Sock production traceability system and method based on big data

Abstract

The invention relates to the technical field of sock production management, in particular to a sock production traceability system and method based on big data. Firstly, a sock production information acquisition unit acquires sock production information of each sock sample; secondly, carrying out quality detection on the sock samples by a sock quality detection unit to obtain a sock quality detection result corresponding to each sock sample; thirdly, a sock abnormity tracing unit performs abnormity tracing according to the sock quality detection result, obtains a defect vector corresponding to each defective sock sample, compares the defect vector with a historical defect vector, and obtains an abnormal equipment number and a corresponding equipment abnormal value; and finally, the sock equipment warning unit warns the abnormal equipment of which the equipment abnormal value is greater than a set threshold value. According to the system, the serial number of the specific defective equipment can be positioned according to the defect vector corresponding to the sock sample and the sock production information, and the accuracy of sock abnormity traceability is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of sock production management, and in particular to a sock production traceability system and method based on big data. Background Art

[0002] As the global economy continues to develop, consumers are increasingly demanding transparency in product quality, origin, and production processes. In particular, in the textile industry, as one of the daily consumer goods, the quality, safety, production process, and environmental protection requirements of socks are attracting more and more attention from consumers and regulators.

[0003] The production process of socks involves multiple links such as spinning, weaving, sealing and shaping. Among them, the technological level of the three links of weaving, sealing and shaping directly determines the appearance, comfort and durability of the socks, and is the key factor affecting the quality of the final product. Therefore, it is necessary to trace the abnormalities of the equipment corresponding to the three links of weaving, sealing and shaping based on the socks with abnormalities produced.

[0004] Most existing sock production traceability methods rely on local monitoring systems, which are difficult to comprehensively monitor the entire production process. In addition, the correlation between equipment failure and quality problems is complex and hidden. Traditional methods find it difficult to effectively extract valid information from massive production data, resulting in inaccurate fault location and reduced accuracy of sock production traceability.

[0005] Therefore, a socks production traceability system and method based on big data is proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a sock production traceability system and method based on big data. First, a sock production information acquisition unit acquires the sock production information of each sock sample; secondly, a sock quality inspection unit performs quality inspection on the sock samples to obtain the sock quality inspection results corresponding to each sock sample; thirdly, a sock abnormality tracing unit performs abnormality tracing based on the sock quality inspection results, obtains the defect vector corresponding to each defective sock sample, and compares it with the historical defect vector to obtain the abnormal device number and the corresponding device abnormal value; finally, a sock device warning unit warns the abnormal device whose device abnormal value is greater than the set threshold.

[0007] To achieve the above object, the present invention provides the following technical solutions: A socks production traceability system based on big data, including: Socks sample acquisition unit: used to acquire multiple socks samples; A socks production information acquisition unit: used to acquire the socks production information of each of the socks samples; Socks quality inspection unit: used to perform quality inspection on the socks samples and obtain the socks quality inspection result corresponding to each socks sample; Socks abnormality tracing unit: used to perform abnormality tracing according to the socks quality test results, and obtain the abnormal device number and the corresponding device abnormal value; the process of obtaining the abnormal device number includes: obtaining socks samples that fail the socks quality test and recording them as defective socks samples; obtaining the defect vector corresponding to each of the defective socks samples; Acquire a historical abnormal socks production data set, wherein the historical abnormal socks production data set includes a plurality of historical abnormal socks data; the historical abnormal socks data includes historical defect vectors of socks that fail a socks quality inspection and corresponding abnormal equipment types; the number of abnormal equipment types corresponding to each defect vector is not less than 1; Calculate the similarity between the defect vector and each of the historical defect vectors in the historical abnormal sock production data set; obtain all abnormal equipment types corresponding to the historical defect vectors whose similarity is greater than a set similarity threshold; obtain the abnormal equipment number corresponding to each of the defective sock samples based on the abnormal equipment type and the sock production information; Sock equipment warning unit: used to warn abnormal equipment whose abnormal value is greater than the set threshold.

[0008] Furthermore, the socks samples include: the production process of each of the socks samples includes a knitting process, a sewing process and a shaping process.

[0009] Furthermore, the sock production information includes sock knitting equipment information, sock sewing equipment information, sock shaping equipment information and sock types; the equipment information includes equipment type and equipment number.

[0010] Further, the socks quality detection unit includes: a socks appearance detection module, a socks air permeability detection module, a socks hygroscopicity detection module, a socks elasticity detection module and a socks quality output module; The sock appearance detection module is used to collect the sock surface image of the sock sample, and obtain the sock appearance detection result based on the sock surface image; the sock air permeability detection module, the sock hygroscopicity detection module and the sock elasticity detection module are used to perform air permeability detection, hygroscopicity detection and elasticity detection on the sock sample respectively, and obtain the air permeability detection result, hygroscopicity detection result and elasticity detection result; the sock quality output module is used to obtain the sock quality detection result based on the detection results output by the four detection modules; the sock quality detection result includes the sock quality detection qualified and the sock quality detection unqualified; when the detection results output by the four detection modules are all qualified, the sock quality detection result is the sock quality detection qualified; otherwise, the sock quality detection result is the sock quality detection unqualified.

