Hardware multi-mode detection method, apparatus, device, and medium

Abstract

The application relates to a hardware multi-mode detection method, device, equipment and medium, and relates to the technical field of embedded hardware circuit board detection.The application comprises the following steps: three-dimensional modeling is performed on an optical image collected by an industrial binocular lens based on a hardware circuit board, and spatial coordinate data is obtained; surface abnormal welding data of welding points of each electronic component is constructed based on radar echo data; three-dimensional fusion data models are constructed through multi-source data fusion processing based on the spatial coordinate data, the surface abnormal welding data and optical image data, the abnormal positions of the radar echo data are judged, the electronic component coordinate data corresponding to the abnormal positions is determined, detection probes are driven to move to the abnormal positions, and the electronic components at the abnormal positions are subjected to automatic detection of electrical parameters and working performance parameters, and a detection report is generated.The application is not prone to causing physical damage to the circuit board, manual welding and positioning are not required, the risk of human error operation is eliminated, the detection efficiency is improved, and online nondestructive detection is realized.

Description

Technical Field

[0001] This application relates to the field of embedded hardware circuit board testing technology, specifically to a hardware multi-mode testing method, apparatus, equipment, and medium. Background Technology

[0002] With the rapid development of industrial control technology, embedded systems are increasingly widely used in industrial automation, intelligent manufacturing, and other fields. To ensure the stability and reliability of industrial control systems, accurate and efficient signal detection and fault detection of embedded hardware circuit boards have become a key aspect of product development and production maintenance.

[0003] Currently, signal detection for micro-components (such as 0402 package resistors, 01005 package components, and high-density BGA chips) in embedded hardware circuit boards mainly relies on traditional flying wire soldering. This method involves manually soldering fine wires to test points on the circuit board, which are then connected to external measuring equipment for detection. However, as electronic components become increasingly miniaturized and highly integrated, this traditional contact-based detection method is gradually revealing the following serious drawbacks: First, flying wire soldering requires penetrating pads or component leads, which can easily cause irreversible physical damage to the circuit board. This is especially true for micro-packaged components such as 0402 and 01005, whose pads are extremely small, making them highly susceptible to detachment during soldering. For BGA chips with pin pitch less than 0.4mm, flying wire soldering can easily cause short circuits between adjacent pins. This permanent hardware damage not only wastes resources but also significantly increases the trial-and-error costs during the R&D phase.

[0004] Secondly, wire bonding is a delicate manual operation that must be performed under a microscope, requiring extremely high operator skills. The error rate exceeds 30%, resulting in low inspection efficiency. Furthermore, this method is only applicable to exposed pads. For components with shielding covers or internally hidden pads, the shielding covers must be removed first, making online non-destructive testing impossible. Summary of the Invention

[0005] In view of this, this application provides a hardware multi-mode testing method, apparatus, equipment and medium. The main purpose is to solve the technical problems of existing flying wire soldering process, which requires penetrating the pads or component pins for operation, which can easily cause irreversible physical damage to the circuit board. Flying wire soldering also requires highly skilled operators, has an error rate of more than 30%, low testing efficiency, and cannot achieve online non-destructive testing.

[0006] Firstly, this application provides a hardware multi-mode detection method, including: Optical images of a hardware circuit board fixed to a testing platform are acquired using an industrial binocular lens. Based on the acquired optical image data, a 3D model is created to obtain the spatial coordinate data of the hardware circuit board relative to the testing platform. The hardware circuit board is scanned by a miniature millimeter-wave radar to obtain radar echo data, and surface abnormal welding data of each electronic component solder joint is constructed based on the radar echo data. The spatial coordinate data, the surface abnormal welding data, and the optical image data are subjected to multi-source data fusion processing to construct a three-dimensional fusion data model for the detection of the hardware circuit board. Based on the three-dimensional fusion data model, the abnormal location of the radar echo data is determined, and the coordinate data of the electronic components corresponding to the abnormal location is determined. Based on the determined coordinate data of the electronic components, the detection probe is driven to move to the abnormal position, and the electrical parameters and operating performance parameters of the electronic components at the abnormal position are automatically detected, generating a detection report.

[0007] Secondly, this application provides a hardware multi-mode detection device, comprising: The acquisition module is used to acquire optical images of a hardware circuit board fixed to the inspection platform through an industrial binocular lens, perform three-dimensional modeling based on the acquired optical image data, and obtain the spatial coordinate data of the hardware circuit board relative to the inspection platform. A construction module is used to scan the hardware circuit board with a micro millimeter-wave radar to obtain radar echo data, and to construct surface abnormal welding data of each electronic component solder joint based on the radar echo data. The fusion module is used to perform multi-source data fusion processing on the spatial coordinate data, the surface abnormal welding data, and the optical image data to construct a three-dimensional fusion data model for the inspection of the hardware circuit board. The determination module is used to determine the abnormal location of the radar echo data based on the three-dimensional fusion data model, and to determine the coordinate data of the electronic components corresponding to the abnormal location; The detection module is used to drive the detection probe to move to the abnormal position according to the determined coordinate data of the electronic component, and to automatically detect the electrical parameters and operating performance parameters of the electronic component at the abnormal position, and generate a detection report.

[0008] Thirdly, this application provides an electronic device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the computer program to implement the hardware multi-mode detection method described in the first aspect.

[0009] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the hardware multi-mode detection method described in the first aspect.

[0010] By employing the above technical solutions, this application provides a hardware multi-mode inspection method, apparatus, equipment, and medium. Compared with existing technologies, this application can acquire optical images of a hardware circuit board fixed on an inspection platform using an industrial binocular lens, perform three-dimensional modeling based on the acquired optical image data, and obtain the spatial coordinate data of the hardware circuit board relative to the inspection platform; scan the hardware circuit board with a miniature millimeter-wave radar to obtain radar echo data, and construct surface abnormal welding data of each electronic component's solder joints based on the radar echo data; perform multi-source data fusion processing on the spatial coordinate data, surface abnormal welding data, and optical image data to construct a three-dimensional fusion data model for hardware circuit board inspection; determine the abnormal location of the radar echo data based on the three-dimensional fusion data model, and determine the coordinate data of the electronic component corresponding to the abnormal location; drive the inspection probe to move to the abnormal location according to the determined electronic component coordinate data, and automatically inspect the electrical parameters and operating performance parameters of the electronic component at the abnormal location, generating an inspection report.

