0040] In a low-impact environment, the optimal detection rate is maintained to continue coal detection. The specific expression is: , where Indicates the actual detection rate when detecting coal in a low-impact environment, represents the optimal detection rate;
[0041] For high-impact environments, the actual detection rate during coal detection is regulated according to the optimal detection rate to reduce the actual detection rate to ensure that in complex environments, the specific regulation expression is: , where Indicates the standardized value of the impact coefficient, which reflects the impact of the current environment on the detection rate. , Indicates the actual detection rate when detecting coal in a high-impact environment, represents the influence coefficient, Indicates the reference threshold of the influence coefficient, represents the maximum value of the influence coefficient, represents the regulatory factor, controlling the decrease in the optimal detection rate;
[0042] Monitor the impact coefficient and actual detection rate in real time, and dynamically adjust the detection strategy. The expression for dynamically adjusting the detection strategy is: , where It indicates the actual detection rate of coal during detection within the detection time window.
[0043] A coal quality detection system based on image data processing, comprising an optimal detection rate setting module, a dust dynamic monitoring module, an image processing and feature extraction module, an image acquisition environment assessment module, an environmental impact classification module and a detection rate dynamic control module;
[0044] The optimal detection rate setting module sets an optimal detection rate, i.e., the rate at which the conveyor belt conveys coal, based on historical experience and previous detection data. At the optimal detection rate, an image acquisition device is used to collect images of coal samples in real time for coal detection;
[0045] The dust dynamic monitoring module defines the conical area between the image acquisition device and the acquisition area as the data acquisition influence area. In the data acquisition influence area, the dynamic changes of dust are monitored in real time to analyze the impact of dust on image acquisition accuracy.
[0046] The image processing and feature extraction module processes the collected coal images, extracts the surface feature information of the coal samples, and understands the surface conditions of different coal samples;
[0047] Image acquisition environment assessment module, which conducts comprehensive analysis on the image acquisition environment during coal detection and assesses the impact of the image acquisition environment on the image detection results;
[0048] Environmental impact classification module, which divides the image acquisition environment into high-impact environment and low-impact environment based on the analysis results;
[0049] The detection rate dynamic control module maintains the optimal detection rate to continue coal detection in low-impact environments. For high-impact environments, the actual detection rate during coal detection is adjusted according to the optimal detection rate, and the actual detection rate is reduced to ensure that the detection system can still accurately capture subtle quality problems of coal in complex environments.
[0050] In the above technical solution, the technical effects and advantages provided by the present invention are:
[0051] The present invention obtains dust particle diameter information and generates a dust particle diameter exceeding standard index by real-time monitoring of dust dynamic changes in the data acquisition influence area, thereby quantifying the influence of dust on image acquisition accuracy. At the same time, surface feature information of coal samples is extracted, the coal surface color is analyzed and a coal surface color difference index is generated to further evaluate the influence of coal surface conditions on image acquisition. The dust particle diameter exceeding standard index and the coal surface color difference index are input into a pre-trained machine learning model, and the influence of the image acquisition environment on the image detection result is evaluated by the generated influence coefficient. Based on the evaluation result, the image acquisition environment is divided into a high-impact environment and a low-impact environment, so that appropriate detection strategies are adopted in different environments to ensure the accuracy of image acquisition and the reliability of detection results, thereby reducing the occurrence of missed detection and false detection.
[0052] The present invention maintains the optimal detection rate and ensures detection efficiency in a low-impact environment. In a high-impact environment, the actual detection rate is reduced to ensure that the detection system can accurately capture subtle quality problems of coal in a complex environment. Through this dynamic control mechanism, the detection accuracy and efficiency can be effectively balanced, the entry of inferior coal into the combustion process can be reduced, the combustion efficiency can be improved, environmental pollution can be reduced, and resource waste can be avoided, thereby improving the economic benefits and resource utilization efficiency of the enterprise. BRIEF DESCRIPTION OF THE DRAWINGS
[0053] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0054] Figure 1 The present invention is a method flow chart of a coal quality detection method based on image data processing.
[0055] Figure 2 The present invention is a module schematic diagram of a coal quality detection system based on image data processing. DETAILED DESCRIPTION
[0056] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.
