Method and system for detecting running state of mining carrier roller

Abstract

The invention relates to the technical field of mining conveying belts, and particularly provides a method and system for detecting the running state of a mining carrier roller, and the method comprises the steps: obtaining an infrared image and a visible light image of the mining carrier roller at the same angle, storing the infrared image and the visible light image at the same time sequence in a one-to-one correspondence manner, generating a time sequence image file, and storing the time sequence image file into a database; performing preprocessing and image analysis on the time sequence image file of each detection period, extracting a first texture feature and a second texture feature, judging whether the first texture feature and the second texture feature belong to abnormal texture features or not, and analyzing the running state of the mining carrier roller in combination with the temperature sensing feature when the abnormal texture features appear. Whether the vibration of the mining carrier roller is normal or not is judged through the first texture feature, whether the surface quality of the mining carrier roller is abnormal or not is judged through the second texture feature, when the abnormal texture feature appears, the abnormity is predicted in combination with the temperature, and then the running state of the mining carrier roller is effectively detected and predicted in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of mine conveyor belts, and in particular, to a method and system for detecting the operating state of mine idlers. Background Art

[0002] A mine idler is a type of idler specifically designed for the mining and quarrying industries, used to support and guide conveyor belt systems for the smooth transportation of heavy materials. Compared with ordinary idlers, mine idlers have stronger durability and load-bearing capacity. However, due to the harsh working conditions in the mine environment and the transportation requirements of heavy materials, mine idlers are usually used to transport heavy materials. Prolonged friction and impact may cause wear on the surface of the idlers. If the material distribution is uneven, some idlers may wear faster than others. Moreover, mine idlers usually need to withstand the impact and pressure of heavy materials. If the impact and load are too large, the structure of the idler may be damaged, resulting in abnormal vibration during operation, thus affecting its normal operation. Therefore, it is very important to detect the operating state of the idlers in a timely manner, not only to reduce the occurrence of accidents, but also to promptly discover abnormalities, prevent further wear of the mine idlers, and protect the normal operation of the mine idlers and conveyor devices.

[0003] Therefore, the detection of the operating state of mine idlers is crucial for the mining and quarrying industries. By timely and accurately monitoring and evaluating the operating state of the idlers, potential failures and accidents can be prevented, and the safety and reliability of the conveyor belt system can be improved. However, due to the complexity of data analysis and fault diagnosis, there is currently no method for comprehensively detecting the operating state of mine idlers.

[0004] Therefore, there is an urgent need to design a method and system for detecting the operating state of mine idlers to solve the technical problem that it is currently difficult to detect the operating state of mine idlers. Summary of the Invention

[0005] In view of this, the present invention proposes a method and system for detecting the operating state of mine idlers, aiming to solve the technical problem that it is difficult to timely and accurately monitor and evaluate the operating state of mine idlers in the prior art.

[0006] The present invention proposes a method for detecting the operating state of mine idlers, including:

[0007] Step S1: Mark feature points on the physical object of the mine idler, obtain infrared images and visible light images of the mine idler at the same angle, and store the infrared images and the visible light images corresponding to the same time sequence one by one to generate a time sequence image file, and construct a feature region for each feature point in each time sequence image file;

[0008] Step S2: Record the sequential image files within each unit of time. Before the start of each unit of time, there is a preparation stage. During the preparation stage, the characteristic points of the mining idler are aligned with preset calibration points for static calibration, and the component information of the mining idler is corresponding to the characteristic points and the characteristic regions; each unit of time also includes a first moment. At the first moment, the mining idler rotates to the calibration point. Starting from the first moment, the infrared images and the visible light images within the unit of time are associated frame by frame. And a detection period is preset according to the total working duration. The unit rotation speed is calibrated according to the rotation speed of the mining idler within the detection period.

[0009] Step S3: Perform preprocessing and image analysis on the sequential image files of each detection period. The image analysis includes performing frequency domain conversion on the preprocessed visible light image to obtain a frequency domain image, obtaining the spectrogram of the mining idler at each unit rotation speed according to the frequency domain image, performing first texture analysis on the spectrogram, extracting first texture features, and analyzing the frequency distribution in the spectrogram, extracting the edge information of the mining idler, and performing second texture analysis on the edge information to extract second texture features.

[0010] Step S4: Use cosine similarity to evaluate the similarity between the first texture features and the second texture features and their preset standard texture features respectively. When the similarity is lower than the standard similarity, it is judged as an abnormal texture feature, and the feature region to which the abnormal texture feature belongs is obtained and set as an abnormal feature region, and the visible light image with the abnormal texture feature is marked as an abnormal state image.

[0011] Step S5: Obtain the infrared image of the same frame associated with the abnormal state image, and extract the feature region corresponding to the abnormal feature region in the infrared image of the same frame for temperature sensing analysis, extract temperature sensing features, and comprehensively analyze the operating state of the mining idler.

[0012] Preferably, in step S1, marking the characteristic points on the physical object of the mining idler includes:

[0013] Obtain the geometric center point of each component of the mining idler as the characteristic point for marking as a structural feature recognition point. The structural feature recognition point is provided with obvious different features, including polygon edges and distinct textures; among them, when the component directly cooperates with the mining conveyor drive component, a temperature isolation coating is used to re-mark the structural feature recognition point as a temperature feature recognition point.

[0014] Construct a feature region for each characteristic point in each sequential image file, including:

[0015] Each of the feature regions covers each outer edge of the component to which the structural feature recognition point or the temperature feature recognition point belongs, and extends 2 mm beyond the outer edge.