[0011] Using the above technical solution, this application acquires optical images and performs 3D modeling through an industrial binocular lens, and combines this with micro millimeter-wave radar scanning to obtain welding point data. This process is completely non-contact, requiring no physical connection or soldering operation on the circuit board. Only after the coordinates of the abnormal location are accurately determined through multi-source data fusion does the system drive the detection probe to move to that specific location for detection, avoiding the risks of pad detachment, circuit breakage, or short circuit between adjacent pins caused by manual soldering, thus achieving non-destructive or low-destructive testing.

[0012] This application demonstrates that the entire process, from optical acquisition, radar scanning, data fusion modeling to probe-driven detection, is automated. Based on a 3D fusion data model, the system automatically identifies anomalies and drives the probe, eliminating the need for manual welding and positioning under a microscope. This automated detection replaces tedious manual operations, eliminates the risk of human error, and significantly improves detection efficiency and data reliability. Through the fusion of radar scanning and visual data, the system can locate anomalies in components with shielding or hidden internal structures, achieving a "see-through" effect without removing the shielding, thus realizing online non-destructive testing.

[0013] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0014] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A schematic flowchart of a hardware multi-mode detection method provided in an embodiment of this application; Figure 2 This application provides an overall structural diagram of a hardware multi-mode detection system. Figure 3 A flowchart of three-dimensional angle point cloud reconstruction and data fusion is provided for embodiments of this application; Figure 4 A schematic diagram illustrating the working principle of a binocular vision optical center plane ranging model provided in this application embodiment; Figure 5 A schematic diagram illustrating the working principle of a theoretical model of the internal space of a camera provided in this application embodiment; Figure 6 A schematic diagram illustrating the working principle of a geometric model of an actual optical center projection point provided in this application embodiment; Figure 7 A schematic diagram illustrating the working principle of a slant distance calculation geometric model provided in this application embodiment; Figure 8 This is a schematic diagram of a hardware multi-mode detection device provided in an embodiment of this application. Detailed Implementation

[0017] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of this application, including various details to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0018] The hardware multi-mode detection method, apparatus, device, and medium of this application are described below with reference to the accompanying drawings.

[0019] This application provides a hardware multi-mode testing method, apparatus, equipment, and medium. The main purpose is to solve the technical problems of existing flying wire soldering processes, which require penetrating pads or component pins for operation, easily causing irreversible physical damage to the circuit board. Flying wire soldering also requires highly skilled operators, has an error rate of over 30%, low testing efficiency, and cannot achieve online non-destructive testing.

[0020] like Figure 1 As shown, an embodiment of this application provides a hardware multi-mode detection method, including: Step 101: Acquire optical images of the hardware circuit board fixed on the inspection platform using an industrial binocular lens, perform 3D modeling based on the acquired optical image data, and obtain the spatial coordinate data of the hardware circuit board relative to the inspection platform.

[0021] In a specific embodiment, this application is used to detect and locate hardware soldering defects on embedded hardware circuit boards, and to drive detection probes to perform automated testing of the hardware circuits. It can automatically detect the output signals of specific component circuits during embedded hardware unit testing, record and analyze them to generate a test report.

[0022] A 3D model of a hardware circuit board fixed to an inspection platform is created using an industrial binocular lens, acquiring its spatial coordinates relative to the platform. A miniature millimeter-wave radar scans the circuit board, using radar echo data to construct surface anomaly data for each electronic component's solder joints. The acquired hardware spatial coordinate data, radar detection data, and optical image data from the binocular vision device are fused to build a 3D fusion data model for the circuit board inspection. The location of anomalies in the radar echo data is determined, identifying the coordinates of the components at those locations, and driving the inspection probe to detect the anomalies. Simultaneously, based on the precise electronic component coordinates of the circuit board, the inspection probe can be driven to automatically inspect the components' electrical parameters and performance parameters according to a pre-defined process, generating a corresponding inspection report.

[0023] This application provides a non-invasive adaptive micro-circuit probe that combines binocular vision 3D reconstruction with millimeter-wave radar microwave scanning technology to achieve 3D modeling of hardware circuit boards and anomaly point determination, enabling precise defect location and testing. The overall system structure is as follows: Figure 2 As shown, it can mainly include a detection probe subsystem, a power supply and servo system, a computing, control and storage subsystem, a radio frequency detection subsystem and an optical modeling subsystem.

[0024] The detection probe subsystem can employ ceramic gold-plated probes, with adaptively adjustable probe spacing and contact pressure controlled within 10g (i.e., pressure ≤ 10g). To improve the signal-to-noise ratio, the probe's outer layer is covered with a nano-ferrite magnetic ring, which can enhance the signal-to-noise ratio by 20dB. The probes possess three-dimensional omnidirectional sliding capability; the two horizontal probes are controlled by stepper motors on the equipment platform for translational sliding, while the vertical probe is controlled by a high-precision worm gear for up-and-down sliding.

[0025] The power supply and servo system can be powered by 24V DC. The servo system controls the horizontal movement of the 3D probe via a servo motor, while simultaneously controlling the millimeter-wave radar module to rotate and detect, thereby achieving synthetic aperture radar imaging.

[0026] The computation, control, and storage subsystem is built on the GD32 series high-performance MCU. This MCU has a built-in 12-bit ADC and hardware FFT, enabling real-time spectrum analysis of signals. The MCU communicates with the host computer via a serial port and is responsible for data storage and device control. This subsystem also includes a host computer, a digital signal processor, a timing control unit, and a data storage unit, used to control the synchronous or asynchronous operation of the optical and radio frequency subsystems to complete data acquisition of the target scene. The optical subsystem can perform data acquisition independently or simultaneously in conjunction with the radio frequency subsystem.

[0027] The radio frequency (RF) detection subsystem can consist of a transceiver antenna array, a power amplifier, a modulator, a frequency source, a mixer, and an analog-to-digital converter. This subsystem is responsible for receiving both the spatial radiation of the transmitted signal and the target scattering signal. It selects the appropriate polarized transmitting and receiving array elements according to a set timing sequence, transmitting the transmitted and echo signals. It can rapidly generate FMCW signals with adjustable signal parameters and high linearity, receive the echo signal and perform descrambling processing, and output an intermediate frequency (IF) signal to the digital subsystem.

