A line arc three-dimensional detection method based on retrieval enhancement generation

Abstract

The application discloses a kind of based on search enhancement generation's line arc three-dimensional detection method, comprising: in multiple view angles acquisition chip pad's first image information, lead frame pad's second image information and equipment state information, respectively to each first image information and each second image information are handled, generate chip pad's three-dimensional position information and lead frame pad's three-dimensional position information, chip pad's three-dimensional position information, lead frame pad's three-dimensional position information, equipment state information is input to preset large language model and is handled, and call search enhancement generation module, obtain three-dimensional arc line model, based on three-dimensional arc line model to the line arc that exceeds preset threshold range is labeled, generates three-dimensional line arc structure detection result.The scheme can be adapted to different packaging factory process, different specification product, system inclusivity is stronger, three-dimensional line arc solution and defect determination accuracy significantly improve.

Description

Technical Field

[0001] This application relates to the field of semiconductor power device packaging and manufacturing technology, and in particular to a three-dimensional detection method for line arcs based on search enhancement generation. Background Technology

[0002] Wire bonding is a core interconnect technology in semiconductor packaging. It uses extremely fine metal wires, employing heat, pressure, and ultrasonic energy to tightly bond metal wires between chip pads and leadframe pins, thereby achieving electrical interconnection and information exchange between the chip and external circuits. Wire bonding is also a critical process step in semiconductor packaging. During bonding, the metal wires form a specific three-dimensional spatial structure between the chip and the leadframe, known in the industry as a wire arc. To ensure packaging reliability, the crossing and contact between different wire arcs must be strictly avoided during packaging manufacturing. Otherwise, it can lead to malfunctions or even complete failure of electronic components.

[0003] Currently, traditional packaging manufacturing mainly relies on the experience of professionals. The process from program design to actual production is cumbersome and time-consuming. The experience and knowledge levels of different personnel vary greatly, resulting in unstable soldering quality and easy omissions and unreasonable aspects.

[0004] To solve practical problems in the manufacturing process and improve product quality, a more scientific testing method is needed to ensure process reliability and enhance packaging robustness. Summary of the Invention

[0005] This application aims to at least solve the technical problems existing in the prior art. To this end, the first aspect of this application proposes a three-dimensional detection method for line arcs based on retrieval enhancement generation, the method comprising: Acquire first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives; Each first image and each second image are processed to generate three-dimensional position information of chip pads and three-dimensional position information of lead frame pads; The three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information are input into the preset large language model for processing, and the retrieval enhancement generation module is called to obtain the three-dimensional arc model. Based on the three-dimensional arc model, the arcs that exceed the preset threshold range are labeled to generate three-dimensional arc structure detection results.

[0006] In one possible implementation, the first image information and the second image information are processed respectively to generate three-dimensional position information of the chip pads and three-dimensional position information of the leadframe pads, including: Feature extraction is performed on each first image information and each second image information to generate first contour information corresponding to chip pads and second contour information corresponding to lead frame pads; The three-dimensional position information of the chip pads is obtained by calculating the first contour information through a preset stereo matching operation; the three-dimensional position information of the lead frame pads is obtained by calculating the second contour information through a preset stereo matching operation.

[0007] In one possible implementation, the three-dimensional position information of the chip pads, the three-dimensional position information of the leadframe pads, and the device status information are input into a preset large language model for processing, and a retrieval enhancement generation module is invoked to obtain a three-dimensional arc model, including: Construct a multimodal knowledge network; wherein, the multimodal knowledge network includes package outline, package wire bonding diagram, and packaging equipment; Based on text similarity, wire diameter, and metal wire material type, an associated retrieval mechanism is used to perform associated retrieval and output an encapsulated knowledge vector set. The three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, the device status information, and the package knowledge vector set are input into the preset large language model. The preset large language model performs text understanding of package professional materials, threshold comparison, and logical reasoning to generate a three-dimensional arc model.

