An intelligent inspection method and system for the BOM form of non-standard parts of manufacturing equipment

Abstract

The present application discloses an intelligent inspection method and system for the BOM table of non-standard parts of manufacturing equipment, belonging to the technical field of industrial design. Among them, historical BOM tables and their corresponding two-dimensional drawing data of non-standard parts are collected, and the two-dimensional drawing data of the non-standard parts are subjected to image preprocessing and multi-modal annotation to obtain an annotated data set. A two-stage field information extraction model is constructed, and the two-stage field information extraction model is trained based on the annotated data set. The two-dimensional drawing data of non-standard parts of a new project and the field names to be extracted are input into the two-stage field information extraction model, and the corresponding field information of the two-dimensional drawing of the non-standard parts is output. The extracted field information of the two-dimensional drawing of the non-standard parts is compared and calculated with the field information in its corresponding BOM table, and an inspection document is output. The inspection method and system provided by the present application realize the automatic extraction and comparison of the field information of the BOM table of non-standard parts, thereby greatly improving the efficiency and accuracy of BOM table inspection.

Description

Technical Field

[0001] This application belongs to the technical field of industrial design, and particularly relates to an intelligent inspection method and system for the BOM table of non-standard parts of manufacturing equipment. Background Art

[0002] The BOM table (Bill of Materials) is an indispensable part in the design and production process of manufacturing equipment, and its accuracy directly affects production efficiency and cost control. Traditional BOM table inspection relies on manual item-by-item verification. Especially for non-standard parts, their 2D drawing structures are complex and the fields are diverse, and the following problems are likely to occur in manual inspection:

[0003] 1. Low efficiency: The number of fields for non-standard part BOM table inspection is huge, and manual verification takes a long time.

[0004] 2. High error rate: The drawings of non-standard parts are complex and the BOM table fields are diverse, so it is easy to miss or make mistakes in manual inspection.

[0005] 3. Complicated operation: When manually checking the field information of the BOM table, it is necessary to open the corresponding 2D drawing data at the same time, resulting in complicated inspection operation steps.

[0006] In the prior art, there is already a table field extraction method based on OCR (Optical Character Recognition), but it has the following limitations:

[0007] 1. Low recognition rate for complex table structures: Traditional OCR is difficult to process complex table structures, and the recognition rates of tables and cells are low.

[0008] 2. Noise interference: Interference factors such as lines and shadows in 2D drawings will significantly reduce the recognition accuracy.

[0009] 3. Incomplete field splicing: Multi-line text or fields across cells cannot be effectively spliced.

[0010] Based on the above actual situation, how to achieve accurate inspection of the BOM table has become an urgent problem to be solved. Summary of the Invention

[0011] In order to solve the above problems existing in the prior art, this application provides an intelligent inspection method and system for the BOM table of non-standard parts of manufacturing equipment, which realizes automatic extraction and comparison of the field information of the non-standard part BOM table, thereby greatly improving the efficiency and accuracy of BOM table inspection.

[0012] The technical solution is as follows:

[0013] On the one hand, an intelligent inspection method for the BOM table of non-standard parts of manufacturing equipment is provided, including the following steps:

[0014] Step S1: Collect historical BOM tables and their corresponding non-standard part 2D drawing data, and perform image preprocessing and multi-modal annotation on the non-standard part 2D drawing data to obtain an annotated dataset;

[0015] Step S2: Construct a two-stage field information extraction model, and train the two-stage field information extraction model based on the annotated dataset;

[0016] Step S3: Input the non-standard part 2D drawing data of a new project and the field names to be extracted into the two-stage field information extraction model, and output the corresponding field information of the non-standard part 2D drawing;

[0017] Step S4: Compare and calculate the field information extracted from the non-standard part 2D drawing with the field information in its corresponding BOM table, and output an inspection document.

[0018] Further, the image preprocessing in Step S1 is to first perform grayscale processing and binarization processing on the non-standard part 2D drawing data to reduce redundant information in the drawing, remove isolated noise points, and retain the integrity of the table lines, and then perform normalization processing on the pixels of the processed image to ensure the stability of model training.