[0028] The optical modeling subsystem can consist of a front polarizing filter, a lens, a photosensor, and an analog-to-digital converter. The polarizing filter filters out strong natural light reflected from the target observation scene, eliminating excessively strong reflected light. Either a linear polarizing filter or a circular polarizing filter can be used. A linear polarizing filter outputs linearly polarized light, with the light wave vibrating in only one direction; a circular polarizing filter outputs circularly polarized light, with the light wave vibrating in all directions. The choice between the two types of polarizing filters depends on the intensity of reflection from the target observation scene.

[0029] In this embodiment, optical images of a hardware circuit board fixed to the inspection platform can be acquired using an industrial binocular lens. Based on the acquired optical image data, a 3D model is created to obtain the spatial coordinate data of the hardware circuit board relative to the inspection platform. The spatial coordinate data may include 3D angle point cloud data, and the process is as follows: Figure 3 As shown.

[0030] Spatial coordinate data may include three-dimensional angle point cloud data.

[0031] In this embodiment of the disclosure, three-dimensional modeling is performed based on the acquired optical image data to obtain the spatial coordinate data of the hardware circuit board relative to the detection platform, which may specifically include: The industrial binocular lens is calibrated to obtain the intrinsic and extrinsic parameters of the binocular camera; Acquire stereo image pairs of the target scene, combine intrinsic and extrinsic parameters to perform distortion correction and epipolar alignment processing, and calculate pixel depth distance; Based on pixel depth distance and combined with the spatial geometry model inside the binocular camera, the horizontal angle, pitch angle and slant distance of each pixel are calculated to generate three-dimensional angle point cloud data.

[0032] In practical implementation, the industrial binocular lens can first be calibrated. More than 20 sets of checkerboard-like photos can be taken using binocular modeling equipment to obtain the intrinsic and extrinsic parameters of the binocular camera and establish its internal geometric model.

[0033] After calibration, a set of stereo image pairs of the target scene can be acquired. Combined with the intrinsic and extrinsic parameters of the binocular modeling device, distortion correction and epipolar alignment processing are performed on the images, and pixel depth distance is calculated. The geometric model of binocular ranging (i.e., the binocular vision optical center plane ranging model) is as follows: Figure 4 As shown, the working principle is as follows: Target point The emitted light rays travel in a straight line through the optical center of the left lens. With the right lens light center Then, on the left pixel plane Point and the right pixel plane Point imaging. (By...) Figure 1 From the geometric relationships presented, we can see that the triangle With triangle Similarly, a similar triangular relationship can be constructed as shown in equation (1): (1) In the formula, For target point B is the vertical distance to the plane formed by the left and right lenses, where B is the actual distance between the two optical centers of the binocular camera, i.e., the baseline distance. The physical focal length of the camera lens. This is the difference in physical distance between corresponding pixels of the same object in the left and right optical views, on the same reference pixel plane. It represents the actual distance of the target point within the same pixel plane, typically on the order of micrometers. Let... Then, from equation (1), we can obtain the binocular ranging formula: (2) In the formula, For target point The vertical distance to the plane of the binoculars formed by the left and right lenses. This is the actual distance between the two optical centers of the binocular camera, i.e., the baseline distance. The physical focal length of the camera lens. For parallax.

[0034] When the focal length of the lens of the binocular camera system is known Baseline distance The target point can be calculated using equation (2). The perpendicular distance between the target and the plane of the binoculars' center, i.e., the target distance. .

[0035] By obtaining the internal parameters of the left camera through calibration, the projection position of the lens optical center relative to the pixel plane is determined, a geometric model of the camera's internal space is established, and the camera's horizontal and pitch angles are calculated. Simultaneously, the Euclidean distance from each pixel in the pixel plane to the projection point of the lens optical center on the pixel plane is calculated. Using the camera's internal spatial geometric model and the actual depth distance of each pixel, the actual slant distance of each pixel is calculated through similar triangle relationships. Finally, the horizontal angle, pitch angle, and slant distance of each pixel are obtained, i.e., 3D angular point cloud data. Combined with the azimuth and range values ​​of the radar data, the nearest neighbor matching principle is used to map the deformation values ​​corresponding to each azimuth and range direction of the radar to each coordinate in the 3D angular point cloud. Different color values ​​represent different deformation ranges, and the radar deformation values ​​are then displayed on the 3D angular point cloud, achieving data fusion.

[0036] In an ideal physical geometry model of camera imaging, the optical center projection point is located at the exact center of the pixel plane, and the horizontal angle of the theoretical model of the camera's internal space is... With pitch angle Spatial geometric models such as Figure 3-7 As shown. By Figure 5 The geometric triangular relationship between the optical center and the actual pixel plane size allows for the calculation of the overall horizontal angle range of the image. With pitch angle range The number of rows for the pixel arrangement is... The number of rows and columns is Column, combined with horizontal angle range With pitch angle range It can calculate the horizontal and vertical angle coordinates of each pixel relative to the center of the image. .

[0037] Due to manufacturing errors, it's impossible to ensure that the projection of the lens's optical center onto the physical pixel plane is perfectly centered. By calibrating the camera, the position of the optical center on the pixel plane can be determined. Based on this position, a camera system can be constructed... Figure 6 The geometric model shown is based on the actual optical center projected onto the pixel plane. This model can be used to calculate the true horizontal and pitch angles of each pixel of the camera.

[0038] Construct a Cartesian coordinate system with the optical center projection point as the origin, and the pitch direction is... The axis, in the horizontal direction is The horizontal angle of view of the camera is divided into axes. and The former is a negative value, and the latter is a positive value; the pitch angle is divided into... and The former is a positive value, and the latter is a negative value. The coordinates of each pixel are... The pitch angle of each pixel can be calculated by combining its coordinates with the physical size of the pixel. Figure 6 medium triangle The target imaging projection point can be obtained from the right-angled triangle relationship. The physical distance from the optical center can be obtained from the physical distance between each pixel and the optical center through a similar relationship. The formulas for calculating the pixel pitch angle, horizontal angle, and physical distance between the pixel and the optical center are shown in equations (3), (4), and (5): (3) (4) (5) In the formula, The pitch angle of a pixel. The horizontal angle of a pixel. The physical distance of a pixel from the optical center. y is the ordinate of the pixel. is the x-coordinate of the pixel. The physical size of a pixel. The focal length calculated for camera calibration. It is the distance from the center of the pixel plane to the projection point of the optical center.