[0008] In one possible implementation, the device status information includes device gyroscope angle information and device camera height information.

[0009] In one possible implementation, the pre-defined large language model is trained based on a semiconductor packaging and testing-specific corpus. The generation process of the pre-defined large language model includes: The deep recurrent neural network is iteratively trained using historical semiconductor packaging and testing data until the error between the model's predicted output and the target output label is less than a preset threshold, thus generating a preset large language model.

[0010] A second aspect of this application proposes a three-dimensional detection system for line arcs based on retrieval enhancement generation, the system comprising: The data acquisition module is used to acquire first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives; The conversion module is used to process each first image information and each second image information respectively to generate the three-dimensional position information of the chip pads and the three-dimensional position information of the lead frame pads. The large language model processing module is used to input the three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information into the preset large language model for processing, and to call the retrieval enhancement generation module to obtain the three-dimensional arc model. The detection result generation module is used to annotate line arcs that exceed the preset threshold range based on the three-dimensional arc model, and generate three-dimensional line arc structure detection results.

[0011] In one possible implementation, the above-mentioned conversion module is specifically used for: Feature extraction is performed on each first image information and each second image information to generate first contour information corresponding to chip pads and second contour information corresponding to lead frame pads; The three-dimensional position information of the chip pads is obtained by calculating the first contour information through a preset stereo matching operation; the three-dimensional position information of the lead frame pads is obtained by calculating the second contour information through a preset stereo matching operation.

[0012] In one possible implementation, the aforementioned large language model processing module is specifically used for: Construct a multimodal knowledge network; wherein, the multimodal knowledge network includes package outline, package wire bonding diagram, and packaging equipment; Based on text similarity, wire diameter, and metal wire material type, an associated retrieval mechanism is used to perform associated retrieval and output an encapsulated knowledge vector set. The three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, the device status information, and the package knowledge vector set are input into the preset large language model. The preset large language model performs text understanding of package professional materials, threshold comparison, and logical reasoning to generate a three-dimensional arc model.

[0013] In one possible implementation, the device status information includes device gyroscope angle information and device camera height information.

[0014] In one possible implementation, the device is also used for: The deep recurrent neural network is iteratively trained using historical semiconductor packaging and testing data until the error between the model's predicted output and the target output label is less than a preset threshold, thus generating a preset large language model.

[0015] A third aspect of this application provides an electronic device comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the retrieval-enhanced three-dimensional detection method for line arcs as described in the first aspect.

[0016] The fourth aspect of this application proposes a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the retrieval-enhanced three-dimensional detection method for line arcs as described in the first aspect.

[0017] The embodiments of this application have the following beneficial effects: This application provides a three-dimensional line arc detection method based on retrieval enhancement generation. The method includes: acquiring first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives; processing each first image information and each second image information to generate three-dimensional position information of chip pads and lead frame pads; inputting the three-dimensional position information of chip pads, lead frame pads, and device status information into a preset large language model for processing; and calling a retrieval enhancement generation module to obtain a three-dimensional arc model; and annotating line arcs exceeding a preset threshold range based on the three-dimensional arc model to generate a three-dimensional line arc structure detection result. This solution introduces an enhanced retrieval architecture for semiconductor packaging and testing line arc scenarios, constructing a multimodal knowledge network for packaging. This network is adaptable to different packaging plant processes and product specifications, resulting in greater system inclusiveness and significantly improved accuracy in 3D line arc calculation and defect identification. Furthermore, by relying on a pre-trained semiconductor domain language model combined with standardized reasoning logic to complete line arc analysis, it no longer excessively depends on engineers' personal experience, unifying judgment standards and ensuring consistent packaging product quality and long-term reliability. In addition, by outputting a high-precision, low-error complete 3D line arc model, complete data records can be retained for various line arc dimensional out-of-limits and deformation anomalies, enabling traceability of process anomalies. At the same time, the 3D model can provide quantitative reference for analyzing molding flow impact and optimizing packaging processes. Attached Figure Description