[0011] Further, the sock appearance detection module includes: the sock surface image includes an outer sock surface image and an inner sock surface image; the outer sock surface image and the corresponding sock type are input into the sock appearance detection model to obtain a sock outer side detection result; after collecting the outer sock surface image of the sock sample, collect the inner sock surface image of the sock sample; input the inner sock surface image and the corresponding sock type into the sock appearance detection model to obtain a sock inner side detection result; when both the outer sock detection result and the inner sock detection result are free of defects, the sock appearance detection result corresponding to the sock sample is qualified for the sock appearance detection; otherwise, the sock appearance detection result corresponding to the sock sample is unqualified for the sock appearance detection.

[0012] Furthermore, the socks appearance detection model includes a socks surface image input layer, a socks surface image segmentation layer, a socks surface image local feature extraction layer, a socks surface image local feature fusion layer, a socks surface image defect classification layer and a socks detection result output layer: The sock surface image input layer is used to input the surface image and the corresponding sock type into the sock appearance detection model; the sock surface image segmentation layer is used to perform semantic segmentation on the sock surface image to obtain the sock toe area image, the sock body area image, the sock heel area image and the cuff area image; The local feature extraction layer of the sock surface image is used to extract feature maps of the four regions segmented by the sock surface image segmentation layer, respectively, to obtain local feature maps corresponding to the four regions; The local feature fusion layer of the sock surface image is used to perform feature fusion on the local feature map corresponding to each area to obtain the local features of the socks corresponding to each area; the local features of the socks include sock texture features and sock color features; The sock surface image defect classification layer is used to obtain the appearance defect type corresponding to each area based on the local features of the socks and the corresponding sock types; the sock detection result output layer is used to output the sock detection results corresponding to the sock surface image; when the sock surface image defect classification layer obtains that each area corresponding to the sock surface image does not have appearance defects, the sock detection result output layer outputs the sock detection result as no defect; otherwise, the sock detection result output layer outputs the area with appearance defects and the corresponding appearance defect type.

[0013] Furthermore, the defect vector includes: generating a defect vector corresponding to each defective sock sample according to the data obtained by the sock quality detection unit; the dimension of the defect vector is N+3; N represents the total number of appearance defect types; when When , the value corresponding to the i-th dimension defect vector is the indicator value of whether the defective socks sample has the i-th type appearance defect; when the defective socks sample has the i-th type appearance defect, the corresponding indicator value is 1; otherwise, the corresponding indicator value is 0; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the air permeability test of the defective socks sample is qualified; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the moisture absorption test of the defective socks sample is qualified; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the elasticity test of the defective socks sample is qualified; when the test is unqualified, the corresponding indication value is 1; otherwise, the corresponding indication value is 0.

[0014] Furthermore, the process of obtaining the device abnormal value includes: obtaining the abnormal device number corresponding to each of the defective socks samples to generate an abnormal device number set; the number of abnormal device numbers in the abnormal device number set is recorded as M; the process of obtaining the device abnormal value corresponding to the i-th abnormal device number includes: Obtaining a sock sample having the jth abnormal device number in the sock production information as the first sock sample; The number of the first sock samples is obtained and recorded as the first number; the number of defective sock samples with the jth abnormal device number in the abnormal devices corresponding to all defective sock samples is obtained and recorded as the second number; based on the first number and the second number, the device abnormality value of the jth abnormal device number is obtained, and the calculation formula is: ; in, The device abnormality value represented by the jth abnormal device number; the value of j ranges from 1 to M; and are respectively represented as the first quantity and the second quantity corresponding to the j-th abnormal device; and They are respectively expressed as the abnormal proportion coefficient of socks and the abnormal quantity coefficient of socks; Expressed as an exponential function with a natural constant as base.

[0015] Acquire multiple hosiery samples; acquire hosiery production information of each of the hosiery samples; Performing quality inspection on the sock samples to obtain a sock quality inspection result corresponding to each sock sample; According to the sock quality inspection result, abnormality tracing is performed to obtain the abnormal device number and the corresponding device abnormal value; the process of obtaining the abnormal device number includes: obtaining sock samples that fail the sock quality inspection and recording them as defective sock samples; obtaining the defect vector corresponding to each defective sock sample; Acquire a historical abnormal socks production data set, wherein the historical abnormal socks production data set includes a plurality of historical abnormal socks data; the historical abnormal socks data includes historical defect vectors of socks that fail a socks quality inspection and corresponding abnormal equipment types; the number of abnormal equipment types corresponding to each defect vector is not less than 1; Calculate the similarity between the defect vector and each of the historical defect vectors in the historical abnormal sock production data set; obtain all abnormal equipment types corresponding to the historical defect vectors whose similarity is greater than a set similarity threshold; obtain the abnormal equipment number corresponding to each of the defective sock samples based on the abnormal equipment type and the sock production information; Warning will be given to abnormal devices whose abnormal values ​​are greater than the set threshold.