[0039] The physical distance from the pixel to the optical center is calculated based on the actual optical center projection point geometric model. Combined with the pixel depth data obtained from binocular ranging, the true slant distance of the target point in each pixel can be calculated. The depth distance of each pixel obtained through binocular ranging is R. The constructed slant distance calculation geometric model is as follows: Figure 7 As shown, the known target The row and column number of the pixel can be used to determine this. The physical distance of each corresponding pixel from the optical center has been calculated using equation (6). .according to Figure 7 The geometric projection relationship in the triangle indicates that... With triangle Similarly, the formula for calculating the slant distance from the target T to the optical center can be obtained: (6) In the formula, This represents the binocular ranging depth corresponding to a pixel, i.e., the distance from the target to the optical center plane. The focal length calculated for camera calibration. This represents the actual slant distance from the target to the optical center within the imaged pixel of that point. The physical distance of each pixel from the optical center.

[0040] Having completed the above steps, we have obtained three-dimensional angular point cloud data in the azimuth, pitch, and slant range directions, with the center of the optical image as the coordinate origin. Through camera spatial geometry, the final slant range is obtained using only one parallax calculation. The accuracy of the pitch coordinate data depends only on the accuracy of the calibration calculation determining the optical center projection position, reducing data errors compared to the three-dimensional rectangular coordinates reconstructed by binoculars. This data can be matched and fused with radar data in the azimuth and slant range directions for display. Furthermore, the three-dimensional angular point cloud data can be converted into three-dimensional rectangular coordinate point cloud data and offset to a specific coordinate system for multi-source data composite analysis.

[0041] The three-dimensional angle point cloud data of the target area reconstructed using the above method essentially has the characteristics of spherical coordinates. The final slant distance is obtained using only one parallax calculation, reducing data errors.

[0042] Step 102: Scan the hardware circuit board with a miniature millimeter-wave radar to obtain radar echo data, and construct surface abnormal welding data of each electronic component solder joint based on the radar echo data.

[0043] This step involves scanning the hardware circuit board with a miniature millimeter-wave radar to obtain radar echo data, and then constructing surface anomaly welding data for each electronic component's solder joints based on the radar echo data. The RF subsystem selects the corresponding polarized transmitting and receiving array elements according to the set timing sequence, transmits the transmitted signal and echo signal, completes deskewing processing, and outputs the intermediate frequency signal.

[0044] Step 103: Perform multi-source data fusion processing on spatial coordinate data, surface abnormal welding data, and optical image data to construct a three-dimensional fusion data model for hardware circuit board inspection.

[0045] In this embodiment, the radar data is detected in the horizontal azimuth and slant range, lacking only the elevation azimuth data compared to spherical coordinates. Therefore, when the origin of the 3D angle point cloud coincides with the radar imaging center, the slant range distances of the two data points in the same azimuth direction are matched using the nearest neighbor principle. At this point, each azimuth and range coordinate of the radar data corresponds to the coordinates of the 3D angle point cloud. Data matching can be achieved by recording the data index between corresponding points of the two types of data. Since the data density of point cloud data is usually greater than that of radar data, a single 2D point in radar data may correspond to multiple 3D coordinates. Therefore, deformation data in the azimuth and range directions of the radar can be assigned multiple 3D coordinates. The 3D point cloud typically contains six dimensions of data: three dimensions of coordinate data and three channels of RGB color data. When the radar detects deformation at a specific distance in a certain azimuth, the deformation can be characterized by changing the RGB information of the 3D point cloud data, thus displaying the deformation information detected by the radar on the 3D point cloud. When fusion is successful, the RGB information of the point cloud data can also be changed to display the area monitored by the radar on the 3D point cloud.

[0046] This method first reads radar imaging data to obtain the azimuth and range resolutions of the radar data, as well as the radar two-dimensional imaging data; then it uses three-dimensional angle point cloud data, and then performs feature point matching between the radar two-dimensional imaging data and the three-dimensional point cloud azimuth-range two-dimensional reconstruction map to obtain the point cloud-radar data matching index matrix; finally, it generates a specific data format to complete the data composite.

[0047] In this embodiment, spatial coordinate data, surface abnormal welding data, and optical image data are fused together to construct a three-dimensional fused data model for hardware circuit board inspection. Specifically, this may include: Acquire radar imaging data and perform multi-frame composite preprocessing on the radar imaging data to generate the mean amplitude matrix of the radar to be composited. Based on the radar azimuth resolution and radar range resolution of radar imaging data, the three-dimensional angle point cloud data is mapped into a two-dimensional azimuth-range reconstruction map matrix. The mean amplitude matrix of the radar to be composited is binarized and matched with the azimuth-range two-dimensional reconstruction map matrix to obtain the mapping relationship between radar imaging data and three-dimensional angle point cloud data in the azimuth and range directions. Based on the matching mapping relationship, the surface abnormal welding data is mapped to the corresponding three-dimensional point cloud coordinates of the three-dimensional angle point cloud data, generating hardware circuit board fusion coordinate data information, and completing the construction of the three-dimensional fusion data model.

[0048] This includes performing multi-frame composite preprocessing on radar imaging data to generate the mean amplitude matrix of the radar to be composited, which may specifically include: Read multiple radar imaging complex matrix data from multiple frames of radar imaging data, calculate the modulus matrix of each radar imaging complex matrix data, and sum and average the multiple modulus matrices to obtain the radar imaging amplitude mean matrix. Multiple radar coherence data matrices from multiple frames of radar imaging data are read, and the multiple radar coherence data matrices are summed and averaged to obtain the radar coherence data mean matrix. A radar coherence mean mask matrix is ​​generated based on the radar coherence data mean matrix that is greater than the radar coherence mean threshold. The radar imaging amplitude mean matrix and the radar coherence mean mask matrix are multiplied by a dot product to generate the radar amplitude mean matrix to be composited.