[0018] Figure 1 A block diagram of a computer device provided in an embodiment of this application; Figure 2 A flowchart illustrating the steps of a three-dimensional detection method for line arcs based on retrieval enhancement provided in this application embodiment; Figure 3 This application provides a schematic diagram of the overall assembly three-dimensional structure of a three-dimensional line arc detection device. Figure 4 An enlarged view of a sphere rotation structure provided in an embodiment of this application; Figure 5 This is a schematic diagram of a concave structure of a tray support provided in an embodiment of this application; Figure 6An enlarged view of a tray support provided in an embodiment of this application.

[0019] Explanation of reference numerals in the attached figures: 1 ~ Display screen; 2 ~ Data cable; 3 ~ Computer host; 4 ~ Screw; 5 ~ Camera; 6 ~ Movable base; 7 ~ Base horizontal movement knob; 8 ~ Base vertical movement knob; 9 ~ Locking structure; 10 ~ Tray support; 11 ~ Ball rotation structure; 12 ~ Operating platform; 13 ~ Joystick; 14 ~ Dedicated lead frame tray; 15 ~ Tray lower convex structure; 16 ~ Tray support upper concave structure; 17 ~ Horizontal drive rubber wheel; 18 ~ Vertical drive rubber wheel; 19 ~ Gyroscope; 20 ~ Tray locking structure; 21 ~ Tray locking structure knob. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0021] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, unless otherwise stated, "a plurality of" means two or more. Furthermore, the use of "based on" or "according to" implies openness and inclusiveness, because processes, steps, calculations, or other actions "based on" or "according to" one or more of the stated conditions or values ​​may in practice be based on additional conditions or beyond the stated values.

[0022] The information retrieval method provided in this application can be applied to computer equipment (electronic devices). The computer equipment can be a server or a terminal. The server can be a single server or a server cluster composed of multiple servers. This application does not specifically limit this. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets and portable wearable devices.

[0023] Taking a computer device as an example, Figure 1 A block diagram of a server is shown, such as Figure 1As shown, the server may include a processor and memory connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. When executed by the processor, the computer program implements a retrieval-enhanced generation-based three-dimensional line arc detection method.

[0024] Those skilled in the art will understand that Figure 1 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the server to which the present application is applied. Optionally, the server may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0025] It should be noted that the execution subject of the embodiments of this application can be a computer device or an information retrieval device. The following method embodiments will be described with a computer device as the execution subject.

[0026] Figure 2 This document provides a flowchart of a three-dimensional line arc detection method based on retrieval enhancement, as illustrated in an embodiment of this application. Figure 2 As shown, the method includes the following steps: Step 202: Acquire first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives.

[0027] Specifically, an image sensor can be used to acquire first image information of the chip pads and second image information of the leadframe pads from at least two different perspectives. Device status information can also be acquired, optionally including device gyroscope angle information and device camera height information.

[0028] Step 204: Process each first image information and each second image information respectively to generate the three-dimensional position information of the chip pads and the three-dimensional position information of the lead frame pads.

[0029] After acquiring the first image information and the second image information, the first image information and the second image information can be processed separately to generate the three-dimensional position information of the chip pads and the three-dimensional position information of the lead frame pads.

[0030] In some optional embodiments, the first image information and the second image information are processed respectively to generate three-dimensional position information of the chip pads and three-dimensional position information of the lead frame pads, including: Feature extraction is performed on each first image and each second image to generate first contour information corresponding to chip pads and second contour information corresponding to lead frame pads. The three-dimensional position information of chip pads is obtained by calculating each first contour information through a preset stereo matching operation. The three-dimensional position information of lead frame pads is obtained by calculating each second contour information through a preset stereo matching operation.

[0031] Specifically, the preset stereo matching operation is a multi-view vision matching algorithm that has completed parameter calibration and logic encapsulation in advance. This operation performs corresponding matching on the same pad contour pixels collected from at least two different shooting angles. Based on the parallax principle, combined with camera intrinsic parameters and shooting pose parameters, it reversely solves the real three-dimensional spatial coordinates of the target object (chip pad, lead frame pad).

