Two-for-one twister abnormal state detection method and system, medium and computer

Abstract

The invention relates to a two-for-one twister abnormal state detection method and system, a medium and a computer, and has the following beneficial effects that a spindle picture is divided into a raw material pipe picture, a yarn passing wheel picture and a ternary picture, and the raw material pipe picture, the yarn passing wheel picture and the ternary picture are respectively detected and judged; compared with an overall spindle picture, the number of pixels is greatly reduced, so that the computing power demand of target detection can be effectively reduced. And in addition, as the abnormal conditions of the raw material pipe, the ternary part and the wire passing wheel are different, the actual conditions of all the parts need to be judged respectively. And when any part is found to be abnormal, an early warning signal is correspondingly sent out, so that related workers can repair and maintain the equipment in time. Compared with manual inspection for judging the working state of the spindle, the method is low in omission ratio, small in computing power requirement and higher in inspection efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of textile equipment detection, and more specifically, to a method, system, medium and computer for detecting abnormal state of a two-for-one twisting machine. Background Art

[0002] The two-for-one twister is a twisting device that can achieve two twists in one turn. The twisting efficiency is doubled compared to traditional twisting equipment. The two-for-one twister can bond two or more single yarns into strands by twisting, and enhance the performance of the original yarn. The two-for-one twister spindle is the heart and key of the two-for-one twister. The biggest difference between the two-for-one twister and the ring twister lies in the spindle. The ordinary spindle can only add one twist per rotation, while the two-for-one twister spindle can produce two twists per rotation.

[0003] At present, the chemical fiber filament weaving industry generally uses double-layer filament two-for-one twisting machines, which are divided into upper and lower layers, with left and right sides. Each layer generally has 128 spindles, and each spindle is a production unit. Each spindle usually includes: three-element, wire wheel and raw material tube; in order to ensure that the spindles of the two-for-one twisting machine can work normally, these parts need to be monitored separately. To ensure that each spindle of the two-for-one twisting machine is in a normal working state.

[0004] Since there are a large number of spindles on the two-for-one twisting machine and the repetitiveness is high, it is easy to make mistakes when checking the spindle status through manual inspection, resulting in the production of defective products. Summary of the invention

[0005] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method, system, medium and computer for detecting abnormal state of a two-for-one twister, so as to overcome the shortcomings of the prior art that there are a large number of spindles on the two-for-one twister and manual detection of the spindle state is prone to omissions.

[0006] The above technical objectives of the present invention are achieved through the following technical solutions: A method for detecting abnormal state of a two-for-one twisting machine, comprising:

[0007] S1. Take a photo of the spindle on the two-for-one twister to obtain a corresponding spindle photo;

[0008] S2. Cut the ingot photo to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo;

[0009] S3. According to the raw material tube photo, the thread passing wheel photo and the ternary photo, it is determined whether the raw material tube, the thread passing wheel and the ternary of the two-for-one twister are all in normal working state. If not, an early warning signal is issued.

[0010] Optionally, judging whether the raw material tube of the two-for-one twister is in a normal working state according to the raw material tube photo includes:

[0011] Acquire a plurality of first training photos for training the model, wherein the first training photos contain raw material tubes in a normal working state or contain raw material tubes in an abnormal working state; fill each first training photo respectively to obtain a corresponding first enhanced photo;

[0012] Acquire channel information of each first enhanced photo respectively to obtain a corresponding first three-dimensional tensor; splice all the first three-dimensional tensors to obtain a corresponding first four-dimensional tensor;

[0013] Inputting the first four-dimensional tensor into a pre-established first neural network model for training to obtain a corresponding second neural network model;

[0014] The raw material pipe photo is input into the second neural network model for prediction to obtain a prediction result, and based on the prediction result, it is determined whether the working state of the raw material pipe in the raw material pipe photo is abnormal.

[0015] Optionally, the step of inputting the raw material tube photo into the second neural network model for prediction to obtain a prediction result further includes:

[0016] Inputting the raw material pipe photo into the second neural network model for prediction to obtain a plurality of first boundary boxes for framing the raw material pipes with abnormal working status;

[0017] Perform non-maximum suppression processing on several of the first bounding boxes to obtain the corresponding first bounding box with the highest confidence, and determine whether the confidence corresponding to the first bounding box with the highest confidence is greater than a preset confidence threshold. If so, it is determined that there is an abnormality in the working state of the raw material tube; if not, it is determined that there is no abnormality in the working state of the raw material tube.