[0016] Furthermore, the socks samples include: the production process of each of the socks samples includes a knitting process, a sewing process and a shaping process.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention adopts a sock appearance detection model to perform appearance defect detection on sock samples; the sock appearance detection model first segments the sock surface image to obtain different areas, and then obtains local features of the socks in different areas. This method performs defect identification based on the features of different areas, which can make the model more focused on the local features of each area, thereby improving the accuracy of defect identification.

[0018] 2. The present invention generates a defect vector based on the data obtained by the sock quality inspection unit, and compares the defect vector with the historical defect vector to obtain the corresponding abnormal device type. This method can identify the abnormal device type corresponding to the historical defect vector with high similarity to the defect vector corresponding to the sock sample based on the historical data, and locate the specific device number in combination with the sock production information corresponding to the sock sample, thereby improving the accuracy of sock abnormality tracing.

[0019] 3. The present invention obtains abnormal values ​​corresponding to abnormal devices with different numbers based on the number of abnormal devices with different numbers that appear in abnormal sock samples and the proportion of abnormal sock samples corresponding to the abnormal devices. The abnormal proportion is an important indicator reflecting the degree of abnormality of the equipment. Combined with the quantity, the abnormality of the equipment can be measured more comprehensively. This method can more accurately identify which equipment has a higher abnormal risk, and then carry out targeted equipment maintenance or optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A structural diagram of a socks production traceability system based on big data provided by an embodiment of the present invention; Figure 2A flow chart of a sock production traceability method based on big data provided by an embodiment of the present invention; Figure 3 A structural diagram of a sample of socks provided in an embodiment of the present invention; Figure 4 This is a structural diagram of a sock quality detection unit provided in an embodiment of the present invention.

[0021] In the figure: 1. Toe area; 2. Body area; 3. Heel area; 4. Cuff area. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of 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.

[0023] The present invention provides a sock production traceability system based on big data. This system is applied to a sock production traceability method based on big data. For specific system structure diagram and method flow chart, please refer to Figure 1 and Figure 2 .

[0024] Embodiment 1 Reference Figure 1 The sock sample acquisition unit in the invention is applied to S10 of a sock production traceability method based on big data.

[0025] Furthermore, in this embodiment, 300 sock samples produced by factory A are collected, and the sock type corresponding to the 300 sock samples is type A; Figure 3 , each sock sample includes a toe area 1, a body area 2, a heel area 3 and a cuff area 4; the sock production equipment numbers corresponding to each production process of the 300 sock samples are not completely the same; the production process of each sock sample includes a knitting process, a sewing process and a shaping process; the number of sock samples corresponding to each sock production equipment number is not less than a set quantity threshold; in this embodiment, the quantity threshold is set to 50, which can also be modified according to actual applications.

[0026] In this step of the present embodiment, multiple sock samples are selected, and different sock samples are knitted, sewn and shaped, and the number of sock samples corresponding to each sock production equipment number is not less than the set quantity threshold. This method can avoid the impact of a single sock sample and missing production process on abnormal traceability of socks, thereby improving the reliability and accuracy of the traceability results.

[0027] Reference Figure 1 The socks production information acquisition unit is applied to S20 of a socks production traceability method based on big data.

[0028] Furthermore, the sock production information acquisition unit includes: the sock production information includes sock knitting information, sock seam information, sock shaping information and sock type; the sock knitting information includes sock knitting yarn information, sock knitting equipment information and equipment knitting parameters; the sock seam information includes sock seam yarn information, sock seam equipment information and equipment seam parameters; the sock shaping information includes sock shaping equipment information and equipment shaping parameters; the equipment information includes equipment type and equipment number.

[0029] This step of the present embodiment obtains the sock production information corresponding to each sock sample, and the sock production information includes sock knitting information, sock seam information, sock shaping information and sock type. This method can provide a data basis for the subsequent tracing of sock abnormalities, so as to improve the accuracy of tracing of sock abnormalities.

[0030] Reference Figure 1 The socks quality detection unit in the invention is applied to S30 of a socks production traceability method based on big data.

[0031] Further, refer to Figure 4 , the socks quality detection unit includes: a socks appearance detection module, a socks air permeability detection module, a socks hygroscopicity detection module, a socks elasticity detection module and a socks quality output module; The sock appearance detection module is used to collect the sock surface image of the sock sample, and obtain the sock appearance detection result based on the sock surface image; the sock air permeability detection module, the sock hygroscopicity detection module and the sock elasticity detection module are used to perform air permeability detection, hygroscopicity detection and elasticity detection on the sock sample respectively, and obtain the air permeability detection result, hygroscopicity detection result and elasticity detection result; the sock quality output module is used to obtain the sock quality detection result based on the detection results output by the four detection modules; the sock quality detection result includes the sock quality detection qualified and the sock quality detection unqualified; when the detection results output by the four detection modules are all qualified, the sock quality detection result is the sock quality detection qualified; otherwise, the sock quality detection result is the sock quality detection unqualified.