[0032] Feature recognition can be performed on each first image and each second image from different perspectives, and the first contour information corresponding to the chip pads and the second contour information corresponding to the lead frame pads from different perspectives can be extracted respectively.

[0033] Next, stereo matching can be performed on the first contour information corresponding to the chip pads and the second contour information corresponding to the lead frame pads from different perspectives. Based on the stereo matching results, the three-dimensional center point coordinates of the chip pads and the substrate pads can be calculated to obtain the three-dimensional position information of the chip pads and the substrate pads.

[0034] Step 206: Input the three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information into the preset large language model for processing, and call the retrieval enhancement generation module to obtain the three-dimensional arc model.

[0035] In some optional embodiments, the three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information are input into a preset large language model for processing, and the retrieval enhancement generation module is invoked to obtain a three-dimensional arc model, including: A multimodal knowledge network is constructed, which includes package outline, package wire bonding diagram, and packaging equipment. Based on text similarity, wire diameter, and metal wire material type, a packaging scenario association retrieval mechanism is used to perform association retrieval and output a set of packaging knowledge vectors. The three-dimensional position information of chip pads, the three-dimensional position information of lead frame pads, equipment status information, and the set of packaging knowledge vectors are input into a preset large language model. The preset large language model is then used to perform text understanding of packaging professional materials, threshold comparison, and logical reasoning to generate a three-dimensional arc model.

[0036] Specifically, a multimodal knowledge network can be constructed first based on relevant information such as package shape, package wire bonding diagram, and packaging equipment. Then, based on text similarity, wire diameter, and metal wire material type, a package scenario association retrieval mechanism can be used to perform association retrieval, thereby outputting a set of package knowledge vectors.

[0037] Next, the 3D position information of the chip pads, the 3D position information of the leadframe pads, the device status information, and the package knowledge vector set can be input into a preset large language model. The preset large language model then performs text understanding of package-related professional materials, threshold comparison, and logical reasoning to generate a 3D arc model. This 3D arc model is a highly accurate 3D model with minimal error compared to the actual dimensions.

[0038] In some optional embodiments, the preset large language model is trained based on a semiconductor packaging and testing-specific corpus. The process of generating the preset large language model includes: iteratively training a deep recurrent neural network using multiple historical semiconductor packaging and testing data until the error between the model's predicted output and the target output label is less than a preset threshold, thereby generating the preset large language model.

[0039] Step 208: Based on the three-dimensional arc model, mark the arcs that exceed the preset threshold range and generate three-dimensional arc structure detection results.

[0040] The preset threshold range can be customized according to actual needs. By comparing the thresholds, lines and arcs that exceed the range can be marked with special colors, and a real three-dimensional line and arc structure with reference value can be output, thus obtaining the final three-dimensional line and arc structure detection result.

[0041] In addition, this application embodiment also provides an overall assembled three-dimensional structure for a dedicated three-dimensional line arc detection device, which is used to implement the above-mentioned three-dimensional line arc detection method based on retrieval enhancement. Figure 3 As shown, Figure 3This application provides a schematic diagram of the overall assembly of a three-dimensional line arc detection device, which includes a display screen (1), a data cable (2), a computer host (3) equipped with the necessary hardware such as CPU / GPU, a screw (4), a camera (5), a movable base (6), a base horizontal movement knob (7), a base vertical movement knob (8), a locking structure (9) to fix the current tray movement angle, a tray support (10), a spherical rotation structure (11), an operating table (12) corresponding to the spherical rotation structure, a rocker arm (13), and a special lead frame tray (14). The tray can change its length and width through a specific telescopic structure to adapt to different frames. The center of the tray is concave to avoid contamination of the chip area by impurities introduced during the observation process. There is also a convex structure on the tray (15) and a concave structure on the tray support (16). The connection between the tray and the support is realized through the triangular structure and the electromagnetic locking switch of the support, and different positions are observed through different combinations of triangular structures. In this embodiment, only a single camera module is used to complete multi-view image acquisition and 3D reconstruction, abandoning the traditional multi-camera array or complex camera moving mechanism, which greatly reduces the equipment procurement and enterprise usage costs. Furthermore, by adopting an architecture that uses a fixed camera and a spherical stage to drive the sample to rotate and shoot in multiple directions, it is different from the traditional moving camera or multi-camera solution, reducing imaging distortion and calibration errors caused by camera displacement, and further improving the accuracy of 3D contour extraction and coordinate calculation.