[0049] In specific application scenarios, N frames of radar imaging complex image data can be read to obtain N radar imaging complex matrix data. The modulus matrix of each complex matrix is ​​calculated, and the N radar imaging modulus matrices are summed to obtain the radar imaging amplitude mean matrix. Next, N frames of radar coherence image data are read to obtain N radar coherence data matrices. The N radar coherence data matrices are summed to obtain the radar coherence data mean matrix. A coherence mean threshold is set, and radar coherence mean data greater than this threshold is retained to generate a coherence mean mask matrix. Finally, the radar imaging amplitude mean matrix and the coherence mean mask matrix are multiplied by a dot product to filter out radar imaging amplitude mean data less than the coherence mean threshold, generating the radar amplitude mean matrix to be composited, and extracting the azimuth range and resolution, as well as the range range and resolution.

[0050] Specifically, the radar azimuth resolution and radar range resolution based on radar imaging data map the three-dimensional angle point cloud data into a two-dimensional azimuth-range reconstruction map matrix, which may include: Step 201: Extract the azimuth coordinates of the 3D angle point cloud data. and distance coordinates Determine the maximum value of the azimuth coordinates respectively. and minimum value and the maximum value of the distance coordinates. and minimum value ; Step 202: Based on radar azimuth resolution Radar range resolution The maximum and minimum values ​​of the spherical coordinate azimuth and distance coordinates are calculated, the number of rows and columns of the azimuth-distance 2D reconstruction matrix are calculated, and a zero matrix of size [number of rows and columns] is initialized as the 2D reconstruction matrix. Specifically, this may include: The difference between the maximum and minimum values ​​of the azimuth coordinates is calculated, and the ratio of this difference to the radar azimuth resolution is used as the column value of the azimuth-range two-dimensional reconstruction map matrix. The difference between the maximum and minimum range coordinates is calculated, and the ratio of this difference to the radar range resolution is used as the row value of the azimuth-range two-dimensional reconstruction map matrix. Based on the column and row values, initialize a zero matrix of size equal to the column and row values ​​as a two-dimensional reconstructed graph matrix; Specifically, the maximum value of the azimuth coordinate can be used. Subtract the minimum value of the azimuth coordinate. Divide by radar range resolution The approximate value of the column number is obtained, and after rounding, the column value (row) is obtained. The maximum value of the distance from the coordinates is then used. Subtract the minimum value of the distance coordinates Divide by radar range resolution The approximate value of the number of rows is obtained, and after rounding, the row value col is obtained. A zero matrix of size col rows and row columns is created as the two-dimensional reconstructed graph matrix. Step 203: Traverse all spherical coordinate azimuth-range points in the 3D angle point cloud data. Based on the coordinate values ​​of the spherical coordinate azimuth-range points and the radar azimuth resolution and radar range resolution, determine the row and column numbers of the 2D reconstruction map matrix. Based on the row and column numbers, modify the corresponding position values ​​of the 2D reconstruction map matrix to 1 to generate a binarized azimuth-range 2D reconstruction map matrix. Specifically, based on the coordinates of the spherical coordinate azimuth-range point and the radar azimuth and range resolutions, the row and column numbers of the two-dimensional reconstruction map matrix are determined, which may include: The difference between the spherical coordinate distance point coordinate value and the minimum value of the distance coordinate is calculated, and the row number of the two-dimensional reconstruction map matrix is ​​determined based on the ratio of the difference result to the radar range resolution. The difference between the spherical coordinate azimuth coordinate value and the minimum azimuth coordinate is calculated, and the column number of the two-dimensional reconstruction map matrix is ​​determined based on the ratio of the difference result to the radar azimuth resolution.

[0051] Specifically, the minimum value of the azimuth coordinate can be subtracted from the azimuth coordinate value of a certain point. Then, divide by the radar azimuth resolution. Adding 1 gives the column number of the two-dimensional reconstructed graph matrix corresponding to the point's coordinates. Subtracting the minimum distance value from the distance coordinate value of a point's sphere coordinates... Then, divide by the radar range resolution. Add 1 to obtain the row number of the 2D reconstruction matrix corresponding to the coordinates of that point. Based on the row and column numbers, modify the corresponding position value in the 2D reconstruction matrix to 1. Repeat this process to traverse all spherical coordinate azimuth-distance points, generating a binarized azimuth-distance 2D reconstruction matrix.

[0052] Subsequently, the mean amplitude matrix of the radar to be composited is binarized and matched with the azimuth-range two-dimensional reconstruction map matrix to obtain the mapping relationship between the radar imaging data and the three-dimensional angle point cloud data in the azimuth and range directions. Specifically, this may include: Binarize the radar amplitude mean matrix to obtain the radar imaging amplitude binary matrix; Two-dimensional reconstruction of radar imaging amplitude binary matrix and three-dimensional point cloud Figure 2 Binary image matching processing can be performed on the value matrix. The matching methods can include template matching methods such as Hamming distance matching, structured matching, feature descriptor matching, squared difference matching, normalized cross-correlation matching, or phase correlation matching. By finding the same or similar feature points in the two matrices, the matching mapping of the radar image and the two-dimensional reconstruction map of the spherical coordinate point cloud on the specific azimuth and range directions is obtained, so as to achieve a one-to-one correspondence between the pixel row and column numbers of the radar image and the pixel row and column numbers of the two-dimensional reconstruction map of the three-dimensional point cloud. Finally, based on the matching mapping relationship, the surface abnormal welding data is mapped to the corresponding three-dimensional point cloud coordinates of the three-dimensional angle point cloud data to generate hardware circuit board fusion coordinate data information and complete the construction of the three-dimensional fusion data model; Since the data density of 3D point cloud data is typically higher than that of radar data, a single 2D point in radar data may correspond to multiple 3D coordinates. Therefore, deformation data in the azimuth and range directions of radar can be assigned multiple 3D coordinates. By modifying the RGB information of the 3D point cloud data to characterize the deformation, the deformation information detected by radar can be displayed on the 3D point cloud.

[0053] Step 104: Based on the three-dimensional fusion data model, determine the abnormal location of the radar echo data and determine the coordinate data of the electronic components corresponding to the abnormal location.

[0054] In the constructed 3D fusion data model, the specific location of radar echo data anomalies is determined by identifying the deformation range represented by RGB color information. Combined with the precise electronic component coordinate data of the hardware circuit board recorded in the spatial coordinate data, the location of the target component on the actual circuit board is directly determined, thus identifying the precise electronic component coordinate data of the component at the abnormal location.

[0055] Step 105: Based on the determined electronic component coordinate data, drive the detection probe to move to the abnormal position, and automatically detect the electrical parameters and operating performance parameters of the electronic component at the abnormal position, generating a detection report.