[0032] In this step of the present embodiment, each sock sample is subjected to appearance testing, air permeability testing, hygroscopicity testing, and elasticity testing, and the sock quality testing results are obtained based on the test results corresponding to the four tests. This method can test all the properties of the sock samples to accurately identify sock samples with unqualified quality, thereby improving the accuracy of abnormal sock tracing.

[0033] Further, the sock appearance detection module includes: the sock surface image includes an outer sock surface image and an inner sock surface image; Inputting the outer sock surface image into a sock appearance detection model to obtain a sock outer side detection result; After acquiring the outer sock surface image of the sock sample, acquiring the inner sock surface image of the sock sample; Inputting the inner side sock surface image into the sock appearance detection model again to obtain the sock inner side detection result; When both the outer socks inspection results and the inner socks inspection results are free of defects, the socks appearance inspection result corresponding to the socks sample is qualified; otherwise, the socks appearance inspection result corresponding to the socks sample is unqualified.

[0034] When the sock shaping equipment is abnormal, it will cause appearance defects such as wrinkles on the outside of the socks. In this step of the present embodiment, the outer sock image is first detected and then the inner sock image is detected. This method can achieve comprehensive sock appearance detection and, in the order of first detecting the outer side and then detecting the inner side, ensure that the outer side image of the socks is not affected by the folding of the sock sample, thereby improving the accuracy of sock appearance defect detection.

[0035] Furthermore, the socks appearance detection model includes a socks surface image input layer, a socks surface image segmentation layer, a socks surface image local feature extraction layer, a socks surface image local feature fusion layer, a socks surface image defect classification layer and a socks detection result output layer: The sock surface image input layer is used to input the surface image and the corresponding sock type into the sock appearance detection model; the surface image has an image label; the image label includes an outer sock surface image and an outer sock surface image; the sock surface image segmentation layer is used to perform semantic segmentation on the sock surface image to obtain a sock toe area image, a sock body area image, a sock heel area image and a cuff area image; The local feature extraction layer of the sock surface image is used to extract feature maps of the four regions segmented by the sock surface image segmentation layer, respectively, to obtain local feature maps corresponding to the four regions; The local feature fusion layer of the sock surface image is used to perform feature fusion on the local feature map corresponding to each area to obtain the local features of the socks corresponding to each area; the local features of the socks include sock texture features and sock color features; The sock surface image defect classification layer is used to obtain the appearance defect type corresponding to each area according to the local features of the socks; the sock detection result output layer is used to output the sock detection result corresponding to the sock surface image; when the sock surface image defect classification layer obtains that there is no appearance defect in each area corresponding to the sock surface image, the sock detection result output layer outputs the sock detection result as no defect; otherwise, the sock detection result output layer outputs the area with appearance defects and the corresponding appearance defect type; the appearance defect type includes but is not limited to: sock tip seam defects, sock body hole defects and sock body color defects.

[0036] In order to verify the accuracy of the socks appearance detection model provided in this embodiment for identifying socks surface image defects, this embodiment also uses three other models to perform defect detection on four different types of socks samples produced by Factory A, and obtains the accuracy of each model for socks defect recognition; there are 300 socks samples corresponding to each type; the calculation formula for the accuracy of socks defect recognition is: ; in, It is expressed as the accuracy of identifying the a-th type of sock defects; the value of a ranges from 1 to 4; It is represented by the number of hosiery samples of type a that have appearance defects; It is represented by the number of correct appearance defect types identified in the dth area of ​​the bth sock sample with appearance defects; It is represented by the number of appearance defect types actually existing in the dth area of ​​the bth sock sample with appearance defects; the accuracy of each model for sock defect recognition is shown in Table 1.

[0037] Table 1 Socks appearance inspection model defect recognition verification table

[0038] In Table 1, model 1 is the sock appearance detection model provided by this embodiment; model 2 is based on model 1 without extracting the texture features of socks; model 3 is based on model 2 without performing semantic segmentation on socks; model 4 is a convolutional neural network model. It can be seen from Table 1 that the sock appearance detection model provided by this embodiment can accurately identify the appearance defects of different areas of the sock samples.

[0039] In this step of the present embodiment, a sock appearance detection model is used to perform appearance defect detection on sock samples; the sock appearance detection model first segments the sock surface image to obtain different areas, and then obtains local features of the socks in different areas. This method performs defect identification based on the features of different areas, which can make the model more focused on the local features of each area, thereby improving the accuracy of defect identification.

[0040] Furthermore, the process of the socks air permeability detection module performing air permeability detection on the socks sample and obtaining the air permeability detection result includes: using an air permeability tester to obtain the speed of air passing through the socks sample; when the speed is less than a set speed threshold, the air permeability detection result is unqualified; otherwise, the air permeability detection result is qualified; the speed threshold is obtained based on experience.

