A PCB quotation parameter extraction method, system, device and medium

Abstract

The application provides a PCB quotation parameter extraction method, system, device and medium, comprising: converting the obtained source file into a picture; dividing the picture into corresponding file types; classifying a user-defined to-be-extracted parameter list to obtain a file source-parameter group-knowledge prompt word dictionary; extracting filter information prompts from the filtered files of the source file; inputting the file source-parameter group-knowledge prompt word dictionary and the filter information prompts into a visual-language model to output a final PCB quotation parameter set.

Description

Technical Field

[0001] This invention relates to the field of PCB pricing technology, and in particular to a method, system, device and medium for extracting PCB pricing parameters. Background Technology

[0002] Timely quotations for PCB intent orders are a crucial factor influencing PCB order conversion. Currently, PCB intent order quotations are primarily based on manual review of intent orders and manual identification of key parameters. This method is inefficient and heavily reliant on expert experience. Its drawbacks include, but are not limited to: 1. Significant efficiency bottleneck: An engineer takes an average of 4 to 6 hours to complete the review of a single project's materials. This slow response time slows down order response and may cause companies to miss business opportunities. 2. High dependence on individual experience: The quality of parameter extraction is directly related to the engineer's professional level and experience. Different engineers may interpret the same design materials differently, leading to inconsistent pricing standards. Once staff turnover occurs, the company may face the risk of experience gaps. 3. Difficulty in controlling quality risks: When manually interpreting drawings and entering parameters, minor errors are amplified at each stage of mass production, potentially leading to batch rework or even customer complaints. 4. Difficulty in collaboration and data silos: Engineering parameters need to be manually and repeatedly entered across different platforms such as quotation and MI data generation, which not only involves repetitive work but also easily leads to data inconsistencies and delayed information updates.

[0003] In existing technologies, optical character recognition (OCR) technology is commonly used to identify PCB pricing parameters. However, this method suffers from extremely low accuracy and inefficiency. Specifically, OCR technology cannot effectively understand diagrams such as overlay diagrams and hole tables, leading to incorrect or missed extractions. After extraction, OCR technology relies on keyword matching, which has extremely poor flexibility. The information in the table requirements and Fabrication Drawing is either very sparse or very dense, and OCR technology performs unpredictably on these documents. Summary of the Invention

[0004] In order to solve the problems existing in the prior art, the present invention aims to provide a method, system, device and medium for extracting PCB quotation parameters, which can realize the automatic and accurate extraction of parameters.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for extracting PCB quotation parameters, including: Convert the obtained source file into an image; Classify the images into their corresponding file types; The user-defined list of parameters to be extracted is categorized to obtain a dictionary of file source, parameter group, and knowledge prompt words; Extract filtering information prompts from the source files that were filtered out. Input the file source, parameter group, knowledge prompt dictionary, and filtered information prompts into the visual-language model, and output the final PCB quotation parameter set.

[0006] As a further improvement of the present invention, the document types include: plain text board making instructions, board making instruction tables, full-page board making requirements tables, overlay diagrams, impedance tables, hole tables, chat and emails, dimension diagrams, panelization diagrams, modification sheets, other text and tables, and other types.

[0007] As a further improvement of the present invention, the step of classifying the user-defined list of parameters to be extracted to obtain a file source-parameter group dictionary includes: The parameters are assigned to their corresponding file types to obtain a file source-parameter group dictionary; For each parameter in the file source-parameter group dictionary, instantiate it according to the predefined domain knowledge hint word template to construct the file source-parameter group-knowledge hint word dictionary.

[0008] As a further improvement of the present invention, the step of extracting filtering information prompts from the source file that has been filtered includes: Extract all text information from the source file that has been filtered out; Extract the text information to obtain the filtered information prompts.

[0009] As a further improvement of the present invention, the step of inputting the document source-parameter group-knowledge prompt word dictionary and filtered information prompt words into the visual-language model and outputting the PCB quotation parameter set includes: Input the file source, parameter group, knowledge prompt dictionary, and filtered information prompts into the visual-language model; For the preliminary PCB quotation parameter set output based on chart information, obtain the coordinate frame of the source file on which the preliminary PCB quotation parameter set is based for inference, as traceability information; For the initial PCB quotation parameter set that relies on textual information for reasoning output, the coordinates of the text are matched from the reasoning process of the initial PCB quotation parameter set as traceability information; The traceability information is combined with the preliminary PCB quotation parameter set to obtain the final PCB quotation parameter set.