[0056] This step can drive the detection probe to move to the abnormal position based on the determined electronic component coordinate data, and automatically detect the electrical parameters and operating performance parameters of the electronic component at the abnormal position, generating a detection report.

[0057] In practice, based on the precise coordinate data of the electronic components on the hardware circuit board, the detection probe is driven to automatically detect the electrical parameters and performance parameters of the components according to a set process. The detection probe subsystem, under the control of a servo system, performs three-dimensional omnidirectional sliding, precisely connecting to the target detection component, recording parameters such as current and voltage, and ultimately generating a corresponding detection report.

[0058] In summary, according to the hardware multi-mode inspection method provided in this application, compared with the existing technology, this application can acquire optical images of a hardware circuit board fixed on an inspection platform using an industrial binocular lens, perform three-dimensional modeling based on the acquired optical image data, and obtain the spatial coordinate data of the hardware circuit board relative to the inspection platform; scan the hardware circuit board with a miniature millimeter-wave radar to obtain radar echo data, and construct surface abnormal welding data of each electronic component's solder joints based on the radar echo data; perform multi-source data fusion processing on the spatial coordinate data, surface abnormal welding data, and optical image data to construct a three-dimensional fusion data model for hardware circuit board inspection; based on the three-dimensional fusion data model, determine the abnormal position of the radar echo data and determine the coordinate data of the electronic component corresponding to the abnormal position; according to the determined electronic component coordinate data, drive the inspection probe to move to the abnormal position, and automatically inspect the electrical parameters and operating performance parameters of the electronic component at the abnormal position, generating an inspection report.

[0059] Using the above technical solution, this application acquires optical images and performs 3D modeling through an industrial binocular lens, and combines this with micro millimeter-wave radar scanning to obtain welding point data. This process is completely non-contact, requiring no physical connection or soldering operation on the circuit board. Only after the coordinates of the abnormal location are accurately determined through multi-source data fusion does the system drive the detection probe to move to that specific location for detection, avoiding the risks of pad detachment, circuit breakage, or short circuit between adjacent pins caused by manual soldering, thus achieving non-destructive or low-destructive testing.

[0060] This application demonstrates that the entire process, from optical acquisition, radar scanning, data fusion modeling to probe-driven detection, is automated. Based on a 3D fusion data model, the system automatically identifies anomalies and drives the probe, eliminating the need for manual welding and positioning under a microscope. This automated detection replaces tedious manual operations, eliminates the risk of human error, and significantly improves detection efficiency and data reliability. Through the fusion of radar scanning and visual data, the system can locate anomalies in components with shielding or hidden internal structures, achieving a "see-through" effect without removing the shielding, thus realizing online non-destructive testing.

[0061] Based on the above Figure 1 The specific implementation of the method shown in this embodiment provides a hardware multi-mode detection device, such as... Figure 8As shown, the device includes: a data acquisition module 31, a data construction module 32, a data fusion module 33, a data determination module 34, and a data detection module 35; The acquisition module 31 is used to acquire optical images of a hardware circuit board fixed on the inspection platform through an industrial binocular lens, perform three-dimensional modeling based on the acquired optical image data, and obtain the spatial coordinate data of the hardware circuit board relative to the inspection platform. The construction module 32 is used to scan the hardware circuit board with a micro millimeter-wave radar to obtain radar echo data, and to construct surface abnormal welding data of each electronic component solder joint based on the radar echo data. The fusion module 33 is used to perform multi-source data fusion processing on the spatial coordinate data, the surface abnormal welding data and the optical image data to construct a three-dimensional fusion data model for the detection of the hardware circuit board. The determination module 34 is used to determine the abnormal location of the radar echo data based on the three-dimensional fusion data model, and to determine the electronic component coordinate data corresponding to the abnormal location; The detection module 35 is used to drive the detection probe to move to the abnormal position according to the determined coordinate data of the electronic component, and to automatically detect the electrical parameters and operating performance parameters of the electronic component at the abnormal position, and generate a detection report.

[0062] In specific application scenarios, the spatial coordinate data includes three-dimensional angle point cloud data; the acquisition module 31 can be used to calibrate the industrial binocular lens to obtain the intrinsic and extrinsic parameters of the binocular camera; acquire stereo image pairs of the target scene, perform distortion correction and epipolar alignment processing in combination with the intrinsic and extrinsic parameters, and calculate pixel depth distance; based on the pixel depth distance and combined with the spatial geometric model inside the binocular camera, calculate the horizontal angle, pitch angle and slant distance of each pixel to generate the three-dimensional angle point cloud data.

[0063] In specific application scenarios, the fusion module 33 can be used to acquire radar imaging data and perform multi-frame composite preprocessing on the radar imaging data to generate a radar amplitude mean matrix to be composited; based on the radar azimuth resolution and radar range resolution of the radar imaging data, the three-dimensional angle point cloud data is mapped into an azimuth-range two-dimensional reconstruction map matrix; the radar amplitude mean matrix to be composited and the azimuth-range two-dimensional reconstruction map matrix are subjected to binarization matching processing to obtain the mapping relationship between the radar imaging data and the three-dimensional angle point cloud data in the azimuth and range directions; based on the matching mapping relationship, the surface abnormal welding data is mapped to the corresponding three-dimensional point cloud coordinates of the three-dimensional angle point cloud data to generate hardware circuit board fusion coordinate data information, thus completing the construction of the three-dimensional fusion data model.

[0064] In a specific application scenario, the fusion module 33 can be used to read multiple radar imaging complex matrix data from multiple frames of radar imaging data, calculate the modulus matrix of each radar imaging complex matrix data, and perform a summation and averaging operation on the multiple modulus matrices to obtain a radar imaging amplitude mean matrix; read multiple radar coherence data matrices from multiple frames of radar imaging data, perform a summation and averaging operation on the multiple radar coherence data matrices to obtain a radar coherence data mean matrix, generate a radar coherence mean mask matrix based on the radar coherence data mean matrix that is greater than the radar coherence mean threshold; and perform a dot product operation between the radar imaging amplitude mean matrix and the radar coherence mean mask matrix to generate the radar amplitude mean matrix to be composited.

