Scraper conveyor chain breakage fault detection method and device

Abstract

The invention relates to the technical field of mechanical fault detection, in particular to a scraper conveyor chain breakage fault detection method and device, and the method comprises the following steps: collecting a picture of a scraper of a scraper conveyor, and building a data set according to the collected picture of the scraper; designing a rotating target detection network model according to the characteristics of the established data set; the invention provides a scraper conveyor chain breakage fault detection method and device, solves the problems that in the prior art, a sensor is prone to abrasion and poor in severe environment resistance, and a visual detection method is poor in real-time performance and not high in applicability, improves the intelligent level of a scraper conveyor, and improves the working efficiency of the scraper conveyor. By means of the detection method and device, faults can be found in advance so as to reduce economic losses, the equipment maintenance period can be shortened, and the method and device have important engineering application value for guaranteeing continuous and safe production of coal, reducing the labor intensity of workers and reducing influences and losses caused by chain breakage faults of the scraper conveyor to the maximum extent.

Description

Technical Field

[0001] The invention relates to the technical field of mechanical fault detection, and in particular to a method and device for detecting chain break fault of a scraper conveyor. Background Art

[0002] Scraper conveyor is the core equipment of the comprehensive mining face, and plays a vital role in the continuous and safe production of coal mines. The chain is the main component of the scraper conveyor. The occurrence of chain breakage will not only lead to production interruption, but also pose a threat to the life safety of operators.

[0003] According to different principles, the detection of broken chain faults can be divided into posture detection method, current detection method, tension detection method, speed detection method, vision-based detection method, etc.

[0004] Posture detection method: detect the inclination of both ends of the scraper. When the inclination exceeds the threshold, it indicates that a chain breakage failure has occurred. The sensor is not convenient to install and is easily damaged or worn.

[0005] Vision-based detection methods: (1) Based on angle histograms, they cannot cope with complex scenes and have poor applicability;

[0006] (2) It is necessary to first extract the scraper profile information, and then obtain the straight line information of the scraper profile by linear fitting. This has poor real-time performance and cannot cope with complex scenarios.

[0007] (3) The lighting module and image acquisition module are fixed under the middle plate, and the installation cannot be guaranteed, resulting in poor accuracy in detecting high-speed chains;

[0008] (4) The image needs to be segmented first and then the pixels are traversed, which cannot guarantee real-time performance;

[0009] In general, the core component sensors of the posture detection method in the existing technology are easy to wear and have poor resistance to harsh environments, while the visual detection method has poor real-time performance and low applicability. Summary of the invention

[0010] The object of the present invention is to provide a method and device for detecting chain breakage faults of a scraper conveyor, so as to solve the problems raised in the above-mentioned background technology.

[0011] To achieve the above object, the present invention provides the following technical solution: a method and device for detecting chain breakage fault of a scraper conveyor, the detection method comprising the following steps:

[0012] S10, collecting pictures of the scraper conveyor, and establishing a data set according to the collected pictures of the scraper;

[0013] S20. Design a rotating target detection network model based on the characteristics of the established data set;

[0014] S30, training the rotation target detection network model using the established data set;

[0015] S40, inputting the acquired video stream into a trained rotating target detection network model, and outputting the offset angle information of the scraper;

[0016] S50: judging whether a chain breakage fault occurs in the scraper conveyor according to the scraper deviation angle information of the scraper conveyor.

[0017] Preferably, in step S10, a certain number of scraper conveyor scraper video images are obtained by using a camera, and the targets on the images are annotated using a picture annotation tool. The data set is divided into a training set, a validation set, and a test set according to a certain ratio. Data collection in various environments is simulated by rotation, random cropping, scale scaling, Gaussian noise, brightness adjustment, and flipping adjustment to enrich data diversity and improve model robustness.

[0018] Preferably, in step S20, the rotating target detection network model includes an angle output dimension, and the angle and other coordinate point information are combined to obtain a rotating rectangular frame of the scraper.

[0019] Preferably, in step S30, the training set and the validation set are input into the rotating target detection network model for training. After the training is completed, the test set is input into the rotating target detection network model, the detection accuracy is calculated, and it is determined whether the requirements are met. If so, the rotating target detection network model and training parameters are saved.

[0020] Preferably, in step S50, the obtained scraper blade offset angle information is compared with a set threshold value, and if the offset angle exceeds the set threshold value, it is determined that a chain break fault occurs.

[0021] Preferably, the threshold is set according to the length, width, spacing and normal wear of the scraper.