[0010] This invention also provides a PCB quotation parameter extraction system for implementing the above-mentioned PCB quotation parameter extraction method, comprising: The file conversion module is used to convert the acquired source files into images; The layout detection and classification module is used to classify images into their corresponding file types; The parameter grouping module by file source is used to group parameters into their corresponding file types, resulting in a file source-parameter group dictionary; The domain knowledge hint word module is used to instantiate each parameter of the file source-parameter group dictionary according to the predefined domain knowledge hint word template, and construct the file source-parameter group-knowledge hint word dictionary; An optical character recognition module is used to identify text information in files that have been filtered out from the source file; The information extraction and summarization module is used to extract and filter out information prompts from text information; A visual-language model is used to obtain a dictionary of file source, parameter group, and knowledge prompt words, filter out information prompt words, and output the final PCB quotation parameter set.

[0011] The present invention also provides one or more computer-readable media having instructions stored thereon, which, when executed by one or more processors, cause the processors to perform the above-described PCB quotation parameter extraction method.

[0012] The present invention also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory is used to store computer programs; and the processor is used to implement the above-mentioned PCB quotation parameter extraction method when executing the program stored in the memory.

[0013] The beneficial effects of this invention are as follows: By converting source files into images and classifying them into corresponding file types, it can accurately adapt to chart-type files such as overlay diagrams and hole tables, solving the problems of optical character recognition technology's inability to effectively understand chart information, easy mis-extraction, and missed extraction; by constructing a dictionary of file source-parameter group-knowledge prompt words based on user-defined parameter classification, and combining it with the input of prompt words extracted from filtered files into a visual-language model, it replaces the keyword matching extraction method of optical character recognition technology, improving the flexibility of parameter extraction, solving the defect of unstable performance of optical character recognition technology for sparse or dense files, realizing automated and accurate parameter extraction, significantly shortening the review time of single-project materials, and breaking through the efficiency bottleneck; it does not rely on keyword matching and human experience, unifies parameter extraction standards, avoids interpretation bias, ensures quotation consistency, and reduces the risk of human input errors. Attached Figure Description

[0014] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0015] Figure 1 This is a flowchart of the PCB quotation parameter extraction method described in this invention; Figure 2This is a schematic diagram of the PCB quotation parameter extraction system described in this invention. Detailed Implementation

[0016] To make the technical problems solved by the present invention, the technical solutions adopted, and the technical effects achieved clearer, the technical solutions of the embodiments of the present invention will be further described in detail below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] This invention provides a method for extracting PCB pricing parameters, such as... Figure 1 As shown, it includes: S1. Convert the acquired source files into images; input source files include: original PCB design documents (PDF, scanned copies, images, etc.), board manufacturing requirements, chat logs, Fabrication Drawing files converted to SVG, etc. All input source files are uniformly converted to PNG image format.

[0018] S2. Classify the images into the corresponding file types; file types include: plain text board making instructions, board making instruction tables, full-page board making requirements tables, overlay diagrams, impedance tables, hole tables, chat and emails, dimension diagrams, panelization diagrams, modification orders, other text and tables not belonging to the above types, and other types.

[0019] Plain text board manufacturing instructions are notes, instructions, and requirements given in plain text form. Board manufacturing instruction tables are Fabrication Notes and board manufacturing instruction tables given in tabular form. Full-page board manufacturing requirement tables are board manufacturing instruction tables given in full-page tabular form. Overlay diagrams are overlay diagrams given in tabular and image form. Impedance tables are impedance tables given in tabular and image form.

[0020] The source files are categorized into twelve types, including plain text PCB fabrication instructions, stack-up diagrams, impedance tables, chat logs, and emails. This allows for refined and standardized classification of PCB quotation-related source files, clearly identifying information from different types and purposes and preventing information from mixing and interfering with parameter extraction. It also lays the foundation for subsequent precise parameter classification and targeted extraction by file type, making parameter extraction more targeted and improving the efficiency of the visual-language model in recognizing and inferring information from different file types, further ensuring the accuracy of parameter extraction.

[0021] S3. Classify the user-defined list of parameters to be extracted to obtain a dictionary of file source-parameter group-knowledge hint words. In this process, first, the parameters are classified into the corresponding file types to obtain a dictionary of file source-parameter group. Then, each parameter in the dictionary of file source-parameter group is instantiated according to a predefined domain knowledge hint word template to construct a dictionary of file source-parameter group-knowledge hint words.

[0022] First, parameters are categorized by file type to obtain a file source-parameter group dictionary. Then, a file source-parameter group-knowledge prompt dictionary is constructed using domain knowledge prompt templates. This achieves a structured and professional organization of user-defined parameters to be extracted, ensuring that parameters are accurately bound to their corresponding file sources and professional knowledge prompts. Integrating domain PCB pricing knowledge into the prompt construction guides the visual-language model to align with PCB industry professional rules for parameter reasoning and extraction, resolving extraction bias issues caused by the lack of industry knowledge support in general models. At the same time, the structured dictionary format facilitates the model's rapid identification and matching of parameters with corresponding file information, improving model reasoning efficiency and the professionalism and accuracy of parameter extraction.