[0016] Preferably, in step S2, a detection period is preset according to the total working duration, and the unit number of revolutions is calibrated according to the rotational speed of the mine idler during the detection period, including:

[0017] Preset the total working duration of the mine idler this time, calibrate the unit time according to the total working duration, and when the total working duration is less than 24 hours, extract 10 minutes every 2 hours after exceeding the 12th hour as the unit time;

[0018] When the total working duration exceeds 24 hours, extract 20 minutes every 1 hour after exceeding the 24th hour as the unit time;

[0019] Install a rotational speed sensor on the mine idler to real-time monitor the rotational speed Δρ (unit: r / min) of the mine idler, compare the rotational speed Δρ r / min of the mine idler with a preset first rotational speed threshold ρ1 and a second rotational speed threshold ρ2, where ρ1 < ρ2, and set the unit number of revolutions according to the comparison result;

[0020] When Δρ < ρ1, preset the unit number of revolutions as the first unit number of revolutions Q1;

[0021] When ρ1 < Δρ ≤ ρ2, preset the unit number of revolutions as the second unit number of revolutions Q2;

[0022] When ρ2 < Δρ, preset the unit number of revolutions as the third unit number of revolutions Q3;

[0023] Wherein, Q1 > Q2 > Q3.

[0024] Preferably, in step S3, the preprocessing of the time-sequence image file for each detection period includes:

[0025] The preprocessing of the infrared image includes using a filter to reduce noise, and using a deconvolution algorithm, motion blur estimation and removal method to restore the clarity of the infrared image;

[0026] The preprocessing of the visible light image includes denoising, enhancement and grayscale conversion in sequence.

[0027] Preferably, in step S3, the first texture analysis of the spectrogram and extraction of the first texture features include:

[0028] Use the gray-level co-occurrence matrix algorithm to obtain the texture information of the spectrogram and generate the first texture features, where the first texture features include the contrast, energy and correlation calculated from the co-occurrence matrix algorithm;

[0029] Analyze the frequency distribution in the spectrogram, extract the edge information of the mine idler, perform a second texture analysis on the edge information, and extract second texture features, including:

[0030] Extract the high-frequency components in the spectrogram, obtain the contrast E and correlation in the first texture features of the high-frequency components, perform edge enhancement on the spectrogram according to the correlation by inputting the Laplacian operator, extract the edge information in the spectrogram, and stratify the edge information according to the contrast E;

[0031] Specifically, compare the contrast E with a preset standard contrast E1, and judge the level of the edge information according to the comparison result;

[0032] When E < E1, judge that the edge information belongs to the inner layer edge information T;

[0033] When E ≥ E1, judge that the edge information belongs to the outer layer edge information R;

[0034] Perform a second texture analysis on the outer layer edge information R using the local binary pattern to generate second texture features, and the second texture features include roughness, directivity, and repeatability calculated from the local binary pattern.

[0035] Preferably, in step S4, use cosine similarity to evaluate the similarity between the first texture feature and the second texture feature and their preset standard texture features, including:

[0036] Normalize the first texture feature and the second texture feature respectively to obtain a first texture vector A and a second texture vector B. A first standard texture feature vector T1 is preset for the first texture vector A, and a second standard texture feature vector T2 is preset for the second texture feature vector B;

[0037] The cosine similarity CS of the first texture vector A A Is calculated by the following formula:

[0038]

[0039] Where A·T1 is the dot product of A and T1, and ∥A∥×∥T1∥ is the norm of A and T1;

[0040] The cosine similarity CS of the second texture vector B B Is calculated by the following formula:

[0041]

[0042] Among them, B·T1 is the dot product of B and T2, and ∥B∥×∥T2∥ is the norm of B and T2.

[0043] Preferably, in step S4, when the similarity is lower than the standard similarity, it is determined as an abnormal texture feature, including:

[0044] The cosine similarity CS of the first texture vector A A is compared with a preset first standard similarity threshold SA1 and a second standard similarity threshold SA2, where SA1 < SA2, and the vibration level is determined according to the comparison result;

[0045] When CS A < SA1, it is determined as an abnormal texture feature and set to a third-level vibration P3;

[0046] When SA1 < CS A ≤ SA2, it is determined as a texture feature to be evaluated and set to a second-level vibration P2;

[0047] When SA2 < CS A , it is determined as a normal texture feature and set to a first-level vibration P1;

[0048] Among them, P1 < P2 < P3.

[0049] Preferably, in step S4, when the similarity is lower than the standard similarity, it is determined as an abnormal texture feature, and further includes:

[0050] The cosine similarity CS of the second texture vector B B is compared with a preset third standard similarity threshold SA3 and a fourth standard similarity threshold SA4, where SA3 < SA4, and the surface quality level is determined according to the comparison result;

[0051] When CS B < SA3, it is determined as an abnormal texture feature and set to a third-level surface quality N3;

[0052] When SA3 < CS B ≤ SA4, it is determined as a texture feature to be evaluated and set to a second-level surface quality N2;

[0053] When SA4 < CS B , it is determined as a normal texture feature and set to a first-level surface quality N1;

[0054] Among them, N1 < N2 < N3.