[0057] The present invention provides Figure 1 A coal quality detection method based on image data processing is shown, comprising the following steps:
[0058] According to historical experience and previous detection data, an optimal detection rate is set, that is, the rate at which the conveyor belt conveys coal. At the optimal detection rate, an image acquisition device (high-definition camera or industrial camera) is used to collect images of coal samples in real time for coal detection;
[0059] The optimal detection rate is set by analyzing previous detection data and historical experience, evaluating the detection accuracy, efficiency and equipment performance under different transmission rates, and selecting a transmission rate that can normally ensure the stable operation of the image acquisition and processing system under specific environmental conditions, accurately identify coal quality characteristics, and maximize production efficiency.
[0060] The conical area between the image acquisition device and the acquisition area is defined as the data acquisition influence area. In the data acquisition influence area, the dynamic changes of dust are monitored in real time to analyze the impact of dust on image acquisition accuracy.
[0061] Define the conical area between the image acquisition device and the acquisition area as the data acquisition influence area. The specific steps are as follows:
[0062] Determine the installation position and angle of the image acquisition device (HD camera or industrial camera);
[0063] The position and angle of the image acquisition device directly affect its field of view. The installation position should be selected above or on the side of the conveyor belt so that it can fully cover the moving path of the coal sample. Adjust the angle of the camera to ensure the best image acquisition viewing angle and avoid image distortion or acquisition blind spots caused by too large or too small angles.
[0064] After determining the position and angle of the image acquisition device, measure the vertical distance from the camera lens to the conveyor belt surface (i.e., the coal sample surface);
[0065] This distance determines the clarity and coverage of the image acquisition. Through actual measurement, accurate data is obtained to calculate the field of view cone of the camera device. Make sure the distance is moderate so that a clear image can still be obtained at a high transmission rate.
[0066] The camera's field of view is calculated based on the camera's viewing angle and the measured vertical distance. The field of view is a cone-shaped area.
[0067] This is done through geometric calculations by calculating the base width and height of a field of view cone based on the camera’s horizontal field of view (FOV) and distance. This cone represents the area of the coal sample that the camera can cover, ensuring that the entire conveyor belt width is within the camera’s field of view.
[0068] Mark the calculated cone area in the actual production environment and use software tools to define the data collection impact area on both sides and the top of the conveyor belt;
[0069] In this area, environmental monitoring sensors, such as dust sensors and light intensity sensors, are installed to monitor the dynamic changes of environmental factors in real time. Ensuring the stability of environmental conditions in this area helps improve the quality of image acquisition and detection accuracy.
[0070] Computer-aided design (CAD) software, image processing software (such as OpenCV), and industrial automation software (such as LabVIEW) can be used to delineate the data acquisition impact area. These tools can help accurately draw and simulate the cone area of the camera's field of view, and combine sensor data to monitor environmental conditions in real time, thereby optimizing the image acquisition process and improving detection accuracy.
[0071] Real-time monitoring of the dynamic changes of dust, including obtaining the dust particle diameter information in the data collection influence area, analyzing the dust particle diameter information within the detection time window, generating a dust particle diameter excess index, and analyzing the impact of dust on image acquisition accuracy through the dust particle diameter excess index.
[0072] When the conical area between the image acquisition device and the acquisition area is designated as the data acquisition impact area, if the dust particles in this area are large in diameter and the coal transmission rate is fast, the accuracy of coal detection will be seriously affected. Large dust particles will scatter and absorb the light of the camera, resulting in blurred images, reduced contrast and loss of details. In addition, high-speed coal transmission will increase motion blur in the image, making it difficult for the system to capture tiny features and defects on the coal surface. In this case, the detection system may not be able to accurately identify the quality characteristics of the coal, such as ash, moisture and sulfur content, resulting in missed detections and false detections.