[0019] Further, the multi-modal annotation in Step S1 is to use a combination of corner point annotation and text box annotation to annotate the table information in the preprocessed 2D drawing data, including:

[0020] First, use corner point annotation to annotate the corner point data of the table in the non-standard part 2D drawing data for positioning the table position, and then use the text box annotation method to annotate the text content in the table. Among them,

[0021] The corner point data includes: the upper left corner point, the upper right corner point, the lower left corner point, the lower right corner point, the corner points of horizontally merged cells, the corner points of vertically merged cells, the corner points of cross-page table connections, the header corner points, and the footer corner points.

[0022] By adopting multi-modal annotation and using a combination of corner point annotation and text box annotation to annotate the table information in the preprocessed 2D drawing data, accurate annotation of the table and text information in the 2D drawing is achieved. Among them, the corner point annotation adds the corner points of horizontally merged cells, the corner points of vertically merged cells, the corner points of cross-page table connections, the header corner points, and the footer corner points on the basis of the upper left corner point, the upper right corner point, the lower left corner point, and the lower right corner point, which not only solves the problem that traditional table recognition methods cannot accurately locate the complex table structure in non-standard part drawings, but also improves the table recognition rate.

[0023] Further, the two-stage field information extraction model includes:

[0024] The first - stage model, which is a convolutional neural network model with a multi - head attention mechanism, is used to identify tables and cell regions in two - dimensional drawing data;

[0025] The second - stage model, which is a convolutional recurrent neural network model, is used to extract and splice the text content within cells.

[0026] Furthermore, the convolutional neural network model with a multi - head attention mechanism sequentially adds a multi - head attention mechanism layer and a BN layer after the convolutional layer of the ResNet - 50 model, where,

[0027] The multi - head attention mechanism layer includes 4 independent attention heads, which are used to focus on feature extraction of table corner points, cell boundaries, merged regions, and cross - page connection regions;

[0028] The BN layer has a built - in Relu activation function, which is used to normalize the convolutional feature map.

[0029] Furthermore, the convolutional recurrent neural network model adds a convolutional neural network at the input end of the recurrent neural network and a translation layer at the output end, where,

[0030] The convolutional neural network includes multiple convolutional layers, which are used to capture image character features to generate a feature map;

[0031] The recurrent neural network layer uses a bidirectional long short - term memory network, which is used to capture the correlation of character sequences in the feature map;

[0032] The translation layer is used to convert individual characters in the feature map into a field sequence.

[0033] By constructing a two - stage field extraction model, the first - stage model focuses on the key region features of the table through the convolutional neural network model with a multi - head attention mechanism, which is used for the accurate positioning of two - dimensional drawing tables. The second - stage model identifies the text content within cells through the convolutional recurrent neural network model and converts individual characters into a field sequence. Through the two - stage field information extraction model, the table position recognition and positioning of two - dimensional drawing data, the accurate extraction and intelligent splicing of field information are realized, avoiding problems such as low recognition rate of complex tables, incomplete extraction of field information, easy confusion of text information, and difficult splicing of field content in traditional OCR technology.

[0034] Furthermore, the non - standard part two - dimensional drawing field information extracted in step S3 includes a field name and a field content.

[0035] Furthermore, the comparison calculation of the non - standard part two - dimensional drawing field information extracted in step S4 with the field information in its corresponding BOM table is to unify the inspection field name of the BOM table and the two - dimensional drawing field name, and then perform a text comparison on their corresponding field contents.

[0036] Furthermore, the unified processing of the inspection field names in the BOM table and the two-dimensional drawing field names is to first use the BERT model to calculate the semantic similarity values of the field names, replace the two-dimensional drawing field names with the BOM table inspection field names with similar semantics, and then perform text comparison on the corresponding field contents. According to the comparison results, it is determined whether the inspected field information is consistent, and the inconsistent field information is marked. This can not only avoid the problem of diverse expressions of field names, but also significantly improve the comparison efficiency of field information. At the same time, it solves the drawback that the traditional technical means cannot complete the end-to-end intelligent inspection of the data information of the non-standard part BOM table.

[0037] Furthermore, the inspection document output in step S4 marks the inconsistent parts of the two-dimensional drawing field information extracted on the basis of the original non-standard part BOM table.