[0055] Preferably, in step S5, obtain the infrared image of the same frame associated with the abnormal state image, and perform temperature sensing analysis on the corresponding feature region of the abnormal feature region in the infrared image of the same frame to extract temperature sensing features, including:

[0056] Obtain the feature region to which the texture to be evaluated in the abnormal state image belongs, use a target detection algorithm to locate the abnormal feature region in the infrared image of the same frame, extract the abnormal feature region as the ROI region, and calculate the temperature of each pixel in the ROI region according to the relationship between the gray value of the pixel in the infrared image and the actual temperature;

[0057] Comprehensively analyze the operating state of the mining idler, including:

[0058] Extract the first high-temperature pixel points with temperature values higher than the preset component temperature alarm threshold after temperature calculation, and compare the first high-temperature pixel points with the occurrence area of the secondary vibration P2 of the texture feature to be evaluated to judge the operating state,

[0059] When it is judged that the third-level vibration P3 appears or the first high-temperature pixel point intersects with the occurrence of the secondary vibration P2, it is judged as an abnormal operating state, and an emergency stop is performed on the mining idler;

[0060] When the first high-temperature pixel point does not intersect with the secondary vibration P2, it is set as the monitoring operating state, a maintenance warning is issued, and the interval between each unit time is cancelled and changed to continuous monitoring;

[0061] Extract the second high-temperature pixel points with temperature values higher than the preset surface temperature alarm threshold after temperature calculation, and compare the second high-temperature pixel points with the occurrence area of the secondary surface quality N2 judged in the texture feature to be evaluated to judge the operating state;

[0062] When it is judged that the third-level surface quality N3 appears or when the second high-temperature pixel point intersects with the secondary surface quality N2, it is judged as an abnormal operating state, and an emergency stop is performed on the mining idler;

[0063] When the second high-temperature pixel point does not intersect with the secondary surface quality N2, it is set as the monitoring operating state, a maintenance warning is issued, and the interval between each unit time is cancelled and changed to continuous monitoring.

[0064] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0065] Combining the information of infrared images and visible light images, as well as various analysis methods such as frequency domain conversion and texture feature extraction, more comprehensive and multi-dimensional information can be obtained, which helps to comprehensively understand the state and operation of the mine idler, and reduces unnecessary computational workload through a reasonable detection cycle plan; by evaluating the similarity of the texture features to the standard similarity in the normal operating state, abnormal conditions can be detected and identified quickly and accurately, improving the efficiency and accuracy of fault diagnosis; by further combining infrared images for temperature sensing analysis, an analysis method with low computational workload based on data association is given, which helps to reduce the computing power cost and timely detect potential thermal problems or fault tendencies. By comprehensively analyzing the operating state of the mine idler, decision-making support can be provided for maintenance personnel to help them better manage equipment, formulate maintenance plans and optimize operation strategies.

[0066] On the other hand, the present application also provides an operating state detection system for a mine idler, which is applied to the above-mentioned operating state detection method of the mine idler and includes:

[0067] An acquisition module, which marks feature points on the physical object of the mine idler, obtains infrared images and visible light images of the mine idler at the same angle, stores the infrared images and the visible light images corresponding to the same time sequence one by one to generate a time sequence image file, and constructs a feature region for each feature point in each time sequence image file;

[0068] A calibration module, which records the time sequence image files per unit time. Before the start of each unit time, there is a preparation stage. In the preparation stage, the feature points of the mine idler are coincided with preset calibration points for static calibration, and the component information of the mine idler is corresponded based on the feature points and the feature regions; each unit time also includes a first moment. At the first moment, the mine idler rotates to the calibration point, and the infrared images and the visible light images within the unit time are associated frame by frame starting from the first moment. According to the total working duration, a detection cycle is preset, and the unit rotation speed is calibrated according to the rotation speed of the mine idler within the detection cycle;

[0069] An image analysis module, which preprocesses and analyzes the time sequence image files of each detection cycle. The image analysis includes performing frequency domain conversion on the preprocessed visible light image to obtain a frequency domain image, obtaining a spectrogram of the mine idler at each unit rotation speed according to the frequency domain image, performing first texture analysis on the spectrogram to extract first texture features, analyzing the frequency distribution in the spectrogram, extracting the edge information of the mine idler, and performing second texture analysis on the edge information to extract second texture features;

[0070] A comparison module uses cosine similarity to evaluate the similarity between the first texture feature and the second texture feature and their preset standard texture features respectively. When the similarity is lower than the standard similarity, it is determined as an abnormal texture feature, and the feature region to which the abnormal texture feature belongs is obtained, set as an abnormal feature region, and the visible light image with the abnormal texture feature is marked as an abnormal state image;

[0071] An operating state detection module obtains the infrared image of the same frame associated with the abnormal state image, extracts the corresponding feature region of the abnormal feature region in the infrared image of the same frame for temperature sensing analysis, extracts temperature sensing features, and comprehensively analyzes the operating state of the mining idler.

[0072] It can be understood that the operating state detection system of the mining idler provided in this application and its corresponding method have the same beneficial effects, which will not be elaborated here. Description of the Drawings

[0073] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0074] Figure 1 is a flowchart of the method for detecting the operating state of the mining idler provided by the embodiment of the present invention;

[0075] Figure 2 is a functional block diagram of the operating state detection system of the mining idler provided by the embodiment of the present invention. Detailed Embodiments

[0076] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be completely conveyed to those skilled in the art. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the drawings and in conjunction with the embodiments.