[0041] Furthermore, the process of the socks hygroscopicity detection module performing hygroscopicity detection on the socks sample and obtaining the hygroscopicity detection result includes: measuring the initial mass of the socks sample; placing the socks sample in a constant temperature and humidity chamber and maintaining it for a set time; after the set time is reached, taking out the socks sample, weighing the mass of the socks sample after absorbing moisture, and calculating the hygroscopicity of the socks sample, and the calculation formula is: ; in, Expressed as the moisture absorption rate of the hosiery sample; and They are respectively expressed as the initial mass and mass after moisture absorption of the hosiery sample; When the moisture absorption rate of the sock sample is less than the set moisture absorption rate threshold, the moisture absorption test result is unqualified; otherwise, the moisture absorption test result is qualified; the moisture absorption rate threshold is obtained based on experience.

[0042] Furthermore, the sock elasticity detection module uses a sock elasticity tester to perform elasticity detection on the sock sample and obtain the elasticity detection result of the sock sample.

[0043] Reference Figure 1 The sock abnormality tracing unit is applied to S40 of a sock production tracing method based on big data.

[0044] Furthermore, the process of obtaining the abnormal device number includes: obtaining sock samples that fail the sock quality inspection and recording them as defective sock samples; obtaining a defect vector corresponding to each of the defective sock samples; Acquire a historical abnormal socks production data set, wherein the historical abnormal socks production data set includes a plurality of historical abnormal socks data; the historical abnormal socks data includes historical defect vectors of socks that fail a socks quality inspection and corresponding abnormal equipment types; the number of abnormal equipment types corresponding to each defect vector is not less than 1; Calculate the similarity between the defect vector and each of the historical defect vectors in the historical abnormal sock production data set; obtain all abnormal equipment types corresponding to the historical defect vectors whose similarity is greater than a set similarity threshold; and obtain the abnormal equipment number corresponding to each of the defective sock samples based on the abnormal equipment type and the sock production information.

[0045] This step of the present embodiment generates a defect vector based on the data obtained by the sock quality inspection unit, and compares the defect vector with the historical defect vector to obtain the corresponding abnormal device type. This method can identify the abnormal device type corresponding to the historical defect vector that has a high similarity with the defect vector corresponding to the sock sample based on the historical data, and locate the number of the specific device in combination with the sock production information corresponding to the sock sample, thereby improving the accuracy of sock abnormality tracing.

[0046] Furthermore, the defect vector includes: generating a defect vector corresponding to each defective sock sample according to the data obtained by the sock quality detection unit; the dimension of the defect vector is N+3; N represents the total number of appearance defect types; when When , the value corresponding to the i-th dimension defect vector is the indicator value of whether the defective socks sample has the i-th type appearance defect; when the defective socks sample has the i-th type appearance defect, the corresponding indicator value is 1; otherwise, the corresponding indicator value is 0; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the air permeability test of the defective socks sample is qualified; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the moisture absorption test of the defective socks sample is qualified; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the elasticity test of the defective socks sample is qualified; when the test is unqualified, the corresponding indication value is 1; otherwise, the corresponding indication value is 0.

[0047] This step of the present implementation assigns corresponding marked values ​​based on the test results of the sock appearance test, elasticity test, air permeability test and hygroscopicity test obtained by the sock quality inspection unit, and obtains the defect vector corresponding to the sock sample. This method can effectively quantify various defects of socks by assigning marked values ​​to different quality indicators, and provide a data basis for tracing the abnormality of socks.

[0048] Furthermore, the process of obtaining the device abnormal value includes: obtaining the abnormal device number corresponding to each of the defective socks samples to generate an abnormal device number set; the number of abnormal device numbers in the abnormal device number set is recorded as M; the process of obtaining the device abnormal value corresponding to the i-th abnormal device number includes: Obtaining a sock sample having the jth abnormal device number in the sock production information as the first sock sample; The number of the first sock samples is obtained and recorded as the first number; the number of defective sock samples with the jth abnormal device number in the abnormal devices corresponding to all defective sock samples is obtained and recorded as the second number; based on the first number and the second number, the device abnormality value of the jth abnormal device number is obtained, and the calculation formula is: ; in, The device abnormality value represented by the jth abnormal device number; the value of j ranges from 1 to M; and are respectively represented as the first quantity and the second quantity corresponding to the j-th abnormal device; and They are respectively expressed as the abnormal proportion coefficient of socks and the abnormal quantity coefficient of socks; Expressed as an exponential function with a natural constant as base.

[0049] This step of the present implementation obtains the abnormal values ​​corresponding to the abnormal devices with different numbers based on the number of abnormal devices with different numbers that appear in the abnormal sock samples and the abnormal sock samples ratio corresponding to the abnormal devices. The abnormal ratio is an important indicator reflecting the degree of abnormality of the equipment. Combined with the quantity, the abnormality of the equipment can be measured more comprehensively. This method can more accurately identify which equipment has a higher abnormal risk, and then carry out targeted equipment maintenance or optimization.

[0050] Reference Figure 1 The socks equipment warning unit is applied to S50 of a socks production traceability method based on big data.

[0051] Furthermore, an abnormal device whose abnormal value is greater than a set threshold is warned, and the threshold is obtained based on experience.