[0073] The steps of obtaining the dust particle diameter information in the data collection influence area and analyzing the dust particle diameter information within the monitoring time window to generate the dust particle diameter exceeding standard index are as follows: within the detection time window, use the laser particle size analyzer to collect the dust particle diameter data in the data collection influence area in real time, establish a data set with the collected dust particle diameter data, and calibrate the data set as D ,but ,in Indicates l The diameter of a dust particle, n Indicates the total number of dust particles obtained within the detection time window;
[0074] The collected dust particle diameter data is divided into several intervals at equal intervals, and the number of particles in each interval is counted.k intervals, each of which has a range of ,in, , Indicates The upper limit of the interval;
[0075] Calculate the proportion of the number of particles in each interval to the total number of particles. u The number of particles in the interval is , then u The proportion of particles in each interval is: , where Indicates u The proportion of particles in each interval;
[0076] In order to reflect the impact of large particles on detection, a weight is assigned to each interval. The calculation expression of the weight is: , where Indicates u The weight of each interval;
[0077] Larger dust particles have a greater impact on image quality and detection results. Therefore, by giving larger particle intervals a higher weight, their influence on the detection process can be highlighted, thereby more accurately reflecting the impact of the environment on coal quality detection.
[0078] Calculate the weighted particle quantity ratio in each interval. The calculation expression is: , where Indicates u The weighted particle quantity ratio of each interval;
[0079] The dust particle diameter excess index is calculated by comprehensively considering the weighted particle quantity ratios in all intervals. The calculation expression is: , where Indicates the dust particle diameter exceeding the standard index.
[0080] It can be seen from the dust particle diameter excess index that within the detection time window, the larger the performance value of the dust particle diameter excess index generated after analyzing the dust particle diameter information, the larger the dust particle diameter is, and the greater the impact on the accuracy of coal detection. On the contrary, the smaller the performance value of the dust particle diameter excess index is, the smaller the dust particle diameter is, and the smaller the impact on the accuracy of coal detection is. Therefore, the dust particle diameter excess index can be used as a key indicator to evaluate the degree of dust impact in the detection environment.
[0081] Process the collected coal images, extract the surface characteristic information of the coal samples, and understand the surface conditions of different coal samples;
[0082] The surface characteristic information of the coal samples is extracted, wherein the surface characteristic information of the coal samples includes the surface color of the coal. Within the detection time window, after analyzing and processing the surface color of the coal, a coal surface color difference index is generated. The surface conditions of different coal samples can be understood through the coal surface color difference index.
[0083] When the surface color difference of coal samples is large and the coal transmission rate is fast, it will have a serious impact on the accuracy of coal detection. High-speed transmission will make it difficult for image acquisition equipment to accurately capture the details of fast-moving coal, especially when the color difference is large. The color change on the coal surface will cause misjudgment of the image processing algorithm. When coal samples with large color difference are transmitted at high speed, the image may be blurred and distorted, resulting in the image processing system being unable to correctly extract the surface feature information of the coal, such as cracks, particle size and shape. This will not only increase the probability of missed detection and false detection, but may also lead to inaccurate quality assessment, affecting the stability of the production process and product quality.
[0084] The steps for analyzing and processing the coal surface color within the detection time window to generate the coal surface color difference index are as follows:
[0085] Preprocessing the images acquired by the image acquisition device within the detection time window;
[0086] Preprocessing includes grayscale transformation, noise filtering and image enhancement, which helps to improve image quality.
[0087] Convert the pre-processed image from RGB color space to CIE Lab* color space for more accurate color analysis;
[0088] The RGB color space is a color model that consists of three basic color channels: red, green, and blue. Various colors are represented by different combinations and proportions. Each channel is usually represented by 8 bits, so an RGB color pixel can have 256 levels of red, green, and blue intensity, which together constitute 16777216 different colors.
[0089] The image is processed in blocks. For each small block, the local color difference value is calculated based on the high-order color difference, combined with the color gradient and edge information. The calculation expression is: , where i , j represents adjacent pixels in the image, , , Indicates that within time t i The Lab* value of pixels, where Indicates that within time t iThe brightness of each pixel, the brightness value ranges from 0 to 100, representing the brightness level from black to white. Indicates that within time t i The chromaticity value of the pixel along the red-green axis, a The axis value range is usually -128 to 127, with negative values representing green and positive values representing red. Indicates that within time t i The chromaticity value of pixels along the yellow-blue axis, b The axis value range is usually -128 to 127, with negative values representing blue and positive values representing yellow. , , Indicates that within time t j The Lab* value of pixels, where Indicates that within time t j The brightness of a pixel, Indicates that within time t j The chromaticity value of a pixel, Indicates that within time t j The chromaticity value of a pixel, and are the horizontal color gradient and vertical color gradient within time t, and Respectively represent the horizontal pixel spacing and vertical pixel spacing, indicating the distance between adjacent pixels, Represents the local color difference value within time t, indicating the color difference between adjacent pixels;
[0090] Combining all local color difference values, the coal surface color difference index is calculated, and the calculation expression is: , Indicates that in the detection time window Internally generated coal surface color difference index, and Respectively represent the start and end time of the detection time window, N and M are the number of rows and columns of the image, respectively. Represents the color weighting coefficient within time t, which is used to adjust the weights of different color differences.