[0018] Optionally, judging whether the three elements of the two-for-one twister are in a normal working state according to the three element photo includes:

[0019] Acquire a number of second training photos for training the model, wherein the second training photos contain ternary elements in a normal working state or ternary elements in an abnormal working state; fill each second training photo respectively to obtain a corresponding second enhanced photo;

[0020] Acquire channel information of each second enhanced photo respectively to obtain a corresponding second three-dimensional tensor; splice all the second three-dimensional tensors to obtain a corresponding second four-dimensional tensor;

[0021] Inputting the second four-dimensional tensor into a pre-established third neural network model for training to obtain a corresponding fourth neural network model;

[0022] The ternary photo is input into the fourth neural network model for prediction to obtain a prediction result, and based on the prediction result, it is determined whether the working status of the ternary in the ternary photo is abnormal.

[0023] Optionally, inputting the ternary photo into the fourth neural network model for prediction to obtain a prediction result includes:

[0024] Inputting the ternary photo into the fourth neural network model for prediction to obtain a plurality of second boundary boxes for ternary elements with abnormal working status;

[0025] Perform non-maximum suppression processing on several of the second bounding boxes to obtain the corresponding second bounding box with the highest confidence, and determine whether the confidence corresponding to the second bounding box with the highest confidence is greater than a preset confidence threshold. If so, it is determined that there is an abnormality in the working state of the ternary element; if not, it is determined that there is no abnormality in the working state of the ternary element.

[0026] Optionally, the determining whether the working state of the wire passing wheel in the wire passing wheel photo is abnormal includes:

[0027] An indicating arrow is arranged on the thread passing wheel of the two-for-one twisting machine; and eight photos of the thread passing wheel are taken at preset time intervals;

[0028] The eight photos of the wire-passing wheel are input into the pre-trained fifth neural network model for detection, the positions of the indicator arrows on the eight wire-passing wheels are identified, and it is determined whether the positions of the indicator arrows on the eight wire-passing wheels are different. If so, it is determined that the wire-passing wheel is in a normal working state. If not, it is determined that the working state of the wire-passing wheel is abnormal.

[0029] Optionally, the ingot photo is cut to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo, including:

[0030] Use the pre-set inspection robot to take pictures of the working status of the two-for-one twister to obtain photos of the spindles;

[0031] According to the position of the wire passing wheel of the two-for-one twister in the spindle photo, the spindle photo is cropped to obtain a corresponding first cropped photo; wherein the width of the first cropped photo is the width of the corresponding spindle photo; and the height of the first cropped photo is the height of the corresponding wire passing wheel in the corresponding spindle photo;

[0032] According to a preset size conversion rule, the first cropped photo is resized to obtain a corresponding second cropped photo; target detection is performed on the second cropped photo to obtain first position information of the corresponding wire passing wheel;

[0033] According to the size transformation rule and the first position information, the second position information of the corresponding wire wheel is determined on the first cutting photo; according to the second position information, the third position information of the three-element of the two-for-one twister and the fourth position information of the raw material tube of the two-for-one twister on the spindle photo are calculated;

[0034] According to the second position information, the corresponding ingot photo is cut to obtain the corresponding wire wheel photo; according to the third position information, the corresponding ingot photo is cut to obtain the corresponding ternary photo; according to the fourth position information, the corresponding ingot photo is cut to obtain the corresponding raw material tube photo.

[0035] A system for detecting abnormal state of a two-for-one twister, comprising:

[0036] Photo shooting module: used to take photos of the spindles on the two-for-one twister to obtain corresponding spindle photos;

[0037] Photo cutting module: used to cut the ingot photo to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo;

[0038] Status judgment module: according to the raw material tube photo, the thread wheel photo and the ternary photo, it is judged whether the raw material tube, the thread wheel and the ternary of the two-for-one twister are all in normal working state. If not, an early warning signal is issued.

[0039] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0040] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0041] In summary, the present invention has the following beneficial effects: in this application, the ingot photo is divided into a raw material tube photo, a wire wheel photo and a ternary photo, and the raw material tube photo, the wire wheel photo and the ternary photo are detected and judged respectively. Compared with the overall ingot photo, the number of pixels is greatly reduced, so the computing power requirement for target detection can be effectively reduced. And in this application, since the abnormal conditions of the raw material tube, the ternary and the wire wheel are different, it is necessary to judge separately according to the actual situation of each component. When any component is found to be abnormal, an early warning signal is issued accordingly, so that the relevant staff can repair and maintain the equipment in time. And compared with manual inspection to judge the working status of the ingot, the method of this application has a low missed detection rate, small computing power requirement and higher inspection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1It is a flow chart of a method for detecting abnormal state of a two-for-one twisting machine of the present invention;

[0043] Figure 2 It is a structural diagram of an abnormal state detection system of a two-for-one twisting machine of the present invention;

[0044] Figure 3 is an internal structure diagram of a computer device in an embodiment of the present invention;

[0045] Figure 4 Schematic diagram of the structure of the inspection robot in an embodiment of the present invention;

[0046] Figure 5 A schematic diagram of shooting a working video of the inspection robot in an embodiment of the present invention;

[0047] Figure 6 It is a schematic diagram of indicating arrows provided on the thread passing wheel in an embodiment of the present invention.