[0065] In specific application scenarios, the fusion module 33 can be used to extract the azimuth coordinates and distance coordinates of the three-dimensional angle point cloud data, and determine the maximum and minimum values ​​of the azimuth coordinates and the distance coordinates respectively; based on the radar azimuth resolution and the radar range resolution, as well as the maximum and minimum values ​​of the azimuth coordinates and the distance coordinates, calculate the number of rows and columns of the azimuth-distance two-dimensional reconstruction map matrix, and initialize a zero matrix of size equal to the number of rows and columns as the two-dimensional reconstruction map matrix; traverse all spherical coordinate azimuth-distance points in the three-dimensional angle point cloud data, determine the row number and column number of the two-dimensional reconstruction map matrix based on the coordinate values ​​of the spherical coordinate azimuth-distance points and the radar azimuth resolution and the radar range resolution, and modify the corresponding position value of the two-dimensional reconstruction map matrix to 1 based on the row number and the column number to generate the binarized azimuth-distance two-dimensional reconstruction map matrix.

[0066] In specific application scenarios, the fusion module 33 can be used to calculate the difference between the maximum and minimum values ​​of the azimuth coordinates, and determine the ratio of the difference to the radar azimuth resolution as the column value of the azimuth-range two-dimensional reconstruction map matrix; calculate the difference between the maximum and minimum values ​​of the range coordinates, and determine the ratio of the difference to the radar range resolution as the row value of the azimuth-range two-dimensional reconstruction map matrix; and initialize a zero matrix with a size equal to the column and row values ​​as the two-dimensional reconstruction map matrix based on the column and row values.

[0067] In specific application scenarios, the fusion module 33 can be used to calculate the difference between the spherical coordinate distance point coordinate value and the minimum value of the distance coordinate, and determine the row number of the two-dimensional reconstruction map matrix based on the ratio of the difference result to the radar range resolution; calculate the difference between the spherical coordinate azimuth coordinate value and the minimum value of the azimuth coordinate, and determine the column number of the two-dimensional reconstruction map matrix based on the ratio of the difference result to the radar azimuth resolution.

[0068] It should be noted that other corresponding descriptions of the functional units involved in the hardware multi-mode detection device provided in this embodiment can be found in [reference]. Figure 1 The corresponding description in [the document] will not be repeated here.

[0069] Based on the above, Figure 1 Accordingly, this embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. Figure 1 The method shown.

[0070] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of this application.

[0071] Based on the above, Figure 1 The method shown, and Figure 8 To achieve the above objectives, the present application also provides an electronic device, comprising a storage medium and a processor; the storage medium for storing a computer program; and the processor for executing the computer program to implement the above-described virtual device embodiments. Figure 1 The method shown.

[0072] Optionally, the aforementioned physical devices may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB interfaces, card reader interfaces, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Wi-Fi interfaces), etc.

[0073] Those skilled in the art will understand that the physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or combine certain components, or have different component arrangements.

[0074] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the aforementioned physical device, supporting the execution of the hardware multi-mode detection program and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software within the hardware multi-mode detection physical device.

[0075] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms, or it can be implemented by hardware. By applying the solution of this embodiment, compared with the existing technology, this application can acquire optical images of a hardware circuit board fixed on a testing platform using an industrial binocular lens, perform three-dimensional modeling based on the acquired optical image data, and obtain the spatial coordinate data of the hardware circuit board relative to the testing platform; scan the hardware circuit board with a miniature millimeter-wave radar to obtain radar echo data, and construct surface abnormal welding data of each electronic component welding point based on the radar echo data; perform multi-source data fusion processing on the spatial coordinate data, surface abnormal welding data, and optical image data to construct a three-dimensional fusion data model for hardware circuit board testing; based on the three-dimensional fusion data model, determine the abnormal position of the radar echo data, and determine the coordinate data of the electronic component corresponding to the abnormal position; according to the determined electronic component coordinate data, drive the detection probe to move to the abnormal position, and automatically detect the electrical parameters and operating performance parameters of the electronic component at the abnormal position, generating a test report.

[0076] Using the above technical solution, this application acquires optical images and performs 3D modeling through an industrial binocular lens, and combines this with micro millimeter-wave radar scanning to obtain welding point data. This process is completely non-contact, requiring no physical connection or soldering operation on the circuit board. Only after the coordinates of the abnormal location are accurately determined through multi-source data fusion does the system drive the detection probe to move to that specific location for detection, avoiding the risks of pad detachment, circuit breakage, or short circuit between adjacent pins caused by manual soldering, thus achieving non-destructive or low-destructive testing.

[0077] This application demonstrates that the entire process, from optical acquisition, radar scanning, data fusion modeling to probe-driven detection, is automated. Based on a 3D fusion data model, the system automatically identifies anomalies and drives the probe, eliminating the need for manual welding and positioning under a microscope. This automated detection replaces tedious manual operations, eliminates the risk of human error, and significantly improves detection efficiency and data reliability. Through the fusion of radar scanning and visual data, the system can locate anomalies in components with shielding or hidden internal structures, achieving a "see-through" effect without removing the shielding, thus realizing online non-destructive testing.

[0078] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprising" or any other variations thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0079] The above are merely specific embodiments of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to these embodiments, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A hardware multi-mode detection method, characterized in that, include: Optical images of a hardware circuit board fixed to a testing platform are acquired using an industrial binocular lens. Based on the acquired optical image data, a 3D model is created to obtain the spatial coordinate data of the hardware circuit board relative to the testing platform. The hardware circuit board is scanned by a miniature millimeter-wave radar to obtain radar echo data, and surface abnormal welding data of each electronic component solder joint is constructed based on the radar echo data. The spatial coordinate data, the surface abnormal welding data, and the optical image data are subjected to multi-source data fusion processing to construct a three-dimensional fusion data model for the detection of the hardware circuit board. Based on the three-dimensional fusion data model, the abnormal location of the radar echo data is determined, and the coordinate data of the electronic components corresponding to the abnormal location is determined. Based on the determined coordinate data of the electronic components, the detection probe is driven to move to the abnormal position, and the electrical parameters and operating performance parameters of the electronic components at the abnormal position are automatically detected, generating a detection report.