[0091] Through the coal surface color difference index, it can be concluded that within the detection time window, the larger the coal surface color difference index generated after analyzing and processing the coal surface color, the greater the change in the surface color of the coal sample, and thus the greater the impact on the accuracy of coal detection. This means that when the coal surface color difference index is high, it may be more difficult for the detection system to accurately identify and classify the quality characteristics of the coal sample, resulting in inaccurate and reduced reliability of the test results. On the contrary, a lower coal surface color difference index indicates that the color change of the coal surface is smaller, which has less impact on the detection accuracy, and the detection system can more reliably perform quality assessment and classification, thereby improving the accuracy and stability of the detection.
[0092] Conduct a comprehensive analysis of the image acquisition environment during coal detection and evaluate the impact of the image acquisition environment on image detection results;
[0093] In the detection time window, the dust particle diameter excess index generated by analyzing the dust particle diameter information during coal detection and the coal surface color difference index generated by analyzing and processing the coal surface color are obtained, and then the dust particle diameter excess index and the coal surface color difference index are input into the pre-trained machine learning model, and the influence coefficient is generated by the machine learning model. The influence coefficient is used to evaluate the impact of the image acquisition environment on the image detection results.
[0094] A pre-trained machine learning model refers to a model that is trained with a large amount of historical data before actual application to learn and establish the relationship between input variables (such as dust particle diameter excess index and coal surface color difference index) and output results (such as the degree to which image detection results are affected by the environment). This model can accurately predict and evaluate the impact of the image acquisition environment on coal quality detection results within the detection time window by analyzing the training data and adjusting its internal parameters. The training process includes data preprocessing, feature extraction, model selection, training and verification to ensure the accuracy and robustness of the model on new data.
[0095] The machine learning model is not specifically limited here, and can generate the influence coefficient by comprehensively analyzing the dust particle diameter excess index and the coal surface color difference index. In order to realize the technical solution of the present invention, the present invention provides a specific implementation method;
[0096] Influence coefficient The resulting calculation formula is: , where , Dust particle diameter excess index and coal surface color difference index The preset scaling factor of , Both are greater than 0.
[0097] It can be seen from the influence coefficient that, within the detection time window, the larger the performance value of the dust particle diameter excess index generated after analyzing the dust particle diameter information, the larger the dust particle diameter is; the larger the coal surface color difference index generated after analyzing and processing the coal surface color, the greater the surface color change of the coal sample, and the greater the impact on the accuracy of coal detection, and vice versa.
[0098] Based on the analysis results, the image acquisition environment is divided into high-impact environment and low-impact environment;
[0099] The influence coefficient generated by comprehensive analysis of the image acquisition environment during coal detection in the detection time window is compared with the pre-set reference threshold of the influence coefficient to divide the image acquisition environment. The specific steps are as follows:
[0100] If the influence coefficient is greater than or equal to the influence coefficient reference threshold, the image acquisition environment is classified as a high-impact environment;
[0101] If the impact coefficient is less than the impact coefficient reference threshold, the image acquisition environment is classified as a low-impact environment.
[0102] In low-impact environments, the optimal detection rate is maintained to continue coal testing. For high-impact environments, the actual detection rate during coal testing is adjusted according to the optimal detection rate, and the actual detection rate is reduced to ensure that the detection system can still accurately capture subtle quality problems of coal in complex environments;
[0103] The specific steps for regulating the actual detection rate during coal detection are as follows:
[0104] In a low-impact environment, the optimal detection rate is maintained to continue coal detection. The specific expression is: , where Indicates the actual detection rate when detecting coal in a low-impact environment, represents the optimal detection rate;
[0105] For high-impact environments, the actual detection rate during coal detection is regulated according to the optimal detection rate to reduce the actual detection rate to ensure that in complex environments, the specific regulation expression is: , where Indicates the standardized value of the impact coefficient, which reflects the impact of the current environment on the detection rate. , Indicates the actual detection rate when detecting coal in a high-impact environment, represents the influence coefficient, Indicates the reference threshold of the influence coefficient, represents the maximum value of the influence coefficient, represents the regulatory factor, controlling the decrease in the optimal detection rate;
[0106] Among them, regulatory factors The specific value of is not limited here and can be set according to actual needs.