[0042] like Figures 4-6 As shown, Figure 4 Figure 1 is an enlarged view of a sphere rotation structure provided in an embodiment of this application. It includes a rubber wheel (17), connected to a corresponding servo motor, to enable horizontal movement of the sphere; a rubber wheel (18), connected to a corresponding servo motor, to enable vertical movement of the sphere; a gyroscope (19), which, through a computer software system algorithm, can determine the angles of horizontal and vertical movement of the sphere; a tray locking structure (20); and a tray locking structure knob (21), through which the position of the tray locking structure (20) can be adjusted to lock the lead frame and prevent it from falling during observation.

[0043] Figure 5 This is a schematic diagram of a concave structure of a tray support provided in an embodiment of this application. Figure 6 An enlarged view of a tray support provided in an embodiment of this application.

[0044] This application provides a three-dimensional line arc detection method based on retrieval enhancement generation. The method includes: acquiring first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives; processing each first image information and each second image information to generate three-dimensional position information of chip pads and lead frame pads; inputting the three-dimensional position information of chip pads, lead frame pads, and device status information into a preset large language model for processing; and calling a retrieval enhancement generation module to obtain a three-dimensional arc model; and annotating line arcs exceeding a preset threshold range based on the three-dimensional arc model to generate three-dimensional line arc structure detection results. This solution introduces an enhanced retrieval architecture for semiconductor packaging and testing line arc scenarios, constructing a multimodal knowledge network for packaging. This network is adaptable to different packaging plant processes and product specifications, resulting in greater system inclusiveness and significantly improved accuracy in 3D line arc calculation and defect identification. Furthermore, by relying on a pre-trained semiconductor domain language model combined with standardized reasoning logic to complete line arc analysis, it no longer excessively depends on engineers' personal experience, unifying judgment standards and ensuring consistent packaging product quality and long-term reliability. In addition, by outputting a high-precision, low-error complete 3D line arc model, complete data records can be retained for various line arc dimensional out-of-limits and deformation anomalies, enabling traceability of process anomalies. At the same time, the 3D model can provide quantitative reference for analyzing molding flow impact and optimizing packaging processes.

[0045] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0046] This application also provides a 3D line arc detection system based on retrieval enhancement generation, the system comprising: The data acquisition module is used to acquire first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives; The conversion module is used to process each first image information and each second image information respectively to generate the three-dimensional position information of the chip pads and the three-dimensional position information of the lead frame pads. The large language model processing module is used to input the three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information into the preset large language model for processing, and to call the retrieval enhancement generation module to obtain the three-dimensional arc model. The detection result generation module is used to annotate line arcs that exceed the preset threshold range based on the three-dimensional arc model, and generate three-dimensional line arc structure detection results.

[0047] Regarding the system in the above embodiments, the specific methods by which each module performs its operations have been described in detail in the embodiments related to the method, and will not be elaborated upon here. Each module in the above information retrieval system can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations of each module.

[0048] In one embodiment of this application, a computer device is provided, the computer device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps: Acquire first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives; Each first image and each second image are processed to generate three-dimensional position information of chip pads and three-dimensional position information of lead frame pads; The three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information are input into the preset large language model for processing, and the retrieval enhancement generation module is called to obtain the three-dimensional arc model. Based on the three-dimensional arc model, the arcs that exceed the preset threshold range are labeled to generate three-dimensional arc structure detection results.

