Belt conveyor detection system and control method of a belt conveyor detection system

Abstract

The application provides a belt conveyor detection system and a control method of the belt conveyor detection system, and relates to the technical field of conveyor belt detection. The belt conveyor detection system comprises a belt conveyor and a host computer. The belt conveyor comprises a fixing frame and an information acquisition module. The fixing frame is located at the middle position of the upper surface of the belt conveyor. The information acquisition module is located on the fixing frame. The information acquisition module is used for acquiring sensing data of the belt conveyor. The sensing data comprises infrared images and sound signals. The host computer is in communication connection with the belt conveyor. The host computer is used for determining whether the belt conveyor is faulty according to the sensing data. The scheme solves the problems of high cost, poor real-time performance and low accuracy of the traditional manual inspection method for the conveyor belt in actual use.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of belt detection, in particular to a belt conveyor detection system and a control method of the belt conveyor detection system. BACKGROUND

[0002] In the production process of coal mines, the coal mined from the coal mining face is mainly transported to the coal bunker by the belt conveyor to realize efficient transportation of the coal. The working environment of the belt conveyor used in coal mines is harsh, complex and dangerous. The working flow of the entire device is large. When a fault occurs, the maintenance workload is large. Linkage adjustment is time-consuming and laborious. Therefore, when the belt conveyor is used, real-time detection and maintenance of the entire device are required.

[0003] The existing belt conveyor is usually manually inspected when in use because the belt conveyor is composed of multiple sections of conveying belts connected by multiple transfer interfaces. However, traditional manual inspection has high cost, poor real-time performance, and low accuracy. Therefore, when in use, it is unstable and has poor detection effect, which can easily lead to remaining safety hazards. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a belt conveyor detection system and a control method of the belt conveyor detection system, which solves the problem of high cost, poor real-time performance, and low accuracy of the traditional manual inspection method for the conveyor belt in actual use.

[0005] To achieve the above purpose, the present application realizes the following technical scheme: a belt conveyor detection system, comprising: a belt conveyor, the belt conveyor comprising a fixed frame and an information acquisition module, wherein the fixed frame is located at the middle position of the upper surface of the belt conveyor, the information acquisition module is located on the fixed frame, the information acquisition module is used to acquire sensing data of the belt conveyor, and the sensing data comprises infrared images and sound signals; an upper computer, in communication connection with the belt conveyor, the upper computer is used to determine whether the belt conveyor is faulty according to the sensing data.

[0006] Optionally, the information acquisition module comprises: an infrared thermal imager located at the middle position of the bottom surface of the fixed frame, the infrared thermal imager is used to acquire the infrared images of the belt conveyor, wherein the middle position is the same distance from the left end point of the fixed frame to the infrared thermal imager and from the right end point of the fixed frame to the infrared thermal imager; a right sound sensor located at one end of the right side of the infrared thermal imager on the bottom surface of the fixed frame; the right sound sensor is used to acquire the first sound signal of the belt conveyor; a left sound sensor located at one end of the left side of the infrared thermal imager on the bottom surface of the fixed frame, the left sound sensor is used to acquire the second sound signal of the belt conveyor.

[0007] Optionally, the belt conveyor detection system comprises a driving motor located at one side of the front face of the belt conveyor, and the driving motor is used to drive the belt conveyor to rotate.

[0008] Optionally, the belt conveyor further comprises a positioning module used to determine the rotation angle of the driving motor, and the positioning module comprises an encoder electrically connected with the driving motor, and the encoder is used to control the rotation of the driving motor.

[0009] Optionally, the belt conveyor detection system comprises a main control module and a data transmitter, the main control module comprises a main control board in communication connection with the belt conveyor, the main control board is used to control the whole system, and a wireless transmission module in communication connection with the main control board is used to transmit data to the data transmitter, and the data transmitter transmits data to the upper computer.