[0107] The maximum value of the impact coefficient refers to the highest value that the impact coefficient can reach under extremely adverse environmental conditions. This value can be determined by analyzing and simulating the worst detection environment in historical data. Specifically, the extreme values of environmental parameters in historical data can be collected, and then the impact coefficients under these extreme conditions can be calculated using a pre-trained machine learning model to finally obtain the maximum value of the impact coefficient. This maximum value is used to calibrate and standardize the calculation of the impact coefficient to ensure effective detection rate adjustment under various environmental conditions.
[0108] Monitor the impact coefficient and actual detection rate in real time, and dynamically adjust the detection strategy. The expression for dynamically adjusting the detection strategy is: , where It indicates the actual detection rate of coal during detection within the detection time window.
[0109] The present invention obtains dust particle diameter information and generates a dust particle diameter exceeding standard index by real-time monitoring of dust dynamic changes in the data acquisition influence area, thereby quantifying the influence of dust on image acquisition accuracy. At the same time, surface feature information of coal samples is extracted, the coal surface color is analyzed and a coal surface color difference index is generated to further evaluate the influence of coal surface conditions on image acquisition. The dust particle diameter exceeding standard index and the coal surface color difference index are input into a pre-trained machine learning model, and the influence of the image acquisition environment on the image detection result is evaluated by the generated influence coefficient. Based on the evaluation result, the image acquisition environment is divided into a high-impact environment and a low-impact environment, so that appropriate detection strategies are adopted in different environments to ensure the accuracy of image acquisition and the reliability of detection results, thereby reducing the occurrence of missed detection and false detection.
[0110] The present invention maintains the optimal detection rate and ensures detection efficiency in a low-impact environment. In a high-impact environment, the actual detection rate is reduced to ensure that the detection system can accurately capture subtle quality problems of coal in a complex environment. Through this dynamic control mechanism, the detection accuracy and efficiency can be effectively balanced, the entry of inferior coal into the combustion process can be reduced, the combustion efficiency can be improved, environmental pollution can be reduced, and resource waste can be avoided, thereby improving the economic benefits and resource utilization efficiency of the enterprise.
[0111] The present invention provides Figure 2A coal quality detection system based on image data processing is shown, comprising an optimal detection rate setting module, a dust dynamic monitoring module, an image processing and feature extraction module, an image acquisition environment assessment module, an environmental impact classification module and a detection rate dynamic control module;
[0112] The optimal detection rate setting module sets an optimal detection rate, i.e., the rate at which the conveyor belt conveys coal, based on historical experience and previous detection data. At the optimal detection rate, an image acquisition device is used to collect images of coal samples in real time for coal detection;
[0113] The dust dynamic monitoring module defines the conical area between the image acquisition device and the acquisition area as the data acquisition influence area. In the data acquisition influence area, the dynamic changes of dust are monitored in real time to analyze the impact of dust on image acquisition accuracy.
[0114] The image processing and feature extraction module processes the collected coal images, extracts the surface feature information of the coal samples, and understands the surface conditions of different coal samples;
[0115] Image acquisition environment assessment module, which conducts comprehensive analysis on the image acquisition environment during coal detection and assesses the impact of the image acquisition environment on the image detection results;
[0116] Environmental impact classification module, which divides the image acquisition environment into high-impact environment and low-impact environment based on the analysis results;
[0117] The detection rate dynamic control module maintains the optimal detection rate to continue coal detection in low-impact environments. For high-impact environments, the actual detection rate during coal detection is regulated according to the optimal detection rate to reduce the actual detection rate, so as to ensure that the detection system can still accurately capture subtle quality problems of coal in complex environments.
[0118] A coal quality detection method based on image data processing provided in an embodiment of the present invention is realized by the above-mentioned coal quality detection system based on image data processing. The specific method and process of a coal quality detection system based on image data processing are detailed in the embodiment of the above-mentioned coal quality detection method based on image data processing, which will not be repeated here.