[0052] Embodiment 2 In the first embodiment, the method of the present invention is combined with the system to realize the abnormal tracing of socks; in the second embodiment of the present application, the method of the present invention will be explained again.

[0053] S10. Obtain multiple sock samples.

[0054] Furthermore, the socks samples include: the production process of each of the socks samples includes a knitting process, a sewing process and a shaping process.

[0055] S20. Acquire the sock production information of each of the sock samples.

[0056] Furthermore, the sock production information includes sock knitting equipment information, sock sewing equipment information, sock shaping equipment information and sock types; the equipment information includes equipment type and equipment number.

[0057] S30: Perform quality inspection on the sock samples to obtain a sock quality inspection result corresponding to each sock sample.

[0058] Furthermore, the quality inspection includes appearance inspection, air permeability inspection, hygroscopicity inspection, elasticity inspection and sock quality output of the socks; the sock quality inspection result is obtained based on the inspection results corresponding to the four inspections; the sock quality inspection result includes qualified sock quality inspection and unqualified sock quality inspection; when the inspection results corresponding to the four inspections are all qualified, the sock quality inspection result is qualified sock quality inspection; otherwise, the sock quality inspection result is unqualified sock quality inspection.

[0059] Furthermore, the sock appearance detection includes: collecting a sock surface image of a sock sample, and obtaining a sock appearance detection result based on the sock surface image; the sock surface image includes an outer sock surface image and an inner sock surface image; inputting the outer sock surface image and the corresponding sock type into a sock appearance detection model to obtain a sock outer side detection result; after collecting the outer sock surface image of the sock sample, collecting an inner sock surface image of the sock sample; inputting the inner sock surface image and the corresponding sock type into a sock appearance detection model to obtain a sock inner side detection result; when both the outer sock detection result and the inner sock detection result are free of defects, the sock appearance detection result corresponding to the sock sample is qualified for the sock appearance detection; otherwise, the sock appearance detection result corresponding to the sock sample is unqualified for the sock appearance detection.

[0060] Furthermore, the socks appearance detection model includes a socks surface image input layer, a socks surface image segmentation layer, a socks surface image local feature extraction layer, a socks surface image local feature fusion layer, a socks surface image defect classification layer and a socks detection result output layer: The sock surface image input layer is used to input the surface image and the corresponding sock type into the sock appearance detection model; the sock surface image segmentation layer is used to perform semantic segmentation on the sock surface image to obtain the sock toe area image, the sock body area image, the sock heel area image and the cuff area image; The local feature extraction layer of the sock surface image is used to extract feature maps of the four regions segmented by the sock surface image segmentation layer, respectively, to obtain local feature maps corresponding to the four regions; The local feature fusion layer of the sock surface image is used to perform feature fusion on the local feature map corresponding to each area to obtain the local features of the socks corresponding to each area; the local features of the socks include sock texture features and sock color features; The sock surface image defect classification layer is used to obtain the appearance defect type corresponding to each area based on the local features of the socks and the corresponding sock types; the sock detection result output layer is used to output the sock detection results corresponding to the sock surface image; when the sock surface image defect classification layer obtains that each area corresponding to the sock surface image does not have appearance defects, the sock detection result output layer outputs the sock detection result as no defect; otherwise, the sock detection result output layer outputs the area with appearance defects and the corresponding appearance defect type.

[0061] S40: perform abnormality tracing based on the sock quality inspection result to obtain the abnormal device number and the corresponding device abnormal value.

[0062] Furthermore, the process of obtaining the abnormal device number includes: obtaining sock samples that fail the sock quality inspection and recording them as defective sock samples; obtaining a defect vector corresponding to each of the defective sock samples; Acquire a historical abnormal socks production data set, wherein the historical abnormal socks production data set includes a plurality of historical abnormal socks data; the historical abnormal socks data includes historical defect vectors of socks that fail a socks quality inspection and corresponding abnormal equipment types; the number of abnormal equipment types corresponding to each defect vector is not less than 1; Calculate the similarity between the defect vector and each of the historical defect vectors in the historical abnormal sock production data set; obtain all abnormal equipment types corresponding to the historical defect vectors whose similarity is greater than a set similarity threshold; and obtain the abnormal equipment number corresponding to each of the defective sock samples based on the abnormal equipment type and the sock production information.

[0063] Furthermore, the defect vector includes: generating a defect vector corresponding to each defective sock sample according to the data obtained by the sock quality detection unit; the dimension of the defect vector is N+3; N represents the total number of appearance defect types; when When , the value corresponding to the i-th dimension defect vector is the indicator value of whether the defective socks sample has the i-th type appearance defect; when the defective socks sample has the i-th type appearance defect, the corresponding indicator value is 1; otherwise, the corresponding indicator value is 0; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the air permeability test of the defective socks sample is qualified; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the moisture absorption test of the defective socks sample is qualified; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the elasticity test of the defective socks sample is qualified; when the test is unqualified, the corresponding indication value is 1; otherwise, the corresponding indication value is 0.

