Performance optimization method of high-order computing power automatic driving FPC (Flexible Printed Circuit) sensing module and related equipment

Abstract

The invention relates to a performance optimization method for a high-order computing power automatic driving FPC sensing module and related equipment, and the method comprises the following steps: scanning a flexible circuit board of the FPC sensing module through a multispectral imaging system, and obtaining multidimensional spectral image data; performing circuit feature extraction on the multi-dimensional spectral image to obtain a circuit wiring feature map; under a given working signal condition, carrying out electromagnetic compatibility analysis on the FPC sensing module based on the circuit wiring characteristic pattern to obtain an electromagnetic interference hotspot distribution pattern; based on the electromagnetic interference hotspot distribution diagram, performing adaptive electromagnetic shielding design on the FPC sensing module to obtain a shielding structure scheme; and optimizing the FPC sensing module based on the shielding structure scheme to obtain a target FPC sensing module, thereby solving the technical problems of sensing data distortion and bit error rate increase caused by the fact that a traditional design method based on an empirical formula or static shielding is difficult to deal with nonlinear interference in high-frequency signal transmission.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of FPC sensing module, and particularly relates to a performance optimization method of high-order computing power automatic driving FPC sensing module and related equipment. BACKGROUND

[0002] Currently, the high-order computing power automatic driving system has high requirements for the precision and reliability of the sensor module, and the electromagnetic compatibility (EMC) performance of the flexible printed circuit (FPC) as the carrier of the core sensing module is directly related to the stability and safety of the automatic driving system. However, with the improvement of circuit integration and the increase of working frequency, the signal integrity problem and cross interference phenomenon of the FPC module in the complex electromagnetic environment are increasingly prominent, and the traditional design method based on empirical formula or static shielding cannot cope with the nonlinear interference in high-frequency signal transmission, resulting in distortion of sensing data, increase of error rate and even system failure. Especially in the automatic driving scene of multi-sensor fusion, the dynamic distribution characteristics of electromagnetic interference (EMI) hotspots make it impossible for the existing technology to realize precise local shielding optimization, which seriously restricts the performance upper limit of the sensor module under high-order computing power. SUMMARY

[0003] The main purpose of the present application is to provide a performance optimization method of high-order computing power automatic driving FPC sensing module, which solves the technical problem that the traditional design method based on empirical formula or static shielding cannot cope with the nonlinear interference in high-frequency signal transmission, resulting in distortion of sensing data and increase of error rate.

[0004] To achieve the above purpose, the present application provides a performance optimization method of high-order computing power automatic driving FPC sensing module, comprising the following steps: scanning the flexible circuit board of the FPC sensing module through a multi-spectral imaging system to obtain multi-dimensional spectral image data; extracting circuit features from the multi-dimensional spectral image to obtain a circuit wiring feature map; under the given working signal condition, performing electromagnetic compatibility analysis on the FPC sensing module based on the circuit wiring feature map to obtain an electromagnetic interference hotspot distribution map; based on the electromagnetic interference hotspot distribution map, performing adaptive electromagnetic shielding design on the FPC sensing module to obtain a shielding structure scheme; optimizing the FPC sensing module based on the shielding structure scheme to obtain a target FPC sensing module.

[0005] Further, the scanning of the flexible circuit board of the FPC sensing module through the multi-spectral imaging system to obtain the multi-dimensional spectral image data comprises: The flexible circuit board of the FPC sensing module is positioned and calibrated through a preset optical alignment device to obtain positioning parameters, and a scanning path of the multispectral imaging system is optimized based on the positioning parameters to obtain an adaptive scanning trajectory map; Based on the adaptive scanning trajectory map, the flexible circuit board of the FPC sensing module is subjected to multi-band synchronous scanning to obtain an original multi-band image data set, and a spectral separation is performed on the original multi-band image data set through a spectral unmixing algorithm to obtain separated visible light, near-infrared and thermal imaging data; Multi-scale image registration is performed on the separated visible light, near-infrared and thermal imaging data to obtain an aligned multi-dimensional spectral image, and dynamic range compression is performed on the aligned multi-dimensional spectral image through an adaptive histogram equalization technique to obtain a multi-dimensional spectral image data.