[0048] In the figure: 1. Photo shooting module; 2. Photo cutting module; 3. Raw material tube photo judgment module; 4. Wire wheel photo judgment module; 5. Ternary photo judgment module. DETAILED DESCRIPTION

[0049] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings. Several embodiments of the present invention are shown in the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein.

[0050] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances. The terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0051] In the present invention, unless otherwise clearly specified and limited, the first feature being "above" or "below" the second feature may include the first and second features being in direct contact, or may include the first and second features being in contact not directly but through another feature between them. Moreover, the first feature being "above", "above" and "above" the second feature includes the first feature being directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. The first feature being "below", "below" and "below" the second feature includes the first feature being directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature. The terms "vertical", "horizontal", "left", "right", "above", "below" and similar expressions are for illustrative purposes only, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0052] The present invention is described in detail below in conjunction with the accompanying drawings and embodiments.

[0053] The present invention provides a method for detecting abnormal state of a two-for-one twisting machine, such as Figure 1 As shown, including:

[0054] S1. Take a photo of the spindle on the two-for-one twister to obtain a corresponding spindle photo;

[0055] S2. Cut the ingot photo to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo;

[0056] S3. According to the raw material tube photo, the thread passing wheel photo and the ternary photo, it is determined whether the raw material tube, the thread passing wheel and the ternary of the two-for-one twister are all in normal working state. If not, an early warning signal is issued.

[0057] Specifically, in this application, the spindle photos are divided into raw material tube photos, wire wheel photos and ternary photos, and the raw material tube photos, wire wheel photos and ternary photos are detected and judged respectively. Compared with the overall spindle photos, the number of pixels is greatly reduced, so the computing power requirements for target detection can be effectively reduced. In this application, since the abnormal conditions of the raw material tube, ternary and wire wheel are different, the raw material tube needs to monitor the raw material residue in the raw material tube, the ternary is monitored to avoid the ternary due to wear affecting the raw material transmission, and the wire wheel is monitored to avoid the existence of broken yarn defects, resulting in no silk thread transmission, and it is necessary to monitor whether the wire wheel is rotating. Therefore, it is necessary to judge separately according to the actual situation of each component. When any component is found to be abnormal, a corresponding early warning signal is issued so that the relevant staff can repair and maintain the equipment in time.

[0058] Further, judging whether the raw material tube of the two-for-one twister is in a normal working state according to the raw material tube photo includes:

[0059] Acquire a plurality of first training photos for training the model, wherein the first training photos contain raw material tubes in a normal working state or contain raw material tubes in an abnormal working state; fill each first training photo respectively to obtain a corresponding first enhanced photo;

[0060] Acquire channel information of each first enhanced photo respectively to obtain a corresponding first three-dimensional tensor; splice all the first three-dimensional tensors to obtain a corresponding first four-dimensional tensor;

[0061] Inputting the first four-dimensional tensor into a pre-established first neural network model for training to obtain a corresponding second neural network model;

[0062] The raw material pipe photo is input into the second neural network model for prediction to obtain a prediction result, and based on the prediction result, it is determined whether the working state of the raw material pipe in the raw material pipe photo is abnormal.

[0063] In practical applications, the first training photos are all photos of raw material tubes. The first training photos are filled mainly to fill the photos of the raw material tubes from rectangles to squares, so that the horizontal and vertical pixel sizes of the photos are consistent. In this application, the photos can be filled by using the open CV algorithm to fill the edge area of ​​the photos so that the short side length of the photos is equal to the long side length, or the photos can be stretched to turn the photos into square photos. Read the channel information of the photo. In this application, for example, if the pixel size of the photo is 220*220, the three-dimensional tensor formed by the channel information of the photo is 220*220*3; in order to facilitate the training of the model, it is also necessary to connect the three-dimensional tensor into a four-dimensional tensor. The four-dimensional tensor can input thousands of training photos into the model at one time, and the model is trained, and its training efficiency is higher. Then the photos of the raw material tubes to be detected are input into the trained neural network model for prediction, and it can be judged whether the raw material tube is in an abnormal state according to the prediction results.

[0064] Furthermore, the step of inputting the photo of the raw material tube into the second neural network model for prediction to obtain a prediction result also includes: inputting the photo of the raw material tube into the second neural network model for prediction to obtain a plurality of first bounding boxes for framing raw material tubes with abnormal working status; performing non-maximum suppression processing on the plurality of first bounding boxes to obtain corresponding first bounding boxes with the highest confidence, and determining whether the confidence corresponding to the first bounding box with the highest confidence is greater than a preset confidence threshold; if so, determining that the working status of the raw material tube is abnormal; if not, determining that the working status of the raw material tube is not abnormal.