[0010] The application discloses a control method of any one of the belt conveyor detection systems, and the belt conveyor detection system further comprises a master control module and a data transmitter, the master control module comprises a master control board and a wireless transmission module, the master control board is in communication connection with the belt conveyor, the master control board is used for controlling the whole system, the wireless transmission module is in communication connection with the master control board, the wireless transmission module is used for transmitting data to the data transmitter, and the data transmitter transmits data to the upper computer, and the method specifically comprises the following steps: S1, a data information acquisition step, the sensing data of the belt conveyor in different operating states is acquired, and the sensing data comprises the infrared image and the sound signal; S2, a data transmission step, the LabVIEW software is started through the upper computer, the sensing data is transmitted through the data transmitter, the sensing data is recorded and stored through the master control board, the MATLABScript script is called through the LabVIEW software, the sensing data collected and stored is processed through the MATLABScript script, important information in the infrared image needs to be highlighted, and unnecessary information needs to be weakened or removed; S3, a data level processing step, the sound signal is used for filtering processing of the infrared image, and a significant region of the infrared image is segmented and feature extracted, and the processed infrared image is obtained, wherein the significant region is a region with a gray value greater than a preset value; S4, a feature level processing step, a first detection model is constructed, wherein the first detection model is obtained through the VNWOA algorithm and the RB algorithm by using a plurality of groups of training data, each group of training data in the plurality of groups of training data comprises a historical infrared image acquired in a historical time period and a first historical detection result; the processed infrared image is input into the first detection model, and a first detection result corresponding to the infrared image is obtained; a second detection model is constructed, wherein the second detection model is obtained through the VNWOA algorithm and the RB algorithm by using a plurality of groups of training data, each group of training data in the plurality of groups of training data comprises a historical sound signal acquired in a historical time period and a second historical detection result; the sound signal is input into the second detection model, and a second detection result corresponding to the sound signal is obtained; S5, a decision level processing step, the first detection result and the second detection result are fused for diagnosis, the detection results are sorted, the sorting serial numbers are averaged, the average sorting serial numbers are normalized, and a target detection result of the belt conveyor is determined; and the target detection result is output, and images, waveforms and fault information are displayed.

[0011] Optionally, in the S3 step, the filtering processing comprises the following steps: an A1 step of determining a maximum size of an expanded window for the infrared image according to the outermost boundary value to ensure that the infrared image can be filled with the window and guarantee the effect of subsequent processing; an A2 step of sequentially traversing points in the infrared image with an initial filtering window of 3x3, taking a pixel value at a center point position of a current window as a center point of an adaptive filtering window, and finding a maximum gray value and a minimum gray value in a filtering window neighborhood of X; an A3 step of performing primary noise judgment, and if the noise is less than or equal to the minimum gray value or the noise is greater than or equal to the maximum gray value, proceeding to secondary noise detection in A4, otherwise, continuing to execute step A3 and increasing the window by a preset step size until W>N, and outputting an original gray value for facilitating subsequent processing; an A4 step of performing secondary noise detection, and when the noise is less than a first gray value or the noise is greater than a second gray value, outputting an improved mean filter value item after filtering, otherwise, increasing the window size, and if the noise is less than the first gray value or the noise is greater than the second gray value, executing step A3 until the window size is Ng, where Ng is a number of pixels of the gray level g in the infrared image, and N is a total number of pixels in the infrared image, and if the value is greater than Ng, a noiseless point is detected and the original gray value in A3 is output; an A5 step of executing steps A2 to A4 to filter all points contained in the infrared image by using the filter, removing the expanded boundary in A1, and processing the infrared image until the processed infrared image.

[0012] Optionally, in the S3 step, a Canny algorithm is used to perform edge extraction on the image segmented based on region saliency to obtain the segmented infrared image.

[0013] Optionally, in the S3 step, the salient region segmentation specifically comprises the following steps: a B1 step of calculating a standard variance of each pixel in the infrared image in an NXN neighborhood window and establishing a variance texture map, where N≥1; a B2 step of taking points with a brightness value greater than a first preset brightness value as boundary points and taking points with a brightness value less than a second preset brightness value as non-boundary points in the variance texture map, where the first preset brightness value is greater than the second preset brightness value; and a B3 step of taking the boundary points in the variance texture map as feature points, performing screening and clustering to obtain a center of a salient region, and obtaining the salient region of the infrared image.

[0014] Optionally, in the S3 step, the feature extraction specifically comprises the following steps: C1, performing target segmentation on the filtered infrared image by using a salient region segmentation method; C2, performing Gaussian low-pass filtering and local entropy texture segmentation on the color image to remove the texture area, obtaining filtered gray images of R, G and B components, clustering and segmenting to automatically divide the three areas of the brightest area, the darkest area and the remaining area of each component, selecting the area with the largest brightness difference between the brightest or darkest area and the remaining area of each color component, detecting the corner points and edge points of the selected area boundary, taking these feature points as salient points, filling and expanding to close the area, extracting the largest connected area, and finally obtaining the salient region to segment the fault area of the belt conveyor; C3, performing edge extraction on the segmented fault area by using a Canny algorithm to obtain the shape feature of the fault area.