2. The method according to claim 1, characterized in that, The spatial coordinate data includes three-dimensional angle point cloud data; The step of performing 3D modeling based on the acquired optical image data to obtain the spatial coordinate data of the hardware circuit board relative to the detection platform includes: The industrial binocular lens is calibrated to obtain the intrinsic and extrinsic parameters of the binocular camera; Acquire stereo image pairs of the target scene, perform distortion correction and epipolar alignment processing in combination with the intrinsic and extrinsic parameters, and calculate pixel depth distance; Based on the pixel depth distance and combined with the spatial geometry model inside the binocular camera, the horizontal angle, pitch angle and slant distance of each pixel are calculated to generate the three-dimensional angle point cloud data.

3. The method according to claim 2, characterized in that, The step of performing multi-source data fusion processing on the spatial coordinate data, the surface abnormal welding data, and the optical image data to construct a three-dimensional fusion data model for hardware circuit board inspection includes: Acquire radar imaging data and perform multi-frame composite preprocessing on the radar imaging data to generate the mean amplitude matrix of the radar to be composited. Based on the radar azimuth resolution and radar range resolution of the radar imaging data, the three-dimensional angle point cloud data is mapped into an azimuth-range two-dimensional reconstruction map matrix; The mean amplitude matrix of the radar to be composited is binarized and matched with the azimuth-range two-dimensional reconstruction map matrix to obtain the mapping relationship between the radar imaging data and the three-dimensional angle point cloud data in the azimuth and range directions. Based on the matching mapping relationship, the surface abnormal welding data is mapped to the corresponding three-dimensional point cloud coordinates of the three-dimensional angle point cloud data to generate hardware circuit board fusion coordinate data information, thus completing the construction of the three-dimensional fusion data model.

4. The method according to claim 3, characterized in that, The step of performing multi-frame composite preprocessing on the radar imaging data to generate the mean amplitude matrix of the radar to be composited includes: Multiple radar imaging complex matrix data from multiple frames of radar imaging data are read, the modulus matrix of each radar imaging complex matrix data is calculated, and the multiple modulus matrices are summed and averaged to obtain the radar imaging amplitude mean matrix. Multiple radar coherence data matrices of multiple frames of radar imaging data are read, and the multiple radar coherence data matrices are summed and averaged to obtain a radar coherence data mean matrix. A radar coherence mean mask matrix is ​​generated based on the radar coherence data mean matrix that is greater than the radar coherence mean threshold. The radar imaging amplitude mean matrix and the radar coherence mean mask matrix are multiplied by a dot product to generate the radar amplitude mean matrix to be composited.

5. The method according to claim 4, characterized in that, The radar azimuth resolution and radar range resolution based on the radar imaging data are used to map the three-dimensional angle point cloud data into a two-dimensional azimuth-range reconstruction map matrix, including: Extract the azimuth coordinates and distance coordinates of the three-dimensional angle point cloud data, and determine the maximum and minimum values ​​of the azimuth coordinates and the distance coordinates respectively; Based on the radar azimuth resolution and the radar range resolution, as well as the maximum and minimum values ​​of the azimuth coordinates and the range coordinates, calculate the number of rows and columns of the azimuth-range two-dimensional reconstruction map matrix, and initialize a zero matrix of size equal to the number of rows and columns as the two-dimensional reconstruction map matrix; Traverse all spherical coordinate azimuth-range points in the three-dimensional angle point cloud data. Based on the coordinate values ​​of the spherical coordinate azimuth-range points and the radar azimuth resolution and radar range resolution, determine the row and column numbers of the two-dimensional reconstruction map matrix. Based on the row and column numbers, modify the corresponding position values ​​of the two-dimensional reconstruction map matrix to 1 to generate the binarized azimuth-range two-dimensional reconstruction map matrix.

6. The method according to claim 5, characterized in that, The step of calculating the number of rows and columns of the azimuth-range two-dimensional reconstruction map matrix based on the radar azimuth resolution, the radar range resolution, and the maximum and minimum values ​​of the azimuth coordinates and the range coordinates, and initializing a zero matrix of size equal to the number of rows and columns as the two-dimensional reconstruction map matrix, includes: The difference between the maximum value and the minimum value of the azimuth coordinate is calculated, and the ratio of the difference to the radar azimuth resolution is determined as the column value of the azimuth-range two-dimensional reconstruction map matrix. The difference between the maximum value and the minimum value of the range coordinate is calculated, and the ratio of the difference to the radar range resolution is determined as the row value of the azimuth-range two-dimensional reconstruction map matrix. Based on the column values ​​and the row values, initialize a zero matrix with a size equal to the column values ​​and the row values ​​as the two-dimensional reconstruction graph matrix.

7. The method according to claim 5, characterized in that, The step of determining the row and column numbers of the two-dimensional reconstruction map matrix based on the coordinate values ​​of the spherical coordinate azimuth-range point and the radar azimuth resolution and radar range resolution includes: The difference between the spherical coordinate distance point coordinate value and the minimum value of the distance coordinate is calculated, and the row number of the two-dimensional reconstruction map matrix is ​​determined based on the ratio of the difference result to the radar range resolution. The difference between the spherical coordinate azimuth coordinate value and the minimum value of the azimuth coordinate is calculated, and the column number of the two-dimensional reconstruction map matrix is ​​determined based on the ratio of the difference result to the radar azimuth resolution.

8. A hardware multi-mode detection device, characterized in that, include: The acquisition module is used to acquire optical images of a hardware circuit board fixed to the inspection platform through an industrial binocular lens, perform three-dimensional modeling based on the acquired optical image data, and obtain the spatial coordinate data of the hardware circuit board relative to the inspection platform. A construction module is used to scan the hardware circuit board with a micro millimeter-wave radar to obtain radar echo data, and to construct surface abnormal welding data of each electronic component solder joint based on the radar echo data. The fusion module is used to perform multi-source data fusion processing on the spatial coordinate data, the surface abnormal welding data, and the optical image data to construct a three-dimensional fusion data model for the inspection of the hardware circuit board. The determination module is used to determine the abnormal location of the radar echo data based on the three-dimensional fusion data model, and to determine the coordinate data of the electronic components corresponding to the abnormal location; The detection module is used to drive the detection probe to move to the abnormal position according to the determined coordinate data of the electronic component, and to automatically detect the electrical parameters and operating performance parameters of the electronic component at the abnormal position, and generate a detection report.

9. An electronic device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the hardware multi-mode detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the hardware multi-mode detection method according to any one of claims 1 to 7.

Images