[0119] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.
[0120] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
[0121] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.
Claims
1. A coal quality detection method based on image data processing, characterized in that: The following steps are involved: According to historical experience and previous detection data, an optimal detection rate is set, that is, the rate at which the conveyor belt conveys coal. At the optimal detection rate, an image acquisition device is used to collect images of coal samples in real time for coal detection; The conical area between the image acquisition device and the acquisition area is defined as the data acquisition influence area. In the data acquisition influence area, the dynamic changes of dust are monitored in real time to analyze the impact of dust on image acquisition accuracy. Real-time monitoring of the dynamic changes of dust, including obtaining the dust particle diameter information in the data acquisition impact area, analyzing the dust particle diameter information within the detection time window, generating a dust particle diameter excess index, and analyzing the impact of dust on image acquisition accuracy through the dust particle diameter excess index; The steps for obtaining the dust particle diameter information within the data collection influence area and analyzing the dust particle diameter information within the detection time window to generate the dust particle diameter excess index are as follows: Within the detection time window, the laser particle size analyzer is used to collect data in real time to collect the dust particle diameter data in the affected area, and the collected dust particle diameter data are used to establish a data set, and the data set is calibrated as D ,but ,in Indicates l The diameter of a dust particle, n Indicates the total number of dust particles obtained within the detection time window; The collected dust particle diameter data is divided into several intervals at equal intervals, and the number of particles in each interval is counted. k intervals, each of which has a range of ,in, , Indicates The upper limit of the interval; Calculate the proportion of the number of particles in each interval to the total number of particles. u The number of particles in the interval is , then u The proportion of particles in each interval is: , where Indicates u The proportion of particles in each interval; In order to reflect the impact of large particles on detection, a weight is assigned to each interval. The calculation expression of the weight is: , where Indicates u The weight of each interval; Calculate the weighted particle quantity ratio in each interval. The calculation expression is: , where Indicates u The weighted particle quantity ratio of each interval; The dust particle diameter excess index is calculated by comprehensively considering the weighted particle quantity ratios in all intervals. The calculation expression is: , where Indicates the dust particle diameter exceeding the standard index; Process the collected coal images, extract the surface characteristic information of the coal samples, and understand the surface conditions of different coal samples; Extracting surface characteristic information of coal samples, wherein the surface characteristic information of coal samples includes coal surface color. After analyzing and processing the coal surface color within the detection time window, a coal surface color difference index is generated, and the surface conditions of different coal samples are understood through the coal surface color difference index. The steps for analyzing and processing the coal surface color within the detection time window to generate the coal surface color difference index are as follows: Preprocessing the images acquired by the image acquisition device within the detection time window; Convert the pre-processed image from RGB color space to CIE Lab* color space for more accurate color analysis; The image is processed in blocks. For each small block, the local color difference value is calculated based on the high-order color difference, combined with the color gradient and edge information. The calculation expression is: , where i , j represents adjacent pixels in the image, , , Indicates that within time t i The Lab* value of pixels, where Indicates that within time t i The brightness of a pixel, Indicates that within time t i The chromaticity value of a pixel, Indicates that within time t i The chromaticity value of a pixel, , , Indicates that within time t j The Lab* value of pixels, where Indicates that within time t j The brightness of a pixel, Indicates that within time t j The chromaticity value of a pixel, Indicates that within time t j The chromaticity value of a pixel, and are the horizontal color gradient and vertical color gradient within time t, and Respectively represent the horizontal pixel spacing and vertical pixel spacing, indicating the distance between adjacent pixels, Represents the local color difference value within time t, indicating the color difference between adjacent pixels; Combining all local color difference values, the coal surface color difference index is calculated, and the calculation expression is: , Indicates that in the detection time window Internally generated coal surface color difference index, and Respectively represent the start and end time of the detection time window, N and M are the number of rows and columns of the image, respectively. Represents the color weighting coefficient within time t, which is used to adjust the weights of different color differences; Conduct a comprehensive analysis of the image acquisition environment during coal detection and evaluate the impact of the image acquisition environment on image detection results; In the detection time window, after obtaining the dust particle diameter excess index generated by analyzing the dust particle diameter information during coal detection and the coal surface color difference index generated by analyzing and processing the