[0015] The present application provides an infrared detection device.

[0016] 1. The present application provides a belt conveyor detection system, compared with the existing detection device, according to the characteristics of high temperature heating during the generation process of the belt conveyor fault, by using the principle that infrared image can convert the object temperature that cannot be perceived by human eyes into image, selecting infrared image as the detection method of belt conveyor fault, collecting the image information of the conveyor belt by infrared thermal imager, and extracting features and denoising through sound signals, image segmentation is also performed before feature extraction, which can improve the efficiency and accuracy of feature extraction, and the shape of the part prone to failure of the belt conveyor is not the same, which can well realize the feature extraction of the image and provide detection feature vector for the fault diagnosis of the belt conveyor.

[0017] 2. The present application provides a control method of a belt conveyor detection system, compared with the existing detection device, due to the inevitable noise pollution in the transportation process of infrared image acquisition, which will affect the fault analysis result in the later stage, an improved mean adaptive median filter (IMAMF) is selected, the algorithm detects noise for each pixel point on the noisy image using an adaptive filtering window, expands the image by copying the pixel value of the image boundary, without changing the pixel value of the original image on the basis of noise detection and filtering of all pixel points, the improved mean filter replaces the original output, the linear filter improved from the traditional mean filter not only maintains the clarity of the image, but also is more effective for high concentration of noise. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 The present application provides a belt conveyor detection system axis measurement schematic diagram;

[0019] Figure 2 It is a hardware structure schematic diagram of the belt conveyor detection system of the application;

[0020] Figure 3 It is a data acquisition and processing flow schematic diagram of the application;

[0021] Figure 4 It is an infrared image noise reduction algorithm flow schematic diagram of the application;

[0022] Figure 5 It is an improved adaptive algorithm flow schematic diagram of the application.

[0023] Among them, 1, the belt conveyor; 2, the fixed frame; 3, the information acquisition module; 4, the host computer; 5, the infrared thermal imager; 6, the right sound sensor; 7, the left sound sensor; 8, the driving motor; 9, the positioning module; 10, the encoder; 11, the main control module; 12, the data transmitter; 13, the main control board; 14, the wireless transmission module. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the application.

[0025] Embodiment:

[0026] As shown in Figure 1 and Figure 2 The embodiment of the application provides a belt conveyor detection system, which comprises a belt conveyor 1 and a host computer 4. The belt conveyor comprises a fixed frame 2 and an information acquisition module 3. The fixed frame is located at the middle position of the upper surface of the belt conveyor. The information acquisition module is located on the fixed frame. The information acquisition module is used for acquiring sensing data of the belt conveyor, and the sensing data comprises infrared images and sound signals. The host computer is in communication connection with the belt conveyor, and is used for determining whether the belt conveyor is faulty according to the sensing data.

[0027] In the scheme, the sensing data can be used to determine whether the belt conveyor is faulty for the fault of the belt conveyor that cannot be perceived by the human eye. The belt conveyor fault generates high temperature, which cannot be perceived by the human eye sometimes. Therefore, the infrared image can be used to detect whether the belt conveyor is faulty. Of course, for some vibration faults of the belt conveyor, the sound signal can also be used to detect whether the belt conveyor is faulty.

[0028] Specifically, compared with the existing detection device, the system generates high temperature according to the characteristics of the needle belt type belt conveyor fault process, selects infrared image as the detection method of the belt conveyor fault through the principle that the human eye cannot perceive the object temperature, collects the image information of the conveyor belt through the infrared thermal imager, and extracts the features and denoises through the sound signal, and the image segmentation is also performed before the feature extraction, which can improve the efficiency and accuracy of the feature extraction, and the shape of the part prone to failure of the belt conveyor is not the same, which can well realize the feature extraction of the image and provide the detection feature vector for the fault diagnosis of the belt conveyor.

[0029] The above information acquisition module includes an infrared thermal imager 5, a right sound sensor 6 and a left sound sensor 7, the infrared thermal imager is located at the middle position of the bottom surface of the fixed frame, the infrared thermal imager is used to collect the infrared image of the belt conveyor, wherein the middle position is the same distance between the infrared thermal imager and the left end point of the fixed frame and the distance between the infrared thermal imager and the right end point of the fixed frame; the right sound sensor is located at one end of the right side of the infrared thermal imager on the bottom surface of the fixed frame; the right sound sensor is used to collect the first sound signal of the belt conveyor; the left sound sensor is located at one end of the left side of the infrared thermal imager on the bottom surface of the fixed frame, and the left sound sensor is used to collect the second sound signal of the belt conveyor.