[0065] In practical applications, the results output by the target detection model usually include several bounding boxes. These bounding boxes usually overlap to a certain extent, that is, the bounding boxes marked by multiple bounding boxes are the same target. Therefore, these bounding boxes need to be deduplicated. First, the bounding box with the highest confidence is obtained, and then the intersection and union ratio between the bounding box and other bounding boxes is calculated. When the intersection and union ratio is greater than the threshold, it means that these bounding boxes have selected the same target. Deduplication of duplicate bounding boxes can obtain a single bounding box with the highest confidence. Then identify the remaining bounding boxes to determine whether the confidence is greater than the preset confidence threshold. If so, it means that the raw material tube selected by the bounding box is an abnormal raw material tube, and relevant personnel need to be prompted to handle it.

[0066] For example: Input the raw material tube photo into the neural network model for prediction, and you can get five bounding boxes ABCDE, among which the confidence of the C bounding box is the largest. Then first calculate the intersection and union ratio of the combination of the four bounding boxes AC, BC, CD, and CE, where the intersection and union ratio of the three combinations of AC, BC, and CD is greater than the preset threshold, which means that the three bounding boxes ABD and C have the same target, but the target selected by the E bounding box is different from that of the C bounding box. Then first remove the three bounding boxes ABD, and only keep the two bounding boxes CE, and then use the confidence to judge the two bounding boxes CE respectively, where the confidence of the C bounding box is greater than the preset confidence threshold, and the confidence of the E bounding box is less than the confidence threshold, so the C box selects the raw material tube with abnormal state, and E is the target of misjudgment. In this application, since the raw material tube photo is cropped by step S2, in this application, the raw material tube photo usually contains only one target, so it is very likely that there is only one bounding box C in the end, then you only need to judge the confidence of C to get the conclusion of whether the raw material tube is working normally.

[0067] Further, judging whether the three elements of the two-for-one twister are in a normal working state according to the three element photo includes:

[0068] Acquire a number of second training photos for training the model, wherein the second training photos contain ternary elements in a normal working state or ternary elements in an abnormal working state; fill each second training photo respectively to obtain a corresponding second enhanced photo;

[0069] Acquire channel information of each second enhanced photo respectively to obtain a corresponding second three-dimensional tensor; splice all the second three-dimensional tensors to obtain a corresponding second four-dimensional tensor;

[0070] Inputting the second four-dimensional tensor into a pre-established third neural network model for training to obtain a corresponding fourth neural network model;

[0071] The ternary photo is input into the fourth neural network model for prediction to obtain a prediction result, and based on the prediction result, it is determined whether the working status of the ternary in the ternary photo is abnormal.

[0072] Furthermore, the step of inputting the ternary photo into the fourth neural network model for prediction to obtain a prediction result includes:

[0073] Inputting the ternary photo into the fourth neural network model for prediction to obtain a plurality of second boundary boxes for ternary elements with abnormal working status;

[0074] Perform non-maximum suppression processing on several of the second bounding boxes to obtain the corresponding second bounding box with the highest confidence, and determine whether the confidence corresponding to the second bounding box with the highest confidence is greater than a preset confidence threshold. If so, it is determined that there is an abnormality in the working state of the ternary element; if not, it is determined that there is no abnormality in the working state of the ternary element.

[0075] In practical applications, ternary elements are also an important component of the ingot. Therefore, in order to detect ternary elements, another model needs to be trained specifically.

[0076] Specifically, in this application, the first neural network model and the third neural network model both adopt the YOLO V5 network model, and modify the original backbone to the new EMO backbone. A similar Transformer mechanism is introduced with a considerable amount of parameters to improve the model accuracy. Moreover, since EMO is a lightweight network, it has a good balance between model accuracy and efficiency, and its performance is better than EdgeViT, Mobile-former and other networks.

[0077] Further, the determining whether the working state of the wire passing wheel in the wire passing wheel photo is abnormal includes:

[0078] An indicating arrow is arranged on the thread passing wheel of the two-for-one twisting machine; and eight photos of the thread passing wheel are taken at preset time intervals;

[0079] The eight photos of the wire-passing wheel are input into the pre-trained fifth neural network model for detection, the positions of the indicator arrows on the eight wire-passing wheels are identified, and it is determined whether the positions of the indicator arrows on the eight wire-passing wheels are different. If so, it is determined that the wire-passing wheel is in a normal working state. If not, it is determined that the working state of the wire-passing wheel is abnormal.