[0038] On the other hand, a smart inspection system for the BOM table of non-standard parts of manufacturing equipment is provided. The system includes:

[0039] A two-dimensional image processing module, which is used to uniformly convert the two-dimensional drawing data into an image format and perform image preprocessing;

[0040] A data annotation module, which is used to provide corner point annotation tools and rotated text box annotation tools, and support nine corner point types and multi-line text annotation;

[0041] A smart calculation and processing module, which is used to extract the field information from the two-dimensional drawing data to obtain the two-dimensional drawing field information;

[0042] A field comparison module, which is used to perform comparison calculations on the inspection field information in the BOM table and the two-dimensional drawing field information to determine whether the field contents are consistent;

[0043] An output module, which is used to display and output the inspection document of the non-standard part BOM table.

[0044] Compared with the prior art, this technical solution has at least the following technical effects:

[0045] This technical solution realizes the accurate annotation of two-dimensional drawing data information by combining corner point annotation and text box annotation. Through the constructed two-stage field information extraction model, it realizes the recognition and positioning of the table position of two-dimensional drawing data, the accurate extraction and intelligent splicing of field information, avoiding problems such as low recognition rate of complex tables, incomplete extraction of field information, easy confusion of text information, and difficult splicing of field content in traditional OCR technology. At the same time, natural language processing technology is used to compare and calculate the inspection fields of the non-standard part BOM table with the field information of the two-dimensional drawing, determine whether the inspected field information is consistent according to the comparison result, and mark the inconsistent field information, solving the drawback that the traditional technical means cannot complete the end-to-end intelligent inspection of the data information of the non-standard part BOM table.

[0046] In addition, through the combination of image processing technology and deep learning technology, this technical solution not only solves the pain point that the BOM table in the manufacturing equipment industry has long relied on manual verification, significantly improves the inspection efficiency and quality of the non-standard part BOM table, but also provides a new idea for the intelligent inspection of BOM tables in related industries. Brief Description of the Drawings

[0047] The accompanying drawings herein are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0048] Figure 1 It is a flowchart of a method for intelligent inspection of a non-standard part BOM table of a manufacturing equipment provided by a preferred embodiment of the present application;

[0049] Figure 2 It is a schematic structural diagram of a system for intelligent inspection of a non-standard part BOM table of a manufacturing equipment provided by a preferred embodiment of the present application;

[0050] Figure 3 It is a schematic structural diagram of a convolutional neural network model with a multi-head attention mechanism provided by a preferred embodiment of the present application;

[0051] Figure 4 It is a schematic structural diagram of a convolutional recurrent neural network model provided by a preferred embodiment of the present application;

[0052] Figure 5 It is the non-standard part two-dimensional drawing data provided by a preferred embodiment of the present application;

[0053] Figure 6 It is the BOM table of non-standard parts provided by a preferred embodiment of the present application. Detailed Description of the Embodiment

[0054] Exemplary embodiments will be described in detail herein, and examples thereof are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0055] The present application provides a method and system for intelligent inspection of the BOM table of non-standard parts of manufacturing equipment. The method includes collecting historical BOM tables and their corresponding two-dimensional drawing data of non-standard parts, performing image preprocessing and multi-modal annotation on the two-dimensional drawing data of non-standard parts to obtain an annotated data set; constructing a two-stage field information extraction model, and training the two-stage field information extraction model based on the annotated data set; inputting the two-dimensional drawing data of non-standard parts of a new project and the field names to be extracted into the two-stage field information extraction model, and outputting the corresponding field information of the two-dimensional drawing of non-standard parts; comparing and calculating the field information extracted from the two-dimensional drawing of non-standard parts with the field information in its corresponding BOM table, and outputting an inspection document.