[0023] S4. Extract filtering information prompts from the source files that are being filtered out. Specifically, first extract all the text information from the source files that are being filtered out, and then refine the text information to obtain the filtering information prompts. It should be noted that the files being filtered out refer to files that do not belong to the above twelve types.

[0024] The entire text information of the filtered source file is extracted and refined into filtering information prompts, enabling effective use of all information in the source file and avoiding the loss of key PCB quotation information hidden in the filtered file. The refined filtering information prompts can serve as supplementary reasoning basis for the visual-language model, helping the model to eliminate invalid information interference, correct reasoning biases, and further improve the information dimensions of parameter extraction, making the final output PCB quotation parameters more comprehensive and accurate, and reducing quotation errors caused by information omissions.

[0025] S5. Input the file source-parameter group-knowledge prompt dictionary and filtered information prompts into the visual-language model, and output the final PCB quotation parameter set.

[0026] Specifically, the document source, parameter group, knowledge prompt dictionary, and filtered information prompts are input into the visual-language model. For the preliminary PCB quotation parameter set output by inference based on chart information, the coordinate frame of the source document on which the preliminary PCB quotation parameter set is based is obtained as the source information. For the preliminary PCB quotation parameter set output by inference based on text information, the coordinates of the text are matched from the inference process of the preliminary PCB quotation parameter set as the source information. The source information is combined with the preliminary PCB quotation parameter set to obtain the final PCB quotation parameter set. Each parameter has corresponding source information and extraction results. The final PCB quotation parameter set can be obtained by putting the source information of each parameter and the preliminary PCB quotation parameter set together according to the parameter fields.

[0027] After inputting relevant prompts into the model, source file coordinate frame traceability information is bound to chart-type inference parameters, and source information is directly labeled for text-type inference parameters. The traceability information is then integrated to obtain the final parameter set, making PCB quotation parameters traceable and verifiable. This ensures that each extracted parameter corresponds to a specific location or textual basis in the source file, facilitating subsequent verification and validation of parameter accuracy by staff and reducing the cost of parameter error troubleshooting. At the same time, binding traceability information enhances the credibility of parameter extraction results, avoids unfounded erroneous inferences by the model, and further guarantees the accuracy and reliability of PCB quotation parameters.

[0028] Furthermore, directly extracting text and coordinate data from the source file provides original and accurate basic data support for subsequent steps such as file type classification, prompt word construction, and traceability information binding, avoiding distortion and deviation problems caused by secondary data conversion. At the same time, the directly extracted text and coordinate data can accurately reflect the original information features of the source file, making the reasoning of the visual-language model and the matching of traceability information more consistent with the actual source file, ensuring the accuracy of each step of the operation, and ultimately improving the overall accuracy of PCB quotation parameter extraction.

[0029] Based on the same inventive concept, this invention provides a PCB quotation parameter extraction system, such as... Figure 2 As shown, the PCB quotation parameter extraction method described above includes: a file conversion module, a layout detection and classification module, a parameter grouping module by file source, a visual-language model, an optical character recognition module, an information extraction and summarization module, a source tracing module, and a text matching module.

[0030] The file conversion module is used to convert the acquired source files into images.

[0031] The layout detection and classification module is used to classify images into their corresponding file types.

[0032] The parameter grouping module by file source is used to assign parameters to the corresponding file types, resulting in a file source-parameter group dictionary.

[0033] The domain knowledge hint module is used to instantiate each parameter of the file source-parameter group dictionary according to a predefined domain knowledge hint template, and construct the file source-parameter group-knowledge hint dictionary.

[0034] The optical character recognition module is used to identify text information in the filtered files from the source file.

[0035] The information extraction and summarization module is used to extract and filter out information prompts from text information.

[0036] A visual-language model is used to obtain a dictionary of file source, parameter group, and knowledge prompt words, filter out information prompt words, and output a preliminary PCB quotation parameter set.

[0037] Furthermore, the image is processed by the optical character recognition module to obtain the text and coordinates in the source file. For parameters that depend on text information, the reasoning process is passed through the text matching module to match the text in the input file, and the coordinates of the matched text are output as the source information of the parameters that depend on text information. For parameters that depend on chart information, the reasoning process is then passed through the source tracing module of the visual-language model to obtain the coordinate frame of the original text on which the reasoning is based, as the source information of the parameters that depend on chart information.

[0038] Finally, the traceability information of all parameters is combined with the preliminary PCB quotation parameter set to obtain the final PCB quotation parameter set, which is output to the front end for display and submitted to the customer for review.