[0077] Refer to Figure 1 As shown, this embodiment provides a method for detecting the operating state of a mining idler, including:

[0078] Step S1: Mark feature points on the physical object of the mine idler, obtain infrared images and visible light images of the mine idler at the same angle, store the infrared images and visible light images corresponding to each other in the same time sequence to generate a time sequence image file, and construct a feature region for each feature point in each time sequence image file;

[0079] Step S2: Record the time sequence image files per unit time. Before the start of each unit time, there is a preparation stage. In the preparation stage, the feature points of the mine idler are coincided with the preset calibration points for static calibration, and the component information of the mine idler is corresponded based on the feature points and the feature regions; Each unit time also includes a first moment. At the first moment, the mine idler is rotated to the calibration point. Starting from the first moment, the infrared images and visible light images within the unit time are associated frame by frame, and the detection period is preset according to the total working duration. The unit revolution number is calibrated according to the rotation speed of the mine idler within the detection period;

[0080] Step S3: Preprocess and perform image analysis on the time sequence image files of each detection period. Among them, the image analysis includes performing frequency domain conversion on the preprocessed visible light image to obtain a frequency domain image, obtaining the spectrogram of the mine idler at each unit revolution number according to the frequency domain image, performing the first texture analysis on the spectrogram, extracting the first texture feature, and analyzing the frequency distribution in the spectrogram, extracting the edge information of the mine idler, and performing the second texture analysis on the edge information to extract the second texture feature;

[0081] Step S4: Use cosine similarity to evaluate the similarity between the first texture feature and the second texture feature and their preset standard texture features respectively. When the similarity is lower than the standard similarity, it is judged as an abnormal texture feature, and the feature region to which the abnormal texture feature belongs is obtained, set as the abnormal feature region, and the visible light image with the abnormal texture feature is marked as an abnormal state image;

[0082] Step S5: Obtain the infrared image associated with the abnormal state image, extract the feature region corresponding to the abnormal feature region in the associated infrared image for thermal sense analysis, extract the thermal sense feature, and comprehensively analyze the operating state of the mine idler.

[0083] It can be understood that in this embodiment, a detection scheme with low computing power cost based on data association is given, which avoids data registration of infrared images and visible light images to reduce the calculation amount, and achieves the technical effect of multi-dimensional detection, improves the reliability and safety of the equipment, reduces the maintenance cost, and prolongs the service life of the equipment, which has important significance for equipment management in industrial fields such as mines.

[0084] In some embodiments of the present application, in step S1, marking feature points on the physical object of the mine idler includes:

[0085] Obtain the geometric center point of each component of the mine idler as a feature point for marking as a structural feature recognition point. The structural feature recognition points are set with distinct features, including polygon edges and distinct textures. Among them, when the component directly cooperates with the mine conveyor drive component, a temperature isolation coating is used to re-mark the structural feature recognition points as temperature feature recognition points.

[0086] For each feature point in each sequential image file, a corresponding feature region is preset, including:

[0087] Each feature region covers each outer edge of the component to which the structural feature recognition point or the temperature feature recognition point belongs, and extends 2 mm beyond the outer edge.

[0088] Specifically, this calibration is directly related to the preset standard texture feature in subsequent step S4. Through reasonable calibration, the comparison and detection with the preset standard texture feature in subsequent step S4 can be made more accurate.

[0089] In some embodiments of the present application, in step S2, when presetting the detection period according to the total working duration and calibrating the unit revolution number according to the rotation speed of the mine idler within the detection period, it includes:

[0090] Preset the total working duration of the mine idler in advance, calibrate the unit time according to the total working duration. When the total working duration is less than 24 hours, starting from the 12th hour, 10 minutes are extracted every 2 hours as the unit time;

[0091] When the total working duration exceeds 24 hours, starting from the 24th hour, 20 minutes are extracted every 1 hour as the unit time;

[0092] Install a rotation speed sensor on the mine idler to monitor the rotation speed Δρ (unit: r / min) of the mine idler in real time. Compare the rotation speed Δρ r / min of the mine idler with the preset first rotation speed threshold ρ1 and second rotation speed threshold ρ2, where ρ1 < ρ2, and set the unit revolution number according to the comparison result;

[0093] When Δρ < ρ1, preset the unit revolution number as the first unit revolution number Q1;

[0094] When ρ1 < Δρ ≤ ρ2, preset the unit revolution number as the second unit revolution number Q2;

[0095] When ρ2 < Δρ, preset the unit revolution number as the third unit revolution number Q3;

[0096] Among them, Q1 > Q2 > Q3.

[0097] Specifically, when the rotation speed of the idler roller is higher at Δρ r / min, the number of detected revolutions per unit is less, and the number of calculation times is increased to achieve a more accurate detection effect. When the rotation speed of the idler roller is lower at Δρ r / min, the number of detected revolutions per unit is more, and the number of calculation times is reduced to save computing power.

[0098] In some embodiments of the present application, in step S3, preprocessing is performed on the sequential image files of each detection period, including:

[0099] The preprocessing of the infrared image includes using a filter to reduce noise, and using a deconvolution algorithm, motion blur estimation, and removal method to restore the clarity of the infrared image;

[0100] The preprocessing of the visible light image includes denoising, enhancement, and grayscale conversion in sequence.