[0030] In this scheme, the infrared thermal imager, the right sound sensor and the left sound sensor can be installed on the fixed frame, so that the infrared thermal imager, the right sound sensor and the left sound sensor can be fixedly installed, and basically the infrared thermal imager, the right sound sensor and the left sound sensor will not be affected by the belt conveyor, ensuring the accuracy of the detection data.

[0031] Specifically, the middle of the upper surface of the belt conveyor is fixedly provided with a fixed frame at the middle position of the conveyor belt, which is convenient for users to fix and install a plurality of detection mechanisms.

[0032] Specifically, the infrared thermal imager can collect the image information of the surface of the belt conveyor.

[0033] Specifically, two sound sensors can be arranged to collect the sound information of the belt conveyor.

[0034] The above belt conveyor detection system includes a driving motor 8, the driving motor is located on one side of the front of the belt conveyor, and the driving motor is used to drive the belt conveyor to rotate.

[0035] The scheme, in the system is provided with a driving motor, so as to provide power for multiple devices in the system, realize the movement and work of multiple devices in the system.

[0036] The above belt conveyor further comprises a positioning module 9 for determining the rotation angle of the driving motor, the positioning module comprising an encoder 10 electrically connected to the driving motor, the encoder being used to control the rotation of the driving motor.

[0037] In the scheme, the encoder can realize the positioning of the position of the belt conveyor by encoding the position information into digital information.

[0038] In the scheme, the positioning of the position of the belt conveyor can realize timely determination of the position of the problem conveyor after detecting the problem.

[0039] Specifically, the driving motor is fixedly arranged on the front side of the belt conveyor, and the encoder is fixedly connected to the upper end of the driving motor, so as to facilitate the positioning of the position of the belt conveyor.

[0040] Specifically, the encoder usually has two types: incremental encoder and absolute encoder.

[0041] The incremental encoder measures the position by detecting the rotation or movement of the device components. It encodes the position information into pulse signals, usually two phase-difference signals. By calculating the number and direction of pulses, the position of the device can be determined. Incremental encoders require a starting position reference point, usually the initial position of the device.

[0042] Absolute encoders can directly provide absolute position information of the device without reference points. They encode each position as a unique digital code. The working principle of absolute encoders usually has different technologies such as optics, magnetism or capacitance. By reading the digital code output by the encoder, the position of the device can be accurately determined.

[0043] The above belt conveyor detection system comprises a main control module 11 and a data transmitter 12, the main control module comprising a main control board 13 and a wireless transmission module 14, the main control board being in communication connection with the belt conveyor, and the main control board being used to control the whole system; the wireless transmission module is in communication connection with the main control board, and the wireless transmission module is used to transmit data to the data transmitter, and the data transmitter transmits data to the upper computer.

[0044] In the scheme, by setting the main control board, the technical effect of overall control and use can be realized, and by setting the wireless transmission module, the module can realize the long-distance data transmission technical effect based on Bluetooth connection technology and wireless network connection technology.

[0045] Specifically, the belt conveyor is wirelessly connected with a main control board, the main control board is connected with a data transmitter through a data line, and the data transmitter is wirelessly connected with an upper computer, so that the functions of data acquisition and data analysis can be realized during use.

[0046] Specifically, the information acquisition module and the positioning module are connected with a main control module, so that positioning can be performed during use by the user, the main control module is connected to the upper computer, the information acquisition module comprises an infrared thermal imager, a left sound sensor and a right sound sensor, the positioning module comprises an encoder, and the main control module comprises a main control board and a wireless transmission module, so that the functions of data acquisition and data analysis of the scheme can be realized, and the detection effect during use of the scheme can be realized.