[0056] By using multi-modal annotation and combining corner point annotation and text box annotation to annotate the table information in the preprocessed two-dimensional drawing data, it solves the problem that traditional table recognition methods cannot accurately locate complex table structures (such as multi-page tables, merged cells) in non-standard part drawings, and realizes accurate annotation of two-dimensional drawing data information; by constructing a two-stage field extraction model, the first-stage model focuses on key region features through a convolutional neural network model with a multi-head attention mechanism, and the second-stage model maintains the coherence of the character sequence through a convolutional recurrent neural network model. Through the constructed two-stage field information extraction model, the table position recognition and positioning of two-dimensional drawing data, accurate extraction and intelligent splicing of field information are realized, avoiding problems such as low recognition rate of complex tables, incomplete extraction of field information, easy confusion of text information, and difficult splicing of field content in traditional OCR technology. Using the BERT model to calculate the semantic similarity value of field names, replacing the field names in the two-dimensional drawing with the BOM table inspection field names with similar semantics, and then performing text comparison on the corresponding field contents, determining whether the inspected field information is consistent according to the comparison result, and annotating the inconsistent field information, which solves the drawback that traditional technical means cannot complete the end-to-end intelligent inspection of the BOM table data information of non-standard parts.

[0057] In one embodiment, the inspection of the BOM table of a single non-standard part is described as a specific embodiment.

[0058] As Figure 1 shown, a method for intelligent inspection of the BOM table of non-standard parts of manufacturing equipment, the specific steps include:

[0059] Step S1: Collect the historical BOM tables and their corresponding 2D drawing data of non-standard parts, and perform image preprocessing and multi-modal annotation on the 2D drawing data to obtain an annotated dataset.

[0060] First, use Solidworks software to export the 2D drawing data of non-standard parts in the historical BOM tables, uniformly convert the 2D drawing data of non-standard parts in all BOM tables into PNG format images, then use the cvtColor function of OpenCV to convert them into grayscale images, and use the local threshold algorithm to process the grayscale images into binary images to reduce redundant information in the drawings, remove isolated noise points, and retain the integrity of the table lines. Finally, perform normalization processing on the image pixels, and process the size into: 224X224 to ensure the stability of model training.

[0061] Then, perform multi-modal annotation on the preprocessed 2D drawing data. The multi-modal annotation is to annotate the table information in the preprocessed 2D drawing data by combining corner point annotation and text box annotation. The specific steps are as follows: First, annotate the nine corner point data of the upper left corner point, upper right corner point, lower left corner point, lower right corner point, horizontally merged cell corner points, vertically merged cell corner points, cross-page table connection corner points, table header corner points, and table tail corner points in the 2D drawing data respectively, and then use the text box annotation method to annotate the text information of the table fields. Finally, obtain 8000 sets of annotated datasets. The text box annotation is to annotate the text content in the table and supports the annotation of inclined text and multi-line text.

[0062] Step S2: Build a two-stage field information extraction model, and use the annotated dataset to train the two-stage field information extraction model.

[0063] First, build a two-stage field information extraction model. The first stage model of the two-stage field information extraction model is a convolutional neural network model with a multi-head attention mechanism. The backbone network of this convolutional neural network model with a multi-head attention mechanism adopts the ResNet-50 network architecture, and a multi-head attention mechanism layer and a BN layer are sequentially added after the convolutional layer of ResNet-50. The multi-head attention mechanism module contains 4 independent attention heads, which are respectively used to focus on the feature extraction of table corner points, cell boundaries, merged areas, and cross-page connection areas; the BN layer is used to normalize the feature map after convolution. This convolutional neural network model with a multi-head attention mechanism mainly includes a convolutional layer, a multi-head attention mechanism layer, a BN layer, a max pooling layer, a Bottleneck layer, an average pooling layer, and a fully connected layer. The specific network structure of this model is as Figure 3 shown.

[0064] The second model of the two-stage field information extraction model uses a convolutional recurrent neural network model. This model adds a convolutional neural network to the input end of the recurrent neural network and a translation layer to the output end, which are used for the generation of feature maps and the generation of field sequences respectively. Among them, the convolutional neural network contains multiple convolutional layers, which are used to capture the character features of the image and generate feature maps. The recurrent neural network layer uses a bidirectional long short-term memory network to capture the correlation of character sequences in the feature maps. Then, the translation layer at the output end maps the character sequences into continuous fields to generate field sequences. The convolutional recurrent neural network model mainly includes a convolutional neural network, a recurrent neural network, and a translation layer. Its specific network structure is as Figure 4 shown.