[0064] Furthermore, the process of obtaining the device abnormal value includes: obtaining the abnormal device number corresponding to each of the defective socks samples to generate an abnormal device number set; the number of abnormal device numbers in the abnormal device number set is recorded as M; the process of obtaining the device abnormal value corresponding to the i-th abnormal device number includes: Obtaining a sock sample having the jth abnormal device number in the sock production information as the first sock sample; The number of the first sock samples is obtained and recorded as the first number; the number of defective sock samples with the jth abnormal device number in the abnormal devices corresponding to all defective sock samples is obtained and recorded as the second number; based on the first number and the second number, the device abnormality value of the jth abnormal device number is obtained, and the calculation formula is: ; in, The device abnormality value represented by the jth abnormal device number; the value of j ranges from 1 to M; and are respectively represented as the first quantity and the second quantity corresponding to the j-th abnormal device; and They are respectively expressed as the abnormal proportion coefficient of socks and the abnormal quantity coefficient of socks; Expressed as an exponential function with a natural constant as base.

[0065] S50. Alert abnormal devices whose abnormal values ​​are greater than a set threshold.

[0066] Furthermore, an abnormal device whose abnormal value is greater than a set threshold is warned, and the threshold is obtained based on experience.

[0067] Furthermore, in order to verify the effectiveness of the sock production traceability system based on big data provided in this embodiment for tracing the abnormality of socks, this embodiment uses the sock production traceability system based on big data to monitor the sock production equipment producing 4 types of socks in Factory B; and collects the accuracy and precision of abnormal equipment warnings within 85 cycles, refer to Table 2 for details.

[0068] Table 2 Socks abnormal equipment warning verification table

[0069] With reference to Table 2, it can be seen that the big data-based sock production traceability method provided in this embodiment has high accuracy and precision in abnormal equipment warnings for socks, that is, this method has high accuracy in tracing abnormal socks.

[0070] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A socks production traceability system based on big data, characterized in that: include: Socks sample acquisition unit: used to acquire multiple socks samples; A socks production information acquisition unit: used to acquire the socks production information of each of the socks samples; Socks quality inspection unit: used to inspect the quality of the socks samples and obtain the socks quality inspection result corresponding to each socks sample; Socks abnormality tracing unit: used to perform abnormality tracing based on the socks quality inspection results, and obtain the abnormal device number and the corresponding device abnormality value; The process of obtaining the abnormal device number includes: obtaining sock samples that fail the sock quality inspection and recording them as defective sock samples; obtaining a defect vector corresponding to each of the defective sock samples; Acquire a historical abnormal socks production data set, wherein the historical abnormal socks production data set includes a plurality of historical abnormal socks data; the historical abnormal socks data includes historical defect vectors of socks that fail a socks quality inspection and corresponding abnormal equipment types; the number of abnormal equipment types corresponding to each defect vector is not less than 1; Calculate the similarity between the defect vector and each of the historical defect vectors in the historical abnormal sock production data set; obtain all abnormal equipment types corresponding to the historical defect vectors whose similarity is greater than a set similarity threshold; obtain the abnormal equipment number corresponding to each of the defective sock samples based on the abnormal equipment type and the sock production information; Socks equipment warning unit: used to warn abnormal equipment whose abnormal value is greater than the set threshold.

2. The big data-based sock production traceability system according to claim 1 is characterized in that: The socks samples include: the production process of each of the socks samples includes a knitting process, a sewing process and a shaping process.

3. The big data-based sock production traceability system according to claim 1 is characterized in that: The socks production information acquisition unit includes: the socks production information includes socks knitting equipment information, socks sewing equipment information, socks shaping equipment information and socks types; the equipment information includes equipment type and equipment number.

4. The big data-based sock production traceability system according to claim 1 is characterized in that: The socks quality detection unit includes: a socks appearance detection module, a socks air permeability detection module, a socks hygroscopicity detection module, a socks elasticity detection module and a socks quality output module; The sock appearance detection module is used to collect the sock surface image of the sock sample, and obtain the sock appearance detection result based on the sock surface image; the sock air permeability detection module, the sock hygroscopicity detection module and the sock elasticity detection module are used to perform air permeability detection, hygroscopicity detection and elasticity detection on the sock sample respectively, and obtain the air permeability detection result, hygroscopicity detection result and elasticity detection result; the sock quality output module is used to obtain the sock quality detection result based on the detection results output by the four detection modules; the sock quality detection result includes the sock quality detection qualified and the sock quality detection unqualified; when the detection results output by the four detection modules are all qualified, the sock quality detection result is the sock quality detection qualified; otherwise, the sock quality detection result is the sock quality detection unqualified.

5. The big data-based sock production traceability system according to claim 4 is characterized in that: The sock appearance detection module includes: the sock surface image includes an outer sock surface image and an inner sock surface image; Inputting the outer sock surface image and the corresponding sock type into a sock appearance detection model to obtain a sock outer side detection result; after collecting the outer sock surface image of the sock sample, collecting the inner sock surface image of the sock sample; The inner side sock surface image and the corresponding sock type are input into the sock appearance detection model to obtain the inner side detection result of the socks; when the outer side sock detection result and the inner side sock detection result are both free of defects, the sock appearance detection result corresponding to the sock sample is qualified; otherwise, the sock appearance detection result corresponding to the sock sample is unqualified.