[0022] A scraper conveyor chain break fault detection device includes a detection device, which includes an image acquisition module, a control module, an alarm module, and a power supply module. The output end of the image acquisition module is electrically connected to the input end of the control module, the output end of the control module is electrically connected to the input end of the alarm module, and the output end of the power supply module is electrically connected to the input ends of the image acquisition module, the control module, and the alarm module.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides a method and device for detecting broken chain faults of a scraper conveyor, which overcomes the problems in the prior art that sensors are easy to wear, have poor resistance to harsh environments, and that visual detection methods have poor real-time performance and low applicability, and improves the intelligence level of the scraper conveyor. Through the detection method and device, faults can be discovered early to reduce economic losses and shorten equipment maintenance cycles, which has important engineering application value for ensuring continuous and safe production of coal, reducing the labor intensity of workers, and minimizing the impact and losses caused by broken chain faults of scraper conveyors. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flow chart of the main steps of the scraper conveyor chain breakage fault detection method provided by the present invention;

[0025] Figure 2 A logic diagram of the working principle of the device provided by the present invention;

[0026] Figure 3 A schematic diagram of a scraper conveyor chain breakage fault detection device provided by the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] See also Figures 1 to 3 The present invention provides a technical solution: a method and device for detecting chain breakage fault of a scraper conveyor, the detection method comprising the following steps:

[0029] S10, collecting pictures of the scraper conveyor, and establishing a data set according to the collected pictures of the scraper;

[0030] S20. Design a rotating target detection network model based on the characteristics of the established data set;

[0031] S30, training the rotation target detection network model using the established data set;

[0032] S40, inputting the acquired video stream into a trained rotating target detection network model, and outputting the offset angle information of the scraper;

[0033] S50: judging whether a chain breakage fault occurs in the scraper conveyor according to the scraper deviation angle information of the scraper conveyor.

[0034] In step S10, a certain number of scraper conveyor scraper video images are obtained by using a camera, and the targets on the images are annotated using a picture annotation tool. The data set is divided into a training set, a validation set, and a test set according to a certain ratio. Data collection in various environments is simulated by rotation, random cropping, scale scaling, Gaussian noise, brightness adjustment, and flip adjustment to enrich data diversity, improve model robustness, and use a picture annotation tool to annotate the targets on the images.

[0035] In step S20, the rotating target detection network model includes an angle output dimension. The angle and other coordinate point information are combined to obtain the rotating rectangular frame of the scraper. The above-mentioned pictures do not contain label information and cannot be used for network model training. It is necessary to annotate each scraper conveyor picture to obtain the position, size, and angle information of the minimum circumscribed rectangle of the scraper. Use the roLabelImg annotation tool to annotate and use the long side representation method to describe the label information. The annotation format of the long side representation method is (x le ,y le ,w le ,h le ,θ le ).

[0036] Among them, it means (x le ,y le ) The coordinates of the center point of the minimum enclosing rectangle, w le is the long side of the minimum circumscribed rectangle, h le is the short side of the minimum circumscribed rectangle, w le It indicates the angle from the x-axis to the long side in the counterclockwise direction. After the annotation is completed, a label file will be generated to record the image information and annotation information.

[0037] Furthermore, in order to enrich data diversity and improve the robustness of the model, data collection in various environments is simulated through rotation, random cropping, scale scaling, Gaussian noise, brightness adjustment and flip adjustment.

[0038] Furthermore, the ratio of training set, validation set and test set is 8:1:1.

[0039] Further, in step S20, according to the characteristics of the data set, a rotating target detection network model is designed. The rotating target detection network model includes an angle output dimension, and the angle and other coordinate point information are combined to obtain a rotating rectangular frame of the scraper.

[0040] In this embodiment, the target detection network Yolov7 is used as the basic architecture, and modifications are made on its basis. In order to avoid the problem of sudden change in boundary angle during angle regression, an annular smooth label is introduced to discretize the continuous angle, and 180° is regarded as 180 angle categories for training. The annular smooth label is based on the periodicity of the angle and uses a Gaussian window function to give the real label the same periodicity. The calculation formula is as follows:

[0041]

[0042] Where: θ refers to the angle of the current bounding box, r is the radius of the window function, g(x) is the window function, and the calculation formula is as follows

[0043]

[0044] Wherein: a, b, r are constant terms, representing the height of the Gaussian curve peak and the center position of the peak respectively, x is the input angle information, and r is the width of the Gaussian curve.

[0045] Furthermore, the output layer is modified to add an angle output dimension to the network prediction.

[0046] In this embodiment, 180 angle classification channels are added to the network output layer, and the output information dimension is adjusted from the original 5+nc to 5+nc+180, where 5 represents the dimension of the horizontal target detection algorithm in the output information, including (x, y, w, h, s), i.e., the center point coordinates, width, height and classification confidence of the detection box, and nc represents the original number of data categories.

[0047] Furthermore, the loss function is modified and angle classification loss is added to learn angle information.

[0048] In this embodiment, the original YOLOv7 loss function includes three parts: bounding box regression loss, category loss, and confidence loss. The modified rotation target detection model adds angle category information, so the loss function needs to be modified accordingly. The total loss calculation formula is as follows:

[0049] Loss = L box +L cls +L obj +L θ

[0050] Where: L box is the bounding box regression loss, using the CIOU loss function, L cls is the category loss, L obj is the confidence loss, L θ is the angle loss, and both use binary cross entropy loss.

[0051] In step S30, the training set and the validation set are input into the rotating target detection network model for training. After the training is completed, the test set is input into the rotating target detection network model, the detection accuracy is calculated, and it is determined whether the requirements are met. If so, the rotating target detection network model and training parameters are saved.