[0065] Finally, use the labeled dataset to train the above two-stage field information extraction model and save the trained model.

[0066] S3. Import the 2D drawing data of non-standard parts of the new project and the field names to be extracted into the two-stage field information extraction model, and extract the corresponding field information of the 2D drawing;

[0067] The 2D drawing data of non-standard parts of the new project in this embodiment is as follows Figure 5 shown. The field names to be extracted in this embodiment are: drawing number, quantity, material, surface treatment.

[0068] Import the above data into the trained two-stage field information extraction model. After being calculated and processed by the model, the field information in the 2D drawing is extracted. The specific content of the field information of the 2D drawing extracted in this embodiment is as follows:

[0069] Drawing number: XJ24ST068-02-01-001-A

[0070] Quantity: 7

[0071] Material: AL5052

[0072] Surface treatment: wire drawing

[0073] Step S4: Compare and calculate the field information of the extracted 2D drawing with the field information in its corresponding BOM table, and output an inspection document. As Figure 6 shown in the BOM table of the non-standard part.

[0074] First, use the pre-trained BERT model to calculate the semantic similarity between the inspection field names in the BOM table and the field names in the 2D drawing, and unify the inspection field names in the BOM table and the field names in the 2D drawing according to the semantic similarity value. The unification process in this embodiment is to replace the "drawing number" in the field name of the 2D drawing with the "document number" in the BOM table. The specific content of the field information of the 2D drawing and the inspection field information of the BOM table after the unification process in this embodiment is shown in Table 1.

[0075] Then, according to the field names, the corresponding BOM table inspection field content is compared with the two-dimensional drawing field content in text. The fields with the same text content are determined as consistent fields; otherwise, they are determined as inconsistent. The determination table of the inspection field information of the BOM table and the two-dimensional drawing field information in this embodiment is shown in Table 1.

[0076] Table 1 Field Consistency Determination Table

[0077]

[0078] Note: The drawing number, quantity, material, and surface treatment are field names, and the text information after the colon is the field content.

[0079] Thus, the inspection of the field information of the BOM table for a single non-standard part is completed.

[0080] If it is necessary to batch-check the field information of the BOM tables of non-standard parts, then repeat the above steps S3 and S4 until all the fields of the BOM tables of non-standard parts are completely checked, and an alignment document is output. This alignment document is based on the original BOM table of non-standard parts, and uses the openpyxl function in python to mark the inconsistent fields. The marked content includes the field name and the field content. This alignment document supports visual display and can be downloaded and edited to facilitate the modification of inconsistent field information.

[0081] In one embodiment, an intelligent inspection system for the BOM table of non-standard parts of manufacturing equipment is provided. This system includes:

[0082] A two-dimensional image processing module, which is used to uniformly convert two-dimensional drawing data into an image format and perform image preprocessing;

[0083] A data annotation module, which is used to provide corner point annotation tools and rotated text box annotation tools, and supports nine corner point types and multi-line text annotation;

[0084] An intelligent calculation and processing module, which is used to extract the field information from the two-dimensional drawing data to obtain the two-dimensional drawing field information;

[0085] A field comparison module, which is used to perform comparison calculations on the inspection field information of the BOM table and the two-dimensional drawing field information to determine whether the field content is consistent;

[0086] An output module, which is used to display and output the inspection document of the BOM table of non-standard parts.