coal surface color, the dust particle diameter excess index and the coal surface color difference index are input into the pre-trained machine learning model, and the influence coefficient is generated by the machine learning model. The influence of the image acquisition environment on the image detection result is evaluated by the influence coefficient; Based on the analysis results, the image acquisition environment is divided into high-impact environment and low-impact environment; The influence coefficient generated by comprehensive analysis of the image acquisition environment during coal detection in the detection time window is compared with the pre-set reference threshold of the influence coefficient to divide the image acquisition environment. The specific steps are as follows: If the influence coefficient is greater than or equal to the influence coefficient reference threshold, the image acquisition environment is classified as a high-impact environment; If the impact coefficient is less than the impact coefficient reference threshold, the image acquisition environment is classified as a low-impact environment; In low-impact environments, the optimal detection rate is maintained to continue coal testing. For high-impact environments, the actual detection rate during coal testing is adjusted according to the optimal detection rate, and the actual detection rate is reduced to ensure that the detection system can still accurately capture subtle quality problems of coal in complex environments; The specific steps for regulating the actual detection rate during coal detection are as follows: In a low-impact environment, the best detection rate is maintained to continue coal detection. The specific expression is: , where Indicates the actual detection rate when detecting coal in a low-impact environment, represents the optimal detection rate; For high-impact environments, the actual detection rate during coal detection is regulated according to the optimal detection rate to reduce the actual detection rate to ensure that in complex environments, the specific regulation expression is: , where Indicates the standardized value of the impact coefficient, which reflects the impact of the current environment on the detection rate. , Indicates the actual detection rate when detecting coal in a high-impact environment, represents the influence coefficient, Indicates the reference threshold of the influence coefficient, represents the maximum value of the influence coefficient, represents the regulatory factor, controlling the decrease in the optimal detection rate; Monitor the impact coefficient and actual detection rate in real time, and dynamically adjust the detection strategy. The expression for dynamically adjusting the detection strategy is: , where It indicates the actual detection rate of coal during detection within the detection time window.
2. A coal quality detection method based on image data processing according to claim 1, characterized in that: Define the conical area between the image acquisition device and the acquisition area as the data acquisition influence area. The specific steps are as follows: Determine the installation position and angle of the image acquisition device; After determining the position and angle of the image acquisition device, measure the vertical distance from the camera lens to the conveyor belt surface; The camera's field of view is calculated based on the camera's viewing angle and the measured vertical distance. The field of view is a cone-shaped area. The calculated cone area is marked in the actual production environment, and the data collection influence area is delineated on both sides and the top of the conveyor belt using software tools.
3. A coal quality detection system based on image data processing, used to implement a coal quality detection method based on image data processing as described in any one of claims 1-2, characterized in that: It includes an optimal detection rate setting module, a dust dynamic monitoring module, an image processing and feature extraction module, an image acquisition environment assessment module, an environmental impact classification module, and a detection rate dynamic control module; The optimal detection rate setting module sets an optimal detection rate, i.e., the rate at which the conveyor belt conveys coal, based on historical experience and previous detection data. At the optimal detection rate, an image acquisition device is used to collect images of coal samples in real time for coal detection; The dust dynamic monitoring module defines the conical area between the image acquisition device and the acquisition area as the data acquisition influence area. In the data acquisition influence area, the dynamic changes of dust are monitored in real time to analyze the impact of dust on image acquisition accuracy. The image processing and feature extraction module processes the collected coal images, extracts the surface feature information of the coal samples, and understands the surface conditions of different coal samples; Image acquisition environment assessment module, which conducts comprehensive analysis on the image acquisition environment during coal detection and assesses the impact of the image acquisition environment on the image detection results; Environmental impact classification module, which divides the image acquisition environment into high-impact environment and low-impact environment based on the analysis results; The detection rate dynamic control module maintains the optimal detection rate to continue coal detection in low-impact environments. For high-impact environments, the actual detection rate during coal detection is adjusted according to the optimal detection rate, and the actual detection rate is reduced to ensure that the detection system can still accurately capture subtle quality problems of coal in complex environments.
Citation Information
Patent Citations
Apple grading identification method based on deep learning
CN111915704A
Coal gangue target detection method based on improved YOLOv5s model
CN114663407A