[0052] In this embodiment, the original anchor frame of the YOLOv7 network cannot meet the needs of the data set of this method, so this method uses the K-means++ clustering algorithm to calculate the prior frame of the data set, and generates prior frames of different numbers and sizes through anchor frame clustering to increase the matching degree between the prior frame and the actual target frame as much as possible, thereby further improving the detection accuracy, and then uses the SGD optimizer to train the model. The model training can be performed on a high-performance GPU of the server. By setting different parameters, different model training effects are obtained. The model is evaluated using a test set. If the detection requirements are met, it is determined to be a deployable network model. If the requirements are not met, the parameters or network results are adjusted in time to retrain the model until a usable network model is obtained.

[0053] In step S40, the acquired video stream is input into the trained rotating target detection network model, and the offset angle information of the scraper is output.

[0054] In this embodiment, the real-time video stream of the scraper conveyor captured by the image acquisition device is input into the trained rotating target detection network model, and the detection frame information of the scraper in the video stream is output, wherein the detection frame information includes (x, y, w, h, s, θ), that is, the center point coordinates, width, height, classification confidence and angle of the detection frame. It is necessary to separate the angle information from the above information, and convert it from radians to degrees, and take its absolute value to facilitate subsequent processing.

[0055] In step S50, the obtained scraper offset angle information is compared with the set threshold. If the offset angle exceeds the set threshold, it is determined that a chain breakage failure has occurred, wherein the threshold is set based on the length, width, spacing and normal wear of the scraper.

[0056] In this embodiment, according to the regulations of the high-strength round link chain for mining, the maximum elongation of the chain is 12%, and the maximum deflection angle threshold of the scraper can be set to 90°×12%, that is, 10°. The above-obtained offset angle value is compared with the set threshold. If the offset angle exceeds the set threshold, it is determined that a chain break failure has occurred.

[0057] A scraper conveyor chain break fault detection device comprises a detection device, the detection device comprises an image acquisition module, a control module, an alarm module, and a power module, the output end of the image acquisition module is electrically connected to the input end of the control module, the image acquisition module is composed of an industrial camera, a lens or a camera, and is used to obtain real-time image information of the scraper conveyor and transmit the image information to the control module, the output end of the control module is electrically connected to the input end of the alarm module, a rotating target detection network model is deployed in the control module to perform image processing, and then the occurrence of the fault is judged and processed in combination with the obtained scraper offset angle information and the set threshold, the output end of the power module is electrically connected to the input end of the image acquisition module, the control module, and the alarm module, in this embodiment, due to the complex coal mining environment, in order to ensure safety, the image acquisition module, the alarm module, and the power module can be selected as intrinsically safe explosion-proof type.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting chain breakage fault of a scraper conveyor, characterized in that: The detection method comprises the following steps: S10, collecting pictures of the scraper conveyor, and establishing a data set according to the collected pictures of the scraper; S20. Design a rotating target detection network model based on the characteristics of the established data set; S30, training the rotation target detection network model using the established data set; S40, inputting the acquired video stream into a trained rotating target detection network model, and outputting the offset angle information of the scraper; S50: judging whether a chain breakage fault occurs in the scraper conveyor according to the scraper deviation angle information of the scraper conveyor.

2. A scraper conveyor chain breakage fault detection method according to claim 1, characterized in that: In step S10, a certain number of scraper conveyor scraper video images are obtained by using a camera, and the targets on the images are annotated using a picture annotation tool. The data set is divided into a training set, a validation set, and a test set according to a certain ratio. Data collection in various environments is simulated by rotation, random cropping, scale scaling, Gaussian noise, brightness adjustment, and flip adjustment to enrich data diversity and improve model robustness.

3. A scraper conveyor chain breakage fault detection method according to claim 2, characterized in that: In step S20, the rotating target detection network model includes an angle output dimension, and the angle and other coordinate point information are combined to obtain a rotating rectangular frame of the scraper.

4. A scraper conveyor chain breakage fault detection method according to claim 3, characterized in that: In step S30, the training set and the validation set are input into the rotating target detection network model for training. After the training is completed, the test set is input into the rotating target detection network model, the detection accuracy is calculated, and it is determined whether the requirements are met. If so, the rotating target detection network model and training parameters are saved.

5. A scraper conveyor chain breakage fault detection method according to claim 4, characterized in that: In the step S50, the obtained scraper blade deviation angle information is compared with a set threshold value, and if the deviation angle exceeds the set threshold value, it is determined that a chain breakage fault occurs.

6. A scraper conveyor chain breakage fault detection method according to claim 5, characterized in that: The threshold is set according to the length, width, spacing and normal wear of the scraper.

7. A scraper conveyor chain breakage fault detection device, used in a scraper conveyor chain breakage fault detection method according to any one of claims 1 to 6, characterized in that: The detection device comprises an image acquisition module, a control module, an alarm module and a power module. The output end of the image acquisition module is electrically connected to the input end of the control module, the output end of the control module is electrically connected to the input end of the alarm module, and the output end of the power module is electrically connected to the input ends of the image acquisition module, the control module and the alarm module.

Images