[0087] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0088] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. An intelligent inspection method for the BOM table of non-standard parts of manufacturing equipment, characterized in that, Including: Step S1: Collect historical BOM tables and their corresponding non-standard part 2D drawing data, perform image preprocessing and multi-modal annotation on the non-standard part 2D drawing data to obtain an annotation dataset. Among them, multi-modal annotation is to annotate the table information in the preprocessed 2D drawing data by combining corner point annotation and text box annotation; Step S2: Construct a two-stage field information extraction model and train the two-stage field information extraction model based on the annotation dataset; The two-stage field information extraction model includes: The first-stage model: a convolutional neural network model with a multi-head attention mechanism, which sequentially adds a multi-head attention mechanism layer and a BN layer after the convolutional layer of the ResNet-50 model. The multi-head attention mechanism layer includes 4 independent attention heads for focusing on feature extraction of table corner points, cell boundaries, merged regions, and cross-page connection regions; The second-stage model: a convolutional recurrent neural network model, which adds a convolutional neural network at the input end of the recurrent neural network and a translation layer at the output end for extracting and splicing the field sequence of the text content within the cell; Step S3: Input the non-standard part 2D drawing data and the field names to be extracted of the new project into the two-stage field information extraction model, and output the corresponding field information of the non-standard part 2D drawing; Step S4: Compare and calculate the field information of the extracted non-standard part 2D drawing with the field information in its corresponding BOM table. First, use the BERT model to unify the inspection field names and 2D drawing field names of the BOM table, and then perform text comparison on the corresponding field contents to output an inspection document.

2. The intelligent inspection method for the non-standard part BOM form of manufacturing equipment according to claim 1, wherein The image preprocessing in Step S1 is to first perform grayscale processing and binarization processing on the non-standard part 2D drawing data to reduce redundant information in the drawing, remove isolated noise points, and retain the integrity of the table lines. Then, perform normalization processing on the pixels of the processed image to ensure the stability of model training.

3. The intelligent inspection method for the non-standard part BOM form of manufacturing equipment according to claim 2, characterized in that, The multi-modal annotation in Step S1 is to annotate the table information in the preprocessed 2D drawing data by combining corner point annotation and text box annotation, including: First, use corner point annotation to annotate the corner point data of the non-standard part 2D drawing data table for positioning the table position, and then use the text box annotation method to annotate the text content in the table. Among them, The corner point data includes: the upper left corner point, the upper right corner point, the lower left corner point, the lower right corner point, the corner points of horizontally merged cells, the corner points of vertically merged cells, the cross-page table connection corner points, the table header corner points, and the table footer corner points.

4. The intelligent inspection method for the non-standard part BOM form of manufacturing equipment according to claim 1, characterized in that The BN layer incorporates a Relu activation function for normalizing the convolutional feature map.

5. The intelligent inspection method for the non-standard part BOM form of manufacturing equipment according to claim 1, characterized in that, The convolutional recurrent neural network model adds a convolutional neural network at the input end of the recurrent neural network and a translation layer at the output end, where The convolutional neural network includes multiple convolutional layers for capturing image character features to generate a feature map; The recurrent neural network layer uses a bidirectional long short-term memory network for capturing the correlation of character sequences in the feature map; The translation layer is used to convert individual characters in the feature map into a field sequence.

6. The intelligent inspection method for the non-standard part BOM form of manufacturing equipment according to claim 1, characterized in that, The two-dimensional drawing field information of non-standard parts extracted in step S3 includes the field name and the field content.

7. The intelligent inspection method for the non-standard part BOM form of manufacturing equipment according to claim 1, characterized in that, The unification process of the inspection field name of the BOM table and the two-dimensional drawing field name is to calculate the semantic similarity value of the field name using the BERT model, and replace the two-dimensional drawing field name with the inspection field name of the BOM table with similar semantics.

8. The intelligent inspection method for the non-standard part BOM form of manufacturing equipment according to claim 7, characterized in that, The inspection document output in step S4 marks the inconsistent parts of the extracted two-dimensional drawing field information on the basis of the original non-standard part BOM table.

9. An intelligent inspection system for the BOM table of non-standard parts of a manufacturing equipment, which adopts the intelligent inspection method for the BOM table of non-standard parts of a manufacturing equipment according to any one of claims 1-8, is characterized in that, The system includes: A two-dimensional image processing module, which is used to uniformly convert two-dimensional drawing data into an image format and perform image preprocessing; A data annotation module, which is used to provide a corner point annotation tool and a rotated text box annotation tool, and support nine corner point types and multi-line text annotation; An intelligent computing and processing module, which is used to extract the field information in the two-dimensional drawing data to obtain the two-dimensional drawing field information; A field comparison module, which is used to compare and calculate the inspection field information of the BOM table and the two-dimensional drawing field information to determine whether the field content is consistent; An output module, which is used to display and output the inspection document of the non-standard part BOM table.

Images