[0049] The computer device provided in this application embodiment has a similar implementation principle and technical effect to the above method embodiment, and will not be described again here.

[0050] In one embodiment of this application, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, it performs the following steps: Acquire first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives; Each first image and each second image are processed to generate three-dimensional position information of chip pads and three-dimensional position information of lead frame pads; The three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information are input into the preset large language model for processing, and the retrieval enhancement generation module is called to obtain the three-dimensional arc model. Based on the three-dimensional arc model, the arcs that exceed the preset threshold range are labeled to generate three-dimensional arc structure detection results.

[0051] The computer-readable storage medium provided in this embodiment is similar in principle and technical effect to the method embodiment described above, and will not be repeated here.

[0052] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0053] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

Claims

1. A three-dimensional detection method for line arcs based on retrieval enhancement generation, characterized in that, The method includes: Acquire first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives; Each of the first image information and each of the second image information are processed to generate the three-dimensional position information of the chip pads and the three-dimensional position information of the lead frame pads; The three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information are input into a preset large language model for processing, and the retrieval enhancement generation module is invoked to obtain a three-dimensional arc model. Based on the three-dimensional arc model, the arcs exceeding the preset threshold range are labeled to generate three-dimensional arc structure detection results.

2. The method according to claim 1, characterized in that, The step of processing the first image information and the second image information respectively to generate the three-dimensional position information of the chip pads and the three-dimensional position information of the lead frame pads includes: Feature extraction is performed on each of the first image information and each of the second image information to generate the first contour information corresponding to the chip pad and the second contour information corresponding to the lead frame pad; The three-dimensional position information of the chip pads is obtained by calculating each of the first contour information through a preset stereo matching operation; and the three-dimensional position information of the lead frame pads is obtained by calculating each of the second contour information through a preset stereo matching operation.

3. The method according to claim 2, characterized in that, The process involves inputting the three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information into a preset large language model for processing, and then calling the retrieval enhancement generation module to obtain a three-dimensional arc model, including: Construct a multimodal knowledge network; wherein, the multimodal knowledge network includes package outline, package wire bonding diagram, and packaging equipment; Based on text similarity, wire diameter, and metal wire material type, an associated retrieval mechanism is used to perform associated retrieval and output an encapsulated knowledge vector set. The three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, the device status information, and the packaging knowledge vector set are input into the preset large language model. The preset large language model is then used to perform text understanding of packaging professional materials, threshold comparison, and logical reasoning to generate the three-dimensional arc model.

4. The method according to any one of claims 1-3, characterized in that, The device status information includes the device gyroscope angle information and the device camera height information.

5. The method according to any one of claims 1-3, characterized in that, The preset large language model is trained based on a semiconductor packaging and testing-specific corpus. The generation process of the preset large language model includes: The deep recurrent neural network is iteratively trained using multiple historical semiconductor packaging and testing data until the error between the model's predicted output and the target output label is less than a preset threshold, thereby generating the preset large language model.

6. A three-dimensional detection system for linear arcs based on retrieval enhancement generation, characterized in that, The system includes: The data acquisition module is used to acquire first image information of chip pads, second image information of lead frame pads, and device status information from multiple perspectives; The conversion module is used to process each of the first image information and each of the second image information respectively to generate the three-dimensional position information of the chip pad and the three-dimensional position information of the lead frame pad; The large language model processing module is used to input the three-dimensional position information of the chip pads, the three-dimensional position information of the lead frame pads, and the device status information into a preset large language model for processing, and to call the retrieval enhancement generation module to obtain a three-dimensional arc model. The detection result generation module is used to annotate the arcs that exceed the preset threshold range based on the three-dimensional arc model, and generate three-dimensional arc structure detection results.

7. An electronic device, characterized in that, The electronic device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The storage medium stores at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the method as described in any one of claims 1-5.

Images