[0080] In practical applications, such as Figure 6 As shown, the wire wheel is a wheel used to transfer and guide the wire. When the wire passes through the wire wheel, it will drive the wire wheel to rotate. Therefore, an indicator arrow is set on the wire wheel to intuitively observe whether the wire wheel is rotating. Specifically, the shutter speed of the camera and the shooting interval of the camera can be set according to the rotation speed of the wire wheel, and then eight photos of the wire wheel are taken continuously to obtain eight photos of the wire wheel with indicator arrows. Then, the pre-trained neural network model is used to identify the position of the arrow on the photo, and it is determined whether the arrow on the wire wheel rotates, so as to determine whether the wire wheel rotates normally.

[0081] Furthermore, the ingot photo is cut to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo, including:

[0082] Use the pre-set inspection robot to take pictures of the working status of the two-for-one twister to obtain photos of the spindles;

[0083] According to the position of the wire passing wheel of the two-for-one twister in the spindle photo, the spindle photo is cropped to obtain a corresponding first cropped photo; wherein the width of the first cropped photo is the width of the corresponding spindle photo; and the height of the first cropped photo is the height of the corresponding wire passing wheel in the corresponding spindle photo;

[0084] According to a preset size conversion rule, the first cropped photo is resized to obtain a corresponding second cropped photo; target detection is performed on the second cropped photo to obtain first position information of the corresponding wire passing wheel;

[0085] According to the size transformation rule and the first position information, the second position information of the corresponding wire wheel is determined on the first cutting photo; according to the second position information, the third position information of the three-element of the two-for-one twister and the fourth position information of the raw material tube of the two-for-one twister on the spindle photo are calculated;

[0086] According to the second position information, the corresponding ingot photo is cut to obtain the corresponding wire wheel photo; according to the third position information, the corresponding ingot photo is cut to obtain the corresponding ternary photo; according to the fourth position information, the corresponding ingot photo is cut to obtain the corresponding raw material tube photo.

[0087] In practical applications, in order to automatically monitor the two-for-one twisting machine to solve the problem of easy omission in manual inspection in the prior art, the present application proposes a method for detecting abnormal state of the two-for-one twisting machine. First, in the actual production process, the two-for-one twisting machine of each specific category, the two-for-one twisting spindles installed thereon are usually located at the same height, and the arrangement of the two-for-one twisting spindles has a certain regularity, that is, the spindles on the two-for-one twisting machine are arranged repeatedly. The inventor found that after fixing the height of the camera, in the photos or videos taken, the two-for-one twisting spindles are all at a predetermined height. For example, the photos or videos taken with a high-definition camera have a pixel width and height size of 2880*1960, then the area where the two-for-one twisting machine spindles are located only appears in a certain lateral area, for example, the area where the two-for-one twisting machine spindles are located usually only appears in the middle area, and the width and height size of this area is 2420*1960. Because the setting status of the ternary, wire wheel and raw material tube of each two-for-one twisting spindle is the same, the areas where each ternary, each wire wheel and each raw material tube appear are independent of each other, but aligned with each other. For example, in a photo, if a wire wheel appears in the area belonging to the wire wheel, then in the area of ​​the raw material tube, there must be a raw material tube corresponding to the wire wheel, and there must be a ternary in the area of ​​the ternary, and the lateral positions of the ternary, raw material tube and wire wheel will also remain consistent. For example, when the position of the wire wheel is offset to the right by 10 pixels, the ternary and the raw material tube will also be offset to the right by 10 pixels.

[0088] Therefore, according to the above ideas, in the actual shooting and recognition process, it is necessary to first set up a mobile device that can fix the camera at a specific height. Figure 4As shown, it is an AGV mobile robot, on which a column is set, and a camera is fixedly installed on the column. The height of the camera needs to be adjusted according to the height of the actual spindle, so that when the AGV robot moves horizontally, the camera can place the spindle in the center of the picture. In the process of shooting a video with a camera, as the AGV robot moves continuously, the spindle will slide horizontally from left to right in the picture. In other words, the area where the spindle appears is a long strip area, the width of which is the same as the actual width of the photo, and the height of the photo is the same as the height of the spindle in the photo. Similarly, in the spindle area, the area where the wire wheel on the spindle is located, the area where the raw material tube is located, and the area where the three elements are located are also the same as the actual width of the photo, and their heights are the same as the pixel heights of the wire wheel, the raw material tube, and the three elements in the photo. Therefore, in the actual detection process, in order to identify the working situation, it is necessary to extract the frame image (that is, the spindle photo in this application) from the video, and it is only necessary to cut the frame image into several areas according to the position where the parts appear, corresponding to the small-sized horizontal strip image. By using the corresponding positions of the parts and components to crop the frame image, you can get the corresponding horizontal stripe photos where the wire wheel is located, the horizontal stripe photos where the three-element is located, and the horizontal stripe photos where the raw material tube is located. Then, perform target detection on the horizontal stripe photos where the wire wheel is located to get the corresponding position of the wire wheel. Then, based on the position of the wire wheel and the corresponding position relationship between the wire wheel, the three-element, and the raw material tube, you can get the position information of the raw material tube and the three-element. Crop the photos according to the position information to get the corresponding raw material tube photos, wire wheel photos, and three-element photos.