6. The big data-based sock production traceability system according to claim 5 is characterized in that: The socks appearance detection model includes a socks surface image input layer, a socks surface image segmentation layer, a socks surface image local feature extraction layer, a socks surface image local feature fusion layer, a socks surface image defect classification layer and a socks detection result output layer: The sock surface image input layer is used to input the surface image and the corresponding sock type into the sock appearance detection model; the sock surface image segmentation layer is used to perform semantic segmentation on the sock surface image to obtain the sock toe area image, the sock body area image, the sock heel area image and the cuff area image; The sock surface image local feature extraction layer is used to extract feature maps of the four regions segmented by the sock surface image segmentation layer, respectively, to obtain local feature maps corresponding to the four regions; The local feature fusion layer of the sock surface image is used to perform feature fusion on the local feature map corresponding to each area to obtain the local features of the socks corresponding to each area; the local features of the socks include sock texture features and sock color features; The sock surface image defect classification layer is used to obtain the appearance defect type corresponding to each area based on the local features of the socks and the corresponding sock types; the sock detection result output layer is used to output the sock detection results corresponding to the sock surface image; when the sock surface image defect classification layer obtains that each area corresponding to the sock surface image does not have appearance defects, the sock detection result output layer outputs the sock detection result as no defect; otherwise, the sock detection result output layer outputs the area with appearance defects and the corresponding appearance defect type.

7. The big data-based sock production traceability system according to claim 1 is characterized in that: The defect vector includes: generating a defect vector corresponding to each defective sock sample according to the data obtained by the sock quality detection unit; the dimension of the defect vector is N+3; N represents the total number of appearance defect types; when When , the value corresponding to the i-th dimension defect vector is the indicator value of whether the defective socks sample has the i-th type appearance defect; when the defective socks sample has the i-th type appearance defect, the corresponding indicator value is 1; otherwise, the corresponding indicator value is 0; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the air permeability test of the defective socks sample is qualified; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the moisture absorption test of the defective socks sample is qualified; when When , the value corresponding to the i-th dimension defect vector is the indication value of whether the elasticity test of the defective socks sample is qualified; when the test is unqualified, the corresponding indication value is 1; otherwise, the corresponding indication value is 0.

8. The big data-based sock production traceability system according to claim 1 is characterized in that: The process of obtaining the device abnormal value includes: obtaining the abnormal device number corresponding to each of the defective socks samples to generate an abnormal device number set; the number of abnormal device numbers in the abnormal device number set is recorded as M; the process of obtaining the device abnormal value corresponding to the i-th abnormal device number includes: obtaining the socks sample with the j-th abnormal device number in the socks production information and recording it as the first socks sample; obtaining the number of the first socks samples and recording it as the first number; obtaining the number of defective socks samples with the j-th abnormal device number in the abnormal devices corresponding to all defective socks samples and recording it as the second number; according to the first number and the second number, obtaining the device abnormal value of the j-th abnormal device number, the calculation formula is: ； in, The device abnormality value represented by the jth abnormal device number; the value of j ranges from 1 to M; and are respectively represented as the first quantity and the second quantity corresponding to the j-th abnormal device; and They are respectively expressed as the abnormal proportion coefficient of socks and the abnormal quantity coefficient of socks; Expressed as an exponential function with a natural constant as base.

9. A method for tracing the production of socks based on big data, characterized in that: include: Obtain multiple hosiery samples; Obtaining the sock production information of each of the sock samples; Performing quality inspection on the sock samples to obtain a sock quality inspection result corresponding to each sock sample; According to the socks quality inspection results, abnormality tracing is performed to obtain the abnormal device number and the corresponding device abnormal value; The process of obtaining the abnormal device number includes: obtaining sock samples that fail the sock quality inspection and recording them as defective sock samples; obtaining a defect vector corresponding to each of the defective sock samples; Acquire a historical abnormal socks production data set, wherein the historical abnormal socks production data set includes a plurality of historical abnormal socks data; the historical abnormal socks data includes historical defect vectors of socks that fail a socks quality inspection and corresponding abnormal equipment types; the number of abnormal equipment types corresponding to each defect vector is not less than 1; Calculate the similarity between the defect vector and each of the historical defect vectors in the historical abnormal sock production data set; obtain all abnormal equipment types corresponding to the historical defect vectors whose similarity is greater than a set similarity threshold; obtain the abnormal equipment number corresponding to each of the defective sock samples based on the abnormal equipment type and the sock production information; Warning will be given to abnormal devices whose abnormal values ​​are greater than the set threshold.

10. The method for tracing the production of socks based on big data according to claim 9, characterized in that: The sock production information includes sock knitting equipment information, sock sewing equipment information, sock shaping equipment information and sock types; the equipment information includes equipment type and equipment number.

Images