[0101] In some embodiments of the present application, in step S3, the first texture analysis is performed on the spectrogram to extract the first texture features, including:

[0102] The gray-level co-occurrence matrix algorithm is used to obtain the texture information of the spectrogram and generate the first texture features, where the first texture features include contrast, energy, and correlation calculated from the co-occurrence matrix algorithm;

[0103] Analyze the frequency distribution in the spectrogram, extract the edge information of the mining idler roller, and perform a second texture analysis on the edge information to extract the second texture features, including:

[0104] Extract the high-frequency components in the spectrogram, obtain the contrast E and correlation in the first texture features of the high-frequency components, perform edge enhancement on the spectrogram by inputting the Laplacian operator according to the correlation, extract the edge information in the spectrogram, and stratify the edge information according to the contrast E;

[0105] Specifically, the contrast E is compared with a preset standard contrast E1, and the level of the edge information is judged according to the comparison result;

[0106] When E < E1, it is judged that the edge information belongs to the inner layer edge information T;

[0107] When E ≥ E1, it is judged that the edge information belongs to the outer layer edge information R;

[0108] Perform a second texture analysis on the outer layer edge information R using the local binary pattern to generate the second texture features, and the second texture features include roughness, directivity, and repeatability calculated from the local binary pattern.

[0109] Specifically, the Laplacian operator is a first-order differential operator used in image processing and computer vision, and is usually used to detect details such as edges and textures in images. The Laplacian operator is widely used in image processing, especially in edge detection and image enhancement; while the Local Binary Pattern (LBP) is a feature extraction method for texture analysis, and is usually used to describe the patterns and structures of local textures in images. LBP generates a binary pattern to represent the texture features of each pixel point by comparing the gray values of the neighborhood around the pixel point in the image. This method is very effective in dealing with tasks such as texture recognition and classification.

[0110] It can be understood that the Laplacian operator and LBP can be applied in spectrogram analysis. Although LBP was originally designed to process grayscale images, it can be extended to spectrograms. In a spectrogram, LBP can be applied to describe the texture features of different local regions in the spectrogram, thereby helping to identify the patterns and structures in the spectrogram.

[0111] In some embodiments of the present application, in step S4, the use of cosine similarity to evaluate the similarity of the first texture feature and the second texture feature with their preset standard texture features includes:

[0112] Normalize the first texture feature and the second texture feature respectively to obtain the first texture vector A and the second texture vector B. A first standard texture feature vector T1 is preset for the first texture vector A, and a second standard texture feature vector T2 is preset for the second texture feature vector B;

[0113] The cosine similarity CS of the first texture vector A A is calculated by the following formula:

[0114]

[0115] where A·T1 is the dot product of A and T1, and ∥A∥×∥T1∥ is the norm of A and T1;

[0116] The cosine similarity CS of the second texture vector B B is calculated by the following formula:

[0117]

[0118] where B·T2 is the dot product of B and T2, and ∥B∥×∥T2∥ is the norm of B and T2.

[0119] In some embodiments of the present application, in step S4, when the similarity is lower than the standard similarity, it is determined as an abnormal texture feature, including:

[0120] Compare the cosine similarity CS of the first texture vector A A with a preset first standard similarity threshold SA1 and a second standard similarity threshold SA2, where SA1 < SA2, and judge the vibration level according to the comparison result;

[0121] When CS A < SA1, judge it as an abnormal texture feature and set it to the third-level vibration P3;

[0122] When SA1 < CS A ≤ SA2, judge it as a texture feature to be evaluated and set it to the second-level vibration P2;

[0123] When SA2 < CS A , judge it as a normal texture feature and set it to the first-level vibration P1;

[0124] where, P1 < P2 < P3.

[0125] In some embodiments of the present application, in step S4, when the similarity is lower than the standard similarity, it is judged as an abnormal texture feature, and it further includes:

[0126] Compare the cosine similarity CS of the second texture vector B B with a preset third standard similarity threshold SA3 and a fourth standard similarity threshold SA4, where SA3 < SA4, and judge the surface quality level according to the comparison result;

[0127] When CS B < SA3, judge it as an abnormal texture feature and set it to the third-level surface quality N3;

[0128] When SA3 < CS B ≤ SA4, judge it as a texture feature to be evaluated and set it to the second-level surface quality N2;

[0129] When SA4 < CS B , judge it as a normal texture feature and set it to the first-level surface quality N1;

[0130] where, N1 < N2 < N3.

[0131] Specifically, the first texture feature reflects the state of the component, and evaluating it can judge the vibration condition of the component; the second texture feature reflects the microscopic structure of the surface, and evaluating it can judge the surface quality of the component, and the higher the value of the cosine similarity CS A or the cosine similarity CS B , the smaller the difference from the preset standard, that is, the smaller the vibration state or the better the surface quality.

[0132] In some embodiments of the present application, in step S5, obtaining the infrared image of the same frame associated with the abnormal state image, and performing temperature sensing analysis on the feature region corresponding to the abnormal feature region in the infrared image of the same frame to extract temperature sensing features, including:

[0133] Obtain the feature region to which the texture to be evaluated in the abnormal state image belongs, use the target detection algorithm to locate the abnormal feature region in the infrared image of the same frame, extract the abnormal feature region as the ROI region (Region of Interest), and calculate the temperature of each pixel in the ROI region according to the relationship between the gray value of the pixel in the infrared image and the actual temperature;

[0134] Specifically, the Region of Interest (ROI) refers to the region in the image or video that is defined by the user or algorithm as having special significance or requiring special processing. The ROI region is an important concept used in image processing, computer vision, and machine learning tasks to focus on the region of interest for more effective analysis, processing, or feature extraction.