[0089] Moreover, this method does not need to use a neural network model for detection, and can determine the position of the raw material tube and the position of the three-element according to the position of the wire wheel. Its detection and positioning effect is accurate enough and does not consume too much computer computing power. After determining the position of the three-element and the raw material tube, the area is cut out, and the three-element and the raw material tube are cut out separately, and target detection is performed to determine whether the three-element and the raw material tube are working normally. The size of the detected photo is small and the target is clear, which can effectively reduce the need for computer computing power. Improving the speed of photo detection can also meet the need for photo clarity.

[0090] In summary, the present invention has the following beneficial effects: in this application, the ingot photo is divided into a raw material tube photo, a wire wheel photo and a ternary photo, and the raw material tube photo, the wire wheel photo and the ternary photo are detected and judged respectively. Compared with the overall ingot photo, the number of pixels is greatly reduced, so the computing power requirement for target detection can be effectively reduced. And in this application, since the abnormal conditions of the raw material tube, the ternary and the wire wheel are different, it is necessary to judge separately according to the actual situation of each component. When any component is found to be abnormal, an early warning signal is issued accordingly, so that the relevant staff can repair and maintain the equipment in time. And compared with manual inspection to judge the working status of the ingot, the method of this application has a low missed detection rate, small computing power requirement and higher inspection efficiency.

[0091] like Figure 2 As shown, the present invention also provides a system for detecting abnormal state of a two-for-one twisting machine, comprising:

[0092] Photo shooting module: used to take photos of the spindles on the two-for-one twister to obtain corresponding spindle photos;

[0093] Photo cutting module: used to cut the ingot photo to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo;

[0094] Status judgment module: according to the raw material tube photo, the thread wheel photo and the ternary photo, it is judged whether the raw material tube, the thread wheel and the ternary of the two-for-one twister are all in normal working state. If not, an early warning signal is issued.

[0095] The specific definition of a two-for-one twister abnormal state detection system can be found in the above definition of a two-for-one twister abnormal state detection method, which will not be repeated here. Each module in the above-mentioned two-for-one twister abnormal state detection system can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0096] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. When the computer program is executed by the processor, a method for detecting abnormal state of a twisting machine is implemented.

[0097] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0098] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0099] S1. Take a photo of the spindle on the two-for-one twister to obtain a corresponding spindle photo;

[0100] S2. Cut the ingot photo to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo;

[0101] S3. According to the raw material tube photo, the thread passing wheel photo and the ternary photo, it is determined whether the raw material tube, the thread passing wheel and the ternary of the two-for-one twister are all in normal working state. If not, an early warning signal is issued.

[0102] In one embodiment, judging whether the raw material tube of the two-for-one twister is in a normal working state according to the raw material tube photo includes:

[0103] Acquire a plurality of first training photos for training the model, wherein the first training photos contain raw material tubes in a normal working state or contain raw material tubes in an abnormal working state; fill each first training photo respectively to obtain a corresponding first enhanced photo;

[0104] Acquire channel information of each first enhanced photo respectively to obtain a corresponding first three-dimensional tensor; splice all the first three-dimensional tensors to obtain a corresponding first four-dimensional tensor;

[0105] Inputting the first four-dimensional tensor into a pre-established first neural network model for training to obtain a corresponding second neural network model;

[0106] The raw material pipe photo is input into the second neural network model for prediction to obtain a prediction result, and based on the prediction result, it is determined whether the working state of the raw material pipe in the raw material pipe photo is abnormal.

[0107] In one embodiment, the step of inputting the raw material tube photo into the second neural network model for prediction to obtain a prediction result also includes: inputting the raw material tube photo into the second neural network model for prediction to obtain a plurality of first bounding boxes for framing raw material tubes with abnormal working status; performing non-maximum suppression processing on the plurality of first bounding boxes to obtain corresponding first bounding boxes with the highest confidence, and determining whether the confidence corresponding to the first bounding box with the highest confidence is greater than a preset confidence threshold; if so, determining that the working status of the raw material tube is abnormal; if not, determining that the working status of the raw material tube is not abnormal.