[0047] A control method of any one of the above belt conveyor detection systems, as shown in Figure 3 and Figure 4 , specifically comprising the following working steps:

[0048] S1. Data information acquisition step, acquiring the above-mentioned sensing data under different operating states of the above-mentioned belt conveyor, the above-mentioned sensing data including the above-mentioned infrared image and the above-mentioned sound signal;

[0049] S2. Data transmission step, starting LabVIEW software through the upper computer, transmitting the above-mentioned sensing data through the above-mentioned data transmitter, recording and storing the above-mentioned sensing data through the above-mentioned main control board, calling MATLABScript script through the above-mentioned LabVIEW software, and processing the collected and stored above-mentioned sensing data through the above-mentioned MATLABScript script, which needs to highlight important information in the above-mentioned infrared image, while weakening or removing unnecessary information;

[0050] S3. Data level processing step, filtering (preprocessing) the above-mentioned infrared image using the above-mentioned sound signal, and segmenting and extracting features of the salient region of the above-mentioned infrared image, to obtain the processed above-mentioned infrared image, wherein the salient region is a region with a gray value greater than a preset value;

[0051] S4. Feature-level processing step: Construct a first detection model, wherein the first detection model is trained using multiple sets of training data through the VNWOA algorithm and the RB algorithm, and each set of training data includes historical infrared images acquired within a historical time period and a first historical detection result; input the processed infrared image into the first detection model to obtain the first detection result corresponding to the infrared image; construct a second detection model, wherein the second detection model is trained using multiple sets of training data through the VNWOA algorithm and the RB algorithm, and each set of training data includes historical sound signals acquired within a historical time period and a second historical detection result; input the sound signal into the second detection model to obtain the second detection result corresponding to the sound signal;

[0052] Specifically, this can be achieved by optimizing the BP neural network using VNWOA, obtaining preliminary subnet diagnosis results through the VNWOA-BP neural network fault diagnosis model, and judging the basic probability of the fault.

[0053] S5. Decision-level processing steps: merge the first detection result and the second detection result for diagnosis, sort the detection results, average the sorted sequence number, normalize the average sorted sequence number, and determine the target detection result of the belt conveyor; output the target detection result and display the image, waveform, and fault information.

[0054] In this scheme, image acquisition and analysis functions are achieved through S1 to S5.

[0055] In step S3 above, such as Figure 5 As shown, the filtering process includes the following steps:

[0056] Step A1: Determine the maximum size of the window for expanding the infrared image based on the outermost boundary value to ensure that the infrared image can fill the window and guarantee the effect of subsequent processing.

[0057] In step A2, the points (x, y) in the infrared image are traversed sequentially using a 3x3 initial filtering window. The pixel value at the current window center point (i, J) is taken as the center point of the adaptive filtering window. The maximum and minimum gray values ​​are found in the neighborhood of the filtering window that forms X.

[0058] In step A3, an initial noise assessment is performed. If the noise (Zv) is less than or equal to the minimum grayscale value (Zmin) or greater than or equal to the maximum grayscale value (Zmax), the process proceeds to the secondary noise detection in step A4. Otherwise, step A3 is repeated, and the window W (W = w + 2)f is increased by a preset step size until W > N. The original grayscale value is then output for subsequent processing.

[0059] A4 step, when the noise is less than the first gray value or the noise is greater than the second gray value, the filtered output is the improved mean filter value item, otherwise increase the window size, if the noise is less than the first gray value or the noise is greater than the second gray value, execute step A3, until the window size is Ng, Ng is the number of pixels with g gray level in the infrared image, N is the total number of pixels in the infrared image, greater than Ng value is that no noise point is detected, output the original gray value in A3; T1 < T2;

[0060] A5 step, execute steps A2 to A4, filter all points contained in the infrared image using the filter, remove the extended boundary in A1, until the processed infrared image. Not only can maintain the clarity of the image, but also more effective for high concentration of noise.

[0061] In this scheme, the infrared image is segmented, which is convenient for subsequent processing.

[0062] In the above S3 step, the Canny algorithm is used to extract the edges of the image segmented based on the region saliency, to obtain the segmented infrared image, which is beneficial to improve the accuracy of image segmentation and feature extraction.

[0063] In this scheme, the Canny algorithm can accurately detect and extract the edges in the image. The edge is one of the important features in the image, and the edge extracted by the Canny algorithm is clear and accurate, which is helpful for subsequent image analysis and processing.

[0064] Specifically, the specific steps of edge extraction and image segmentation by Canny algorithm are as follows:

[0065] Step S31, convert to gray image: convert the original image to a gray image for subsequent processing;

[0066] Step S32, Gaussian filtering: apply Gaussian filtering to the gray image to reduce the influence of noise. Gaussian filtering can be realized by convolution operation, using a Gaussian kernel with appropriate size;

[0067] Step S33, calculate gradient: calculate the gradient amplitude and direction of each pixel in the image using Sobel operator. Usually, horizontal and vertical Sobel operators are used to convolve the image to obtain the gradient value and direction of each pixel;