[0135] The comprehensive analysis of the operating state of the mine idler includes:

[0136] Extract the first high-temperature pixel points whose temperature values are higher than the preset component temperature alarm threshold after temperature calculation, and compare the first high-temperature pixel points with the occurrence region of the secondary vibration P2 of the texture feature to be evaluated to judge the operating state,

[0137] When it is judged that the third-level vibration P3 appears or when the first high-temperature pixel point intersects with the secondary vibration P2, it is judged as an abnormal operating state, and the mine idler is stopped immediately;

[0138] When the first high-temperature pixel point does not intersect with the secondary vibration P2, it is set as the monitoring operating state, a maintenance warning is issued, and the interval between each unit time is cancelled and changed to continuous monitoring;

[0139] Extract the second high-temperature pixel points whose temperature values are higher than the preset surface temperature alarm threshold after temperature calculation, and compare the second high-temperature pixel points with the occurrence region of the secondary surface quality N2 judged in the texture feature to be evaluated to judge the operating state;

[0140] When it is judged that the third-level surface quality N3 appears or when the second high-temperature pixel point intersects with the secondary surface quality N2, it is judged as an abnormal operating state, and the mine idler is stopped immediately;

[0141] When the second high-temperature pixel point does not intersect with the secondary surface quality N2, it is set as the monitoring operating state, a maintenance warning is issued, and the interval between each unit time is cancelled and changed to continuous monitoring.

[0142] In summary, compared with the prior art, the beneficial effects of this embodiment are as follows:

[0143] By combining the information of infrared images and visible light images, as well as various analysis methods such as frequency domain conversion and texture feature extraction, more comprehensive and multi-dimensional information can be obtained, which helps to comprehensively understand the state and operation of the mine idler, and reduces unnecessary computational complexity through a reasonable detection cycle plan; by evaluating the similarity of the texture features to the standard similarity in the normal operating state, abnormal conditions can be quickly and accurately detected and identified, improving the efficiency and accuracy of fault diagnosis; by further combining infrared images for temperature sensing analysis, an analysis method with low computational complexity based on data association is given, which helps to reduce the computing power cost and timely detect potential thermal problems or fault tendencies. By comprehensively analyzing the operating state of the mine idler, decision-making support can be provided for maintenance personnel to help them better manage equipment, formulate maintenance plans, and optimize operation strategies.

[0144] Refer to Figure 2 As shown, this embodiment also provides a system for detecting the operating state of a mine idler, which is applied to the above method for detecting the operating state of a mine idler, and includes:

[0145] An acquisition module marks feature points on the physical object of the mine idler, obtains infrared images and visible light images of the mine idler at the same angle, stores the infrared images and visible light images corresponding to the same time sequence one by one to generate a time sequence image file, and constructs a feature region for each feature point in each time sequence image file;

[0146] A calibration module records the time sequence image files per unit time. Before the start of each unit time, there is a preparation stage. In the preparation stage, the feature points of the mine idler are aligned with the preset calibration points for static calibration, and the component information of the mine idler is based on the feature points and feature regions; Each unit time also includes a first moment. At the first moment, the mine idler rotates to the calibration point, and the infrared images and visible light images within the unit time are associated frame by frame starting from the first moment. According to the total working duration, the detection period is preset, and the unit rotation speed is calibrated according to the rotation speed of the mine idler within the detection period;

[0147] An image analysis module performs preprocessing and image analysis on the time sequence image files of each detection period. Among them, the image analysis includes performing frequency domain conversion on the preprocessed visible light image to obtain a frequency domain image, obtaining the spectrogram of the mine idler at each unit rotation speed according to the frequency domain image, performing first texture analysis on the spectrogram, extracting the first texture feature, analyzing the frequency distribution in the spectrogram, extracting the edge information of the mine idler, and performing second texture analysis on the edge information to extract the second texture feature;

[0148] The comparison module uses cosine similarity to evaluate the similarities between the first texture feature and the second texture feature and their preset standard texture features respectively. When the similarity is lower than the standard similarity, it is determined as an abnormal texture feature, and the feature region to which the abnormal texture feature belongs is obtained, set as the abnormal feature region, and the visible light image with the abnormal texture feature is marked as an abnormal state image;

[0149] The operating state detection module obtains the infrared image of the same frame associated with the abnormal state image, performs thermal sense analysis on the corresponding feature region of the abnormal feature region in the infrared image of the same frame, extracts thermal sense features, and comprehensively analyzes the operating state of the mine idler.

[0150] It can be understood that the operating state detection system of the mine idler provided in this embodiment has the same beneficial effects as the corresponding method, which will not be elaborated here.

[0151] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0152] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of processes and / or blocks in the flowcharts and / or block diagrams can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0153] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device implements the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0154] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions for implementing the functions specified in one process or a plurality of processes and / or blocks Figure 1 one process or a plurality of processes and / or blocks Figure 1 steps of the functions specified in one block or a plurality of blocks.