[0108] In one embodiment, judging whether the three elements of the two-for-one twister are in a normal working state according to the three element photo includes:

[0109] Acquire a number of second training photos for training the model, wherein the second training photos contain ternary elements in a normal working state or ternary elements in an abnormal working state; fill each second training photo respectively to obtain a corresponding second enhanced photo;

[0110] Acquire channel information of each second enhanced photo respectively to obtain a corresponding second three-dimensional tensor; splice all the second three-dimensional tensors to obtain a corresponding second four-dimensional tensor;

[0111] Inputting the second four-dimensional tensor into a pre-established third neural network model for training to obtain a corresponding fourth neural network model;

[0112] The ternary photo is input into the fourth neural network model for prediction to obtain a prediction result, and based on the prediction result, it is determined whether the working status of the ternary in the ternary photo is abnormal.

[0113] In one embodiment, the step of inputting the ternary photo into the fourth neural network model for prediction to obtain a prediction result includes:

[0114] Inputting the ternary photo into the fourth neural network model for prediction to obtain a plurality of second boundary boxes for ternary elements with abnormal working status;

[0115] Perform non-maximum suppression processing on several of the second bounding boxes to obtain the corresponding second bounding box with the highest confidence, and determine whether the confidence corresponding to the second bounding box with the highest confidence is greater than a preset confidence threshold. If so, it is determined that there is an abnormality in the working state of the ternary element; if not, it is determined that there is no abnormality in the working state of the ternary element.

[0116] In one embodiment, determining whether the working state of the wire passing wheel in the wire passing wheel photo is abnormal includes:

[0117] An indicating arrow is arranged on the thread passing wheel of the two-for-one twisting machine; and eight photos of the thread passing wheel are taken at preset time intervals;

[0118] The eight photos of the wire-passing wheel are input into the pre-trained fifth neural network model for detection, the positions of the indicator arrows on the eight wire-passing wheels are identified, and it is determined whether the positions of the indicator arrows on the eight wire-passing wheels are different. If so, it is determined that the wire-passing wheel is in a normal working state. If not, it is determined that the working state of the wire-passing wheel is abnormal.

[0119] In one embodiment, the ingot photo is cut to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo, including:

[0120] Use the pre-set inspection robot to take pictures of the working status of the two-for-one twister to obtain photos of the spindles;

[0121] According to the position of the wire passing wheel of the two-for-one twister in the spindle photo, the spindle photo is cropped to obtain a corresponding first cropped photo; wherein the width of the first cropped photo is the width of the corresponding spindle photo; and the height of the first cropped photo is the height of the corresponding wire passing wheel in the corresponding spindle photo;

[0122] According to a preset size conversion rule, the first cropped photo is resized to obtain a corresponding second cropped photo; target detection is performed on the second cropped photo to obtain first position information of the corresponding wire passing wheel;

[0123] According to the size transformation rule and the first position information, the second position information of the corresponding wire wheel is determined on the first cutting photo; according to the second position information, the third position information of the three-element of the two-for-one twister and the fourth position information of the raw material tube of the two-for-one twister on the spindle photo are calculated;

[0124] According to the second position information, the corresponding ingot photo is cut to obtain the corresponding wire wheel photo; according to the third position information, the corresponding ingot photo is cut to obtain the corresponding ternary photo; according to the fourth position information, the corresponding ingot photo is cut to obtain the corresponding raw material tube photo.

[0125] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0126] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for detecting abnormal state of a two-for-one twisting machine, characterized in that: include: S1. Take a photo of the spindle on the two-for-one twister to obtain a corresponding spindle photo; S2. Cut the ingot photo to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo; S3. According to the raw material tube photo, the thread passing wheel photo and the ternary photo, it is determined whether the raw material tube, the thread passing wheel and the ternary of the two-for-one twister are all in normal working state. If not, an early warning signal is issued.

2. A method for detecting abnormal state of a two-for-one twisting machine according to claim 1, characterized in that: The step of judging whether the raw material tube of the two-for-one twister is in a normal working state according to the raw material tube photo comprises: Acquire a plurality of first training photos for training the model, wherein the first training photos contain raw material tubes in a normal working state or contain raw material tubes in an abnormal working state; fill each first training photo respectively to obtain a corresponding first enhanced photo; Acquire channel information of each first enhanced photo respectively to obtain a corresponding first three-dimensional tensor; splice all the first three-dimensional tensors to obtain a corresponding first four-dimensional tensor; Inputting the first four-dimensional tensor into a pre-established first neural network model for training to obtain a corresponding second neural network model; The raw material pipe photo is input into the second neural network model for prediction to obtain a prediction result, and based on the prediction result, it is determined whether the working state of the raw material pipe in the raw material pipe photo is abnormal.