[0039] For details on the implementation process of the PCB quotation parameter extraction system, please refer to the previous text; it will not be repeated here.

[0040] Based on the same inventive concept, the present invention provides one or more computer-readable media having instructions stored thereon that, when executed by one or more processors, cause the processors to perform the above-described PCB quotation parameter extraction method.

[0041] Optionally, the computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), solid-state drive (SSD), or optical disc, etc.

[0042] Random access memory can include resistive random access memory (ReRAM) and dynamic random access memory (DRAM).

[0043] Based on the same inventive concept, the present invention also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory is used to store computer programs; and the processor is used to implement the above-mentioned PCB quotation parameter extraction method when executing the program stored in the memory.

[0044] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0045] The communication interface is used for communication between the aforementioned terminal and other devices.

[0046] The memory may include random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0047] The aforementioned processors can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0048] In this description, references to terms such as "an embodiment," "example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example.

[0049] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style of the specification is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0050] The technical principles of the present invention have been described above with reference to specific embodiments. These descriptions are merely for explaining the principles of the invention and should not be construed as limiting the scope of protection of the invention in any way. Based on this explanation, those skilled in the art can readily conceive of other specific embodiments of the invention without inventive effort, and these embodiments will all fall within the scope of protection of the present invention.

Claims

1. A method for extracting PCB quotation parameters, characterized in that, include: Convert the obtained source file into an image; Classify the images into their corresponding file types; The user-defined list of parameters to be extracted is categorized to obtain a dictionary of file source, parameter group, and knowledge prompt words; Extract filtering information prompts from the source files that were filtered out. Input the file source, parameter group, knowledge prompt dictionary, and filtered information prompts into the visual-language model, and output the final PCB quotation parameter set.

2. The PCB quotation parameter extraction method according to claim 1, characterized in that, The file types include: plain text board making instructions, board making instruction tables, full-page board making requirements tables, overlay diagrams, impedance tables, hole tables, chat and emails, dimension diagrams, panelization diagrams, modification orders, other text and tables, and other types.

3. The PCB quotation parameter extraction method according to claim 1, characterized in that, The step of classifying the user-defined list of parameters to be extracted to obtain a dictionary of file sources and parameter groups includes: The parameters are assigned to their corresponding file types to obtain a file source-parameter group dictionary; For each parameter in the file source-parameter group dictionary, instantiate it according to the predefined domain knowledge hint word template to construct the file source-parameter group-knowledge hint word dictionary.

4. The PCB quotation parameter extraction method according to claim 1, characterized in that, The step of extracting filtering information prompts from the source file of the filtered file includes: Extract all text information from the source file that has been filtered out; Extract the text information to obtain the filtered information prompts.

5. The PCB quotation parameter extraction method according to claim 4, characterized in that, The step of inputting the file source-parameter group-knowledge prompt dictionary and filtered information prompts into the visual-language model and outputting the PCB quotation parameter set includes: Input the file source, parameter group, knowledge prompt dictionary, and filtered information prompts into the visual-language model; For the preliminary PCB quotation parameter set output based on chart information, obtain the coordinate frame of the source file on which the preliminary PCB quotation parameter set is based for inference, as traceability information; For the initial PCB quotation parameter set that relies on textual information for reasoning output, the coordinates of the text are matched from the reasoning process of the initial PCB quotation parameter set as traceability information; The traceability information is combined with the preliminary PCB quotation parameter set to obtain the final PCB quotation parameter set.

6. A PCB quotation parameter extraction system, characterized in that, The method for extracting PCB pricing parameters as described in any one of claims 1 to 5 includes: The file conversion module is used to convert the acquired source files into images; The layout detection and classification module is used to classify images into their corresponding file types; The parameter grouping module by file source is used to group parameters into their corresponding file types, resulting in a file source-parameter group dictionary; The domain knowledge hint word module is used to instantiate each parameter of the file source-parameter group dictionary according to the predefined domain knowledge hint word template, and construct the file source-parameter group-knowledge hint word dictionary; An optical character recognition module is used to identify text information in files that have been filtered out from the source file; The information extraction and summarization module is used to extract and filter out information prompts from text information; A visual-language model is used to obtain a dictionary of file source, parameter group, and knowledge prompt words, filter out information prompt words, and output the final PCB quotation parameter set.

7. One or more computer-readable media, characterized in that, It stores instructions that, when executed by one or more processors, cause the processors to perform the PCB quotation parameter extraction method as described in any one of claims 1-5.

8. An electronic device, characterized in that, The device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; the memory is used to store computer programs; and the processor, when executing the program stored in the memory, implements the PCB quotation parameter extraction method as described in any one of claims 1 to 5.

Images