[0155] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific embodiments of the present invention. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A method for detecting the operating state of a mine idler, characterized in that, Including: Step S1: Mark feature points on the physical object of the mining idler, obtain the infrared image and visible light image of the mining idler at the same angle, and store the infrared image and the visible light image of the same time sequence in one-to-one correspondence to generate a time sequence image file; And construct a feature region for each of the feature points in each of the time sequence image files; Step S2: Record the time sequence image files per unit time. Before the start of each unit time, there is a preparation stage. In the preparation stage, the feature points of the mining idler are coincided with the preset calibration points for static calibration, and the component information of the mining idler is corresponded based on the feature points and the feature regions; The unit time also includes a first moment. At the first moment, the mining idler rotates to the calibration point. Starting from the first moment, the infrared images and the visible light images within the unit time are associated frame by frame. And according to the total working duration, a detection period is preset. According to the rotation speed of the mining idler within the detection period, the unit rotation number is calibrated; Step S3: Preprocess and perform image analysis on the time sequence image files of each detection period. Among them, the image analysis includes performing frequency domain conversion on the preprocessed visible light image to obtain a frequency domain image, obtaining the spectrogram of the mining idler at each unit rotation number according to the frequency domain image, performing first texture analysis on the spectrogram, extracting the first texture feature, and analyzing the frequency distribution in the spectrogram, extracting the edge information of the mining idler, and performing second texture analysis on the edge information to extract the second texture feature; Step S4: Use cosine similarity to evaluate the similarity between the first texture feature and the second texture feature and their preset standard texture features respectively. When the similarity is lower than the standard similarity, it is judged as an abnormal texture feature, and the feature region to which the abnormal texture feature belongs is obtained, set as an abnormal feature region, and the visible light image with the abnormal texture feature is marked as an abnormal state image; Step S5: Obtain the infrared image of the same frame associated with the abnormal state image, extract the feature region corresponding to the abnormal feature region in the infrared image of the same frame for temperature sensing analysis, extract the temperature sensing feature, and comprehensively analyze the operating state of the mining idler.

2. The method for detecting the operating state of the mine idler according to claim 1, wherein In the step S1, the marking of the feature points on the physical object of the mining idler includes: Obtain the geometric center point of each component of the mining idler as the feature point for marking as a structural feature recognition point. The structural feature recognition point is provided with obvious different features, including polygon edges and different textures; Among them, when the component directly cooperates with the mining conveyor drive component, use a temperature isolation coating to re-mark the structural feature recognition point as a temperature feature recognition point; The constructing of a feature region for each of the feature points in each of the time sequence image files includes: Each of the feature regions covers each outer edge of the component to which the structural feature recognition point or the temperature feature recognition point belongs, and exceeds the outer edge by 2 mm.

3. The method for detecting the operating state of the mine idler according to claim 2, characterized in that, In step S2, presetting the detection period according to the total working duration and calibrating the unit rotation speed based on the rotation speed of the mine idler within the detection period includes: Presetting the total working duration of the mine idler this time and calibrating the unit time according to the total working duration; When the total working duration is less than 24 hours, extract 10 minutes every 2 hours after the 12th hour as the unit time; When the total working duration exceeds 24 hours, extract 20 minutes every 1 hour after the 24th hour as the unit time; Install a rotation speed sensor on the mine idler to monitor the rotation speed Δρ of the mine idler in real time, compare the rotation speed Δρ of the mine idler with a preset first rotation speed threshold ρ1 and a second rotation speed threshold ρ2, where ρ1 < ρ2, and set the unit rotation speed according to the comparison result; When Δρ < ρ1, preset the unit rotation speed as the first unit rotation speed Q1; When ρ1 < Δρ ≤ ρ2, preset the unit rotation speed as the second unit rotation speed Q2; When ρ2 < Δρ, preset the unit rotation speed as the third unit rotation speed Q3; Wherein, Q1 > Q2 > Q3.

4. The method for detecting the operating state of the mine idler according to claim 3, characterized in that, In step S3, preprocessing the sequential image file of each detection period includes: The preprocessing of the infrared image includes using a filter to reduce noise, and using a deconvolution algorithm, motion blur estimation and removal method to restore the clarity of the infrared image; The preprocessing of the visible light image includes denoising, enhancement and grayscale conversion in sequence.

5. The method for detecting the operating state of the mine idler according to claim 4, wherein, In step S3, performing a first texture analysis on the spectrogram and extracting first texture features includes: Using the gray-level co-occurrence matrix algorithm to obtain the texture information of the spectrogram and generate first texture features, where the first texture features include contrast, energy and correlation calculated from the co-occurrence matrix algorithm; Analyzing the frequency distribution in the spectrogram, extracting the edge information of the mine idler, and performing a second texture analysis on the edge information to extract second texture features includes: Extracting the high-frequency components in the spectrogram, obtaining the contrast E and correlation in the first texture features of the high-frequency components, performing edge enhancement on the spectrogram by inputting the correlation into the Laplacian operator to detect the edge information in the spectrogram, and extracting the edge information in the spectrogram, and stratifying the edge information according to the contrast E. Specifically, compare the contrast E with a preset standard contrast E1, and judge the level of the edge information according to the comparison result; When E < E1, judge that the edge information belongs to the inner layer edge information T; When E ≥ E1, judge that the edge information belongs to the outer layer edge information R; Performing a second texture analysis on the outer layer edge information R using the local binary pattern to generate second texture features, where the second texture features include roughness, directionality and repeatability calculated from the local binary pattern.

6. The method for detecting the operating state of the mine idler according to claim 1, wherein, In step S4, using the cosine similarity to evaluate the similarity between the first texture features and the second texture features and their preset standard texture features includes: Normalize the first texture feature and the second texture feature respectively to obtain a first texture vector A and a second texture vector B. A first standard texture feature vector T1 is preset for the first texture vector A, and a second standard texture feature vector T2 is preset for the second texture feature vector B; The cosine similarity CS of the first texture vector A A is calculated by the following formula: Wherein, A·T1 is the dot product of A and T1, and ∥A∥×∥T1∥ is the norm of A and T1; The cosine similarity CS of the second texture vector B B is obtained by calculating with the following formula: Wherein, B·T1 is the dot product of B and T2, and ∥B∥×∥T2∥ is the norm of B and T2.