3. A method for detecting abnormal state of a two-for-one twisting machine according to claim 2, characterized in that: The step of inputting the raw material tube photo into the second neural network model for prediction to obtain a prediction result further includes: Inputting the raw material pipe photo into the second neural network model for prediction to obtain a plurality of first boundary boxes for framing the raw material pipes with abnormal working status; Perform non-maximum suppression processing on several of the first bounding boxes to obtain the corresponding first bounding box with the highest confidence, and determine whether the confidence corresponding to the first bounding box with the highest confidence is greater than a preset confidence threshold. If so, it is determined that there is an abnormality in the working state of the raw material tube; if not, it is determined that there is no abnormality in the working state of the raw material tube.

4. The method for detecting abnormal state of a two-for-one twisting machine according to claim 1, characterized in that: The step of judging whether the three elements of the two-for-one twister are in a normal working state according to the three element photo comprises: Acquire a number of second training photos for training the model, wherein the second training photos contain ternary elements in a normal working state or ternary elements in an abnormal working state; fill each second training photo respectively to obtain a corresponding second enhanced photo; Acquire channel information of each second enhanced photo respectively to obtain a corresponding second three-dimensional tensor; splice all the second three-dimensional tensors to obtain a corresponding second four-dimensional tensor; Inputting the second four-dimensional tensor into a pre-established third neural network model for training to obtain a corresponding fourth neural network model; The ternary photo is input into the fourth neural network model for prediction to obtain a prediction result, and based on the prediction result, it is determined whether the working status of the ternary in the ternary photo is abnormal.

5. A method for detecting abnormal state of a two-for-one twisting machine according to claim 4, characterized in that: The step of inputting the ternary photo into the fourth neural network model for prediction to obtain a prediction result includes: Inputting the ternary photo into the fourth neural network model for prediction to obtain a plurality of second boundary boxes for ternary elements with abnormal working status; Perform non-maximum suppression processing on several of the second bounding boxes to obtain the corresponding second bounding box with the highest confidence, and determine whether the confidence corresponding to the second bounding box with the highest confidence is greater than a preset confidence threshold. If so, it is determined that there is an abnormality in the working state of the ternary element; if not, it is determined that there is no abnormality in the working state of the ternary element.

6. A method for detecting abnormal state of a two-for-one twisting machine according to claim 1, characterized in that: The determining whether the working state of the wire-passing wheel in the wire-passing wheel photo is abnormal includes: An indicating arrow is arranged on the thread passing wheel of the two-for-one twisting machine; and eight photos of the thread passing wheel are taken at preset time intervals; The eight photos of the wire-passing wheel are input into the pre-trained fifth neural network model for detection, the positions of the indicator arrows on the eight wire-passing wheels are identified, and it is determined whether the positions of the indicator arrows on the eight wire-passing wheels are different. If so, it is determined that the wire-passing wheel is in a normal working state. If not, it is determined that the working state of the wire-passing wheel is abnormal.

7. The method for detecting abnormal state of a two-for-one twisting machine according to claim 1, characterized in that: The ingot photo is cut to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo, including: Use the pre-set inspection robot to take pictures of the working status of the two-for-one twister to obtain photos of the spindles; According to the position of the wire passing wheel of the two-for-one twister in the spindle photo, the spindle photo is cropped to obtain a corresponding first cropped photo; wherein the width of the first cropped photo is the width of the corresponding spindle photo; and the height of the first cropped photo is the height of the corresponding wire passing wheel in the corresponding spindle photo; According to a preset size conversion rule, the first cropped photo is resized to obtain a corresponding second cropped photo; target detection is performed on the second cropped photo to obtain first position information of the corresponding wire passing wheel; According to the size transformation rule and the first position information, the second position information of the corresponding wire wheel is determined on the first cutting photo; according to the second position information, the third position information of the three-element of the two-for-one twister and the fourth position information of the raw material tube of the two-for-one twister on the spindle photo are calculated; According to the second position information, the corresponding ingot photo is cut to obtain the corresponding wire wheel photo; according to the third position information, the corresponding ingot photo is cut to obtain the corresponding ternary photo; according to the fourth position information, the corresponding ingot photo is cut to obtain the corresponding raw material tube photo.

8. A two-for-one twister abnormal state detection system, characterized in that: include: Photo shooting module: used to take photos of the spindles on the two-for-one twister to obtain corresponding spindle photos; Photo cutting module: used to cut the ingot photo to obtain the corresponding raw material tube photo, wire wheel photo and ternary photo; Status judgment module: according to the raw material tube photo, the thread wheel photo and the ternary photo, it is judged whether the raw material tube, the thread wheel and the ternary of the two-for-one twister are all in normal working state. If not, an early warning signal is issued.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

Images