[0068] Step S34, non-maximum suppression: perform non-maximum suppression on the gradient magnitude image to extract the edges corresponding to the local maximum gradient values. For each pixel, compare its two neighboring pixels in the gradient direction, and keep the pixel with the maximum gradient magnitude, and set the other pixels to 0;

[0069] Step S35, double threshold processing: perform threshold processing on the non-maximum suppressed image according to the set high threshold and low threshold. Pixels higher than the high threshold are considered as strong edges, pixels lower than the low threshold are considered as non-edges, and pixels between the two thresholds are determined as weak edges according to whether they are connected to strong edge pixels;

[0070] Step S36, edge connection: connect weak edge pixels to strong edge pixels according to their connectivity to form complete edge lines;

[0071] Step S37, image segmentation: according to the obtained edge image, image segmentation can be performed, such as target recognition or region segmentation by closed regions of edges, etc.

[0072] Specifically, the Canny algorithm can reduce noise interference in the edge extraction process through multiple thresholding and non-maximum suppression steps. This can improve the accuracy of edge extraction and avoid misjudging noise as edges. The Canny algorithm can connect broken edges through thresholding and edge connection steps to form complete edge contours. This can better describe objects and structures in the image and provide a basis for subsequent target recognition, segmentation, and tracking tasks. The Canny algorithm reduces computational complexity through Gaussian filtering and non-maximum suppression steps to improve algorithm efficiency. This is particularly important for large-scale image data processing and real-time applications.

[0073] In addition, the Canny operator can be used to extract features, and the specific steps of edge extraction and image segmentation using the Canny operator are as follows:

[0074] Step S37, convert the image to a grayscale image, as the Canny operator can only process grayscale images;

[0075] Step S38, perform Gaussian filtering on the grayscale image to reduce noise in the image. Gaussian filtering can be implemented using the GaussianBlur function in the OpenCV library;

[0076] Step S39, use the Canny operator for edge detection. The specific steps of the Canny operator are as follows:

[0077] Step S40, gradient calculation is performed on the image to obtain the gradient amplitude and direction of each pixel point in the image. The Sobel function in the OpenCV library can be used to calculate the gradient of the image;

[0078] Step S41, non-maximum suppression is performed on the gradient amplitude to refine the edges;

[0079] Step S42, a double-threshold processing is applied to divide the edges into three parts: strong edges, weak edges, and non-edges. The Canny function in the OpenCV library can be used to achieve this;

[0080] Step S43, further processing and segmentation are performed on the edge image obtained by the Canny operator. Suitable methods such as threshold segmentation and region growing can be selected according to specific requirements.

[0081] The Canny operator can accurately detect edges in an image by using gradient information to determine edge position and direction. It can accurately identify the intensity and direction of the edge, providing more accurate edge detection results. Before performing edge detection, the Canny operator first smoothes the image using a Gaussian filter to suppress noise interference. This makes the Canny operator have a certain robustness to noise during edge detection, improving the accuracy of edge detection. The Canny operator uses non-maximum suppression to accurately locate the edge. This method can preserve the most obvious edges and eliminate multiple responses of the edges. This can result in more accurate and continuous edges, which helps subsequent image segmentation and target recognition. The Canny operator can reduce the false detection rate and the missed detection rate during edge detection. It adjusts the sensitivity of edge detection by determining appropriate high and low thresholds, thereby reducing the possibility of false detection and false segmentation.

[0082] The above S3 step specifically includes the following steps:

[0083] B1 step, calculate the standard deviation of each pixel of the above infrared image in the NXN neighborhood window, and establish a variance texture map, where N≥1;

[0084] B2 step, in the above variance texture map, points with a brightness value greater than a first preset brightness value are taken as boundary points, and points with a brightness value less than a second preset brightness value are taken as non-boundary points, and the first preset brightness value is greater than the second preset brightness value (in the figure, the brighter points represent the boundary points with larger gray values, and the darker points represent the non-boundary points with smaller gray values);

[0085] B3, the boundary points in the variance texture map are taken as feature points, screening and clustering are performed to obtain the center of the saliency region, and the saliency region of the infrared image is obtained (the brighter points in the variance texture map of the saliency region are taken as feature points, the feature points are screened and clustered to obtain the center of the saliency region, and the saliency region of the infrared image is obtained, which facilitates the region segmentation function of the device in use, and improves the effect of subsequent feature extraction).