7. The method for detecting the operating state of a mine idler according to claim 6, characterized in that In step S4, when the similarity is lower than the standard similarity, it is determined as an abnormal texture feature, including: Compare the cosine similarity CS of the first texture vector A A with a preset first standard similarity threshold SA1 and a second standard similarity threshold SA2, where SA1 < SA2, and determine the vibration level according to the comparison result; When CS A <SA1, it is determined as an abnormal texture feature and set as the third-level vibration P3; When SA1 < CS A ≤ SA2, it is determined as a texture feature to be evaluated and set as the secondary vibration P2; When SA2 < CS A it is determined as a normal texture feature and set as the first-level vibration P1; Wherein, P1 < P2 < P3.

8. The method for detecting the operating state of a mine idler according to claim 7, characterized in that, In step S4, when the similarity is lower than the standard similarity, it is determined as an abnormal texture feature, and further includes: Compare the cosine similarity CS of the second texture vector B B with a preset third standard similarity threshold SA3 and a fourth standard similarity threshold SA4, where SA3 < SA4, and determine the surface quality level according to the comparison result; When CS B When < SA3, it is determined to be an abnormal texture feature and set to the third-level surface quality N3; When SA3 < CS B ≤ SA4, it is determined as a texture feature to be evaluated and set as the secondary surface quality N2; When SA4 < CS B it is determined as a normal texture feature and set as the first-level surface quality N1; Wherein, N1 < N2 < N3.

9. The method for detecting the operating state of the mine idler according to claim 8, characterized in that, In step S5, obtaining the infrared image of the same frame associated with the abnormal state image, and performing a temperature sensing analysis on the corresponding feature region of the abnormal feature region in the infrared image of the same frame to extract temperature sensing features, including: Obtain the feature region to which the texture to be evaluated in the abnormal state image belongs, use an object detection algorithm to locate the abnormal feature region in the infrared image of the same frame, extract the abnormal feature region as the ROI region, and calculate the temperature of each pixel in the ROI region according to the relationship between the gray value of the pixel in the infrared image and the actual temperature; The comprehensive analysis of the operating state of the mining idler includes: Extract the first high-temperature pixel points with temperature values higher than the preset component temperature alarm threshold after temperature calculation, and compare the first high-temperature pixel points with the occurrence area of the secondary vibration P2 of the texture feature to be evaluated to judge the operating state, When it is judged that the third-level vibration P3 appears or when the first high-temperature pixel point intersects with the occurrence of the secondary vibration P2, it is judged as an abnormal operating state, and an emergency stop is performed on the mining idler; When the first high-temperature pixel point does not intersect with the secondary vibration P2, it is set as the monitoring operating state, a maintenance warning is issued, and the interval between each unit time is cancelled and changed to continuous monitoring; Extract the second high-temperature pixel points with temperature values higher than the preset surface temperature alarm threshold after temperature calculation, and compare the second high-temperature pixel points with the occurrence area of the secondary surface quality N2 judged in the texture feature to be evaluated to judge the operating state; When it is judged that the third-level surface quality N3 appears or when the second high-temperature pixel point intersects with the occurrence of the secondary surface quality N2, it is judged as an abnormal operating state, and an emergency stop is performed on the mining idler; When the second high-temperature pixel point does not intersect with the secondary surface quality N2, it is set as the monitoring operating state, a maintenance warning is issued, and the interval between each unit time is cancelled and changed to continuous monitoring.

10. An operating state detection system for a mine idler, which is applied to the operating state detection method of the mine idler according to any one of claims 1-9, is characterized in that Including: The acquisition module marks feature points on the physical object of the mine idler, obtains the infrared image and visible light image of the mine idler at the same angle, stores the infrared image and the visible light image corresponding to the same time sequence one by one, generates a time sequence image file, and constructs a feature region for each feature point in each time sequence image file; The calibration module records the time sequence image files per unit time. Before the start of each unit time, there is a preparation stage. In the preparation stage, the feature points of the mine idler are coincided with the preset calibration points for static calibration, and the component information of the mine idler is corresponded based on the feature points and the feature regions; The unit time also includes a first moment. At the first moment, the mine idler rotates to the calibration point. Starting from the first moment, the infrared images and the visible light images within the unit time are associated frame by frame, and a detection period is preset according to the total working duration. The unit rotation speed of the mine idler is calibrated according to the rotation speed of the mine idler within the detection period; The image analysis module preprocesses and analyzes the time sequence image files of each detection period. Among them, the image analysis includes performing frequency domain conversion on the preprocessed visible light image to obtain a frequency domain image, obtaining the spectrogram of the mine idler at each unit rotation speed according to the frequency domain image, performing first texture analysis on the spectrogram, extracting first texture features, analyzing the frequency distribution in the spectrogram, extracting the edge information of the mine idler, and performing second texture analysis on the edge information to extract second texture features; The comparison module uses cosine similarity to evaluate the similarity of the first texture feature and the second texture feature with their preset standard texture features respectively. When the similarity is lower than the standard similarity, it is judged as an abnormal texture feature, and the feature region to which the abnormal texture feature belongs is obtained, set as an abnormal feature region, and the visible light image with the abnormal texture feature is marked as an abnormal state image; The operating state detection module obtains the infrared image of the same frame associated with the abnormal state image, extracts the feature region corresponding to the abnormal feature region in the infrared image of the same frame for thermal sense analysis, extracts thermal sense features, and comprehensively analyzes the operating state of the mine idler.

Images