[0086] Screening: the variance texture map contains boundary points, non-boundary points, and points that are not determined as boundary points. The undetermined points are traversed. If the points around the point contain boundary points, the point is a boundary point, otherwise it is a non-boundary point. Only boundary points and non-boundary points are included in the output result, and the screening is completed.

[0087] Clustering: through the DBSCAN algorithm, the radius and the number of density points are set as threshold values, and the Euclidean distance between multiple points is calculated, so that the regions in the texture map can be divided.

[0088] In this scheme, the accuracy of image segmentation is improved through the above steps.

[0089] In the above S3 step, the feature extraction specifically includes the following steps:

[0090] C1, a saliency region segmentation method is used to segment the target of the filtered infrared image.

[0091] C2, first, the color image is subjected to Gaussian low-pass filtering and local entropy texture segmentation to remove the texture region, obtaining a filtered gray image of R, G, and B components. Clustering segmentation automatically divides the three regions of the brightest region, the darkest region, and the remaining region of each component. The region with the largest brightness difference between the selected brightest or darkest region and the remaining region is selected. The boundary of the selected region is detected for corner points and edge points. These feature points are taken as salient points. Filling and expansion are performed to close the region, and the largest connected region is extracted. Finally, the saliency region is obtained, and the fault region of the belt conveyor is segmented.

[0092] Through contour detection, all contours in the color component can be obtained, and the points of each contour are stored in an array. All arrays are traversed, the area of each array is calculated, and the array with the largest connected region area is extracted to complete the segmentation. The fault region generally has a certain temperature feature, which is represented in the form of a digital matrix. The region without temperature feature is a non-fault region.

[0093] C3, the Canny algorithm is used to extract the edge of the segmented fault region, and the shape feature of the fault region is obtained (which is conducive to improving the detection accuracy and detection effect of the device in use).

[0094] In this scheme, the basic feature extraction function is implemented.

[0095] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary and that changes in form and detail can be made without departing from the principles and spirit of the application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A control method of a belt conveyor detection system, characterized by, The belt conveyor detection system comprises a belt conveyor, an upper computer, a master control module and a data transmitter, the master control module comprises a master control board and a wireless transmission module, the master control board is in communication connection with the belt conveyor, the master control board is used for controlling the whole system, the wireless transmission module is in communication connection with the master control board, the wireless transmission module is used for transmitting data to the data transmitter, the data transmitter transmits data to the upper computer, and specifically comprises the following steps: S1. Data information acquisition step, collecting sensing data of the belt conveyor in different operating states, the sensing data comprising infrared images and sound signals; S2. Data transmission step, starting LabVIEW software through the upper computer, transmitting the sensing data through the data transmitter, recording and storing the sensing data through the master control board, calling MATLABScript script through the LabVIEW software, and processing the collected and stored sensing data through the MATLABScript script, wherein the important information in the infrared image is highlighted, and the unnecessary information is weakened or removed; S3. Data level processing step, filtering the infrared image using the sound signal, segmenting and extracting features of the significant area of the infrared image, and obtaining the processed infrared image, wherein the significant area is an area with a gray value greater than a preset value; S4. Feature level processing step, constructing a first detection model, wherein the first detection model is trained by VNWOA algorithm and BP algorithm using multiple sets of training data, each set of training data in the multiple sets of training data comprising historical infrared images and first historical detection results obtained in a historical time period; The processed infrared image is input into the first detection model to obtain a first detection result corresponding to the infrared image; Constructing a second detection model, wherein the second detection model is trained by VNWOA algorithm and BP algorithm using multiple sets of training data, each set of training data in the multiple sets of training data comprising historical sound signals and second historical detection results obtained in a historical time period; The sound signal is input into the second detection model to obtain a second detection result corresponding to the sound signal; S5. Decision level processing step, fusing the first detection result and the second detection result for diagnosis, sorting the detection results, averaging the sorting sequence numbers, normalizing the average sorting sequence numbers, and determining a target detection result of the belt conveyor; outputting the target detection result and displaying the image, waveform and fault information; In the S3 step, the filtering processing comprises the following steps: A1. Step, according to the outermost boundary value, determine the maximum size of the expansion window of the infrared image, to ensure that the infrared image can fill the window, and ensure the effect of subsequent processing; A2, sequentially traversing the points in the infrared image with a 3x3 initial filtering window, taking the pixel value at the center point of the current window as the center point of the adaptive filtering window, finding the maximum and minimum gray scale values in the filtering window neighborhood of X; A3, performing a primary noise judgment, if the noise is less than or equal to the minimum gray scale value or the noise is greater than or equal to the maximum gray scale value, then going to the secondary noise detection in A4, otherwise continuing to execute step A3 in a loop, and increasing the window size by a preset step size until W>N, outputting the original gray scale value for subsequent processing; A4, when performing the secondary noise detection, if the noise is less than the first gray scale value or the noise is greater than the second gray scale value, then outputting the improved mean filter value item after filtering, otherwise increasing the window size, if the noise is not less than the first gray scale value or the noise is not greater than the second gray scale value, then executing step A3 in a loop until the window size is Ng, Ng being the number of pixels of the gray scale g in the infrared image, N being the total number of pixels in the infrared image, and greater than Ng being a noiseless point detected, outputting the original gray scale value in A3; A5, executing steps A2 to A4 in a loop to filter all points contained in the infrared image using the filter, removing the extended boundary in A1 until the processed infrared image; The feature extraction in S3 specifically includes the following steps: C1, performing target segmentation on the filtered infrared image using a salient region segmentation method; C2, first performing Gaussian low-pass filtering and local entropy texture segmentation on the color image to remove the texture region, obtaining a filtered gray scale image of R, G, and B components, clustering and segmenting to automatically divide the three regions of the brightest region, the darkest region, and the remaining region for each component, selecting the region with the largest brightness difference between the brightest or darkest region and the remaining region for each color component, detecting the corner points and edge points of the selected region boundary, taking these feature points as salient points, filling and dilating to close the region, extracting the largest connected region, and finally obtaining the salient region, segmenting the fault region of the belt conveyor; C3, using the Canny algorithm to perform edge extraction on the segmented fault region to obtain the shape feature of the fault region.

2. The control method of a belt conveyor detection system according to claim 1, characterized in that: In S3, the Canny algorithm is used to perform edge extraction on the image segmented based on region saliency to obtain the segmented infrared image.

3. The control method of the belt conveyor detection system according to claim 1, characterized in that, The salient region segmentation in S3 specifically includes the following steps: B1, calculating the standard deviation of each pixel in the infrared image in an NXN neighborhood window, and establishing a variance texture map, where N≥1; B2, taking the points with a brightness value greater than a first preset brightness value as boundary points and the points with a brightness value less than a second preset brightness value as non-boundary points in the variance texture map, the first preset brightness value being greater than the second preset brightness value; B3, filtering and clustering the boundary points in the variance texture map as feature points to obtain the center of the salient region, and obtaining the salient region of the infrared image.

4. A belt conveyor detection system characterized by, The belt conveyor detection system is configured to perform the control method of the belt conveyor detection system according to any one of claims 1 to 3, and the system comprises: The belt conveyor comprises a fixed frame and an information acquisition module, wherein the fixed frame is located at a middle position of an upper surface of the belt conveyor, and the information acquisition module is located on the fixed frame and configured to acquire sensing data of the belt conveyor, the sensing data comprising infrared images and sound signals; The upper computer is in communication connection with the belt conveyor, and is configured to determine whether the belt conveyor is faulty according to the sensing data.

5. The belt conveyor detection system of claim 4, wherein, The information acquisition module comprises: The infrared thermal imager is located at a middle position of a bottom surface of the fixed frame, and is configured to acquire the infrared images of the belt conveyor, wherein the middle position is the same distance from a left end point of the fixed frame as from a right end point of the fixed frame; The right sound sensor is located at one end of the bottom surface of the fixed frame close to the right side of the infrared thermal imager, and is configured to acquire a first sound signal of the belt conveyor; The left sound sensor is located at one end of the bottom surface of the fixed frame close to the left side of the infrared thermal imager, and is configured to acquire a second sound signal of the belt conveyor.

6. The belt conveyor detection system of claim 4, wherein, The belt conveyor detection system comprises: The driving motor is located at one side of a front face of the belt conveyor, and is configured to drive the belt conveyor to rotate.

7. The belt conveyor detection system of claim 6, wherein, The belt conveyor further comprises a positioning module configured to determine a rotation angle of the driving motor, and the positioning module comprises: The encoder is in electrical connection with the driving motor, and is configured to control the rotation of the driving motor.

8. The belt conveyor detection system of claim 4, wherein, The belt conveyor detection system comprises a main control module and a data transmitter, and the main control module comprises: The main control board is in communication connection with the belt conveyor, and is configured to control the whole system; The wireless transmission module is in communication connection with the main control board, and is configured to transmit data to the data transmitter, and the data transmitter transmits data to the upper computer.

Images