[0006] Further, the circuit feature extraction on the multi-dimensional spectral image is performed to obtain a circuit wiring feature map, including: Multi-scale wavelet decomposition is performed on the multi-dimensional spectral image to obtain image feature subsets in different frequency domains, and adaptive threshold segmentation is performed on the multi-dimensional spectral image based on the image feature subsets to obtain a preliminary circuit contour map; Noise suppression and edge enhancement are performed on the preliminary circuit contour map to obtain an optimized circuit contour map, and Hough transform is performed based on the optimized circuit contour map to obtain a circuit segment parameter set; Based on the circuit segment parameter set, the circuit topology structure of the flexible circuit board is reconstructed to obtain a preliminary circuit wiring diagram, and the preliminary circuit wiring diagram is geometrically corrected; Edge detection is performed on the geometrically corrected preliminary circuit wiring diagram to obtain circuit edge positioning data, and circuit width and spacing measurement is performed based on the circuit edge positioning data to obtain a circuit wiring feature map.

[0007] Further, the adaptive threshold segmentation on the multi-dimensional spectral image based on the image feature subsets to obtain a preliminary circuit contour map includes: Local entropy value calculation is performed on the image feature subsets to obtain a circuit region gray scale distribution feature map, and multi-level clustering is performed based on the circuit region gray scale distribution feature map to obtain a circuit region hierarchical data set; Region labeling is performed on the circuit region hierarchical data set to obtain a circuit contour region label map; Two-dimensional curvature analysis is performed on the circuit contour region label map to obtain circuit surface curvature feature data, and local optimal threshold calculation is performed based on the circuit surface curvature feature data to obtain a dynamic threshold mapping matrix; Segment the circuit contour region label map based on the dynamic threshold mapping matrix to obtain a circuit region binary result, and perform morphological post-processing on the circuit region binary result to obtain a preliminary circuit contour map.

[0008] Further, under the given working signal condition, the FPC sensing module is analyzed for electromagnetic compatibility based on the circuit wiring feature map to obtain an electromagnetic interference hotspot distribution map, including: An S parameter measurement is performed on the FPC sensing module under the given working signal condition by a preset vector network analyzer to obtain a frequency domain scattering parameter matrix, and time domain gating transformation is performed on the frequency domain scattering parameter matrix to obtain time domain reflection characteristic data; The geometric parameters of the conductors of the circuit wiring feature map are extracted to obtain a three-dimensional conductor structure parameter set, and dipole current distribution modeling is performed based on the three-dimensional conductor structure parameter set to obtain an equivalent current source distribution map; Near-field coupling calculation is performed on the equivalent current source distribution map by the method of moments to obtain spatial electromagnetic field intensity distribution data; Transmission line crosstalk analysis is performed on the FPC sensing module based on the time domain reflection characteristic data to obtain an inter-line coupling coefficient matrix, and far-field radiation transformation is performed on the spatial electromagnetic field intensity distribution data to obtain a radiation emission power spectral density map; The eigenvalues of the inter-line coupling coefficient matrix are decomposed to obtain main crosstalk mode parameters, and ground plane common mode current inversion calculation is performed based on the radiation emission power spectral density map to obtain a common mode current distribution vector, and the main crosstalk mode parameters and the common mode current distribution vector are weighted and fused by the superposition principle to obtain the electromagnetic interference hotspot distribution map.

[0009] Further, the transmission line crosstalk analysis of the FPC sensing module based on the time domain reflection characteristic data to obtain an inter-line coupling coefficient matrix includes: The time domain reflection characteristic data is differentiated to obtain a reflection waveform slope map, and impedance discontinuity detection is performed on the flexible circuit board of the FPC sensing module based on the reflection waveform slope map to obtain an impedance change node distribution map; The physical structure of the impedance change node distribution map is reconstructed by geometric inversion technology to generate a conductor geometric deformation distribution map, and electric field distribution calculation is performed based on the conductor geometric deformation distribution map in combination with the circuit wiring feature map to obtain an inter-conductor electric field coupling strength map; Green function integration is performed on the inter-conductor electric field coupling strength map to obtain a mutual inductance coefficient distribution matrix, and the crosstalk transmission path of the transmission line in the FPC sensing module is analyzed based on the mutual inductance coefficient distribution matrix to obtain a crosstalk propagation characteristic vector; Perform matrix eigenvalue decomposition based on the crosstalk propagation eigenvector to obtain an inter-line coupling coefficient matrix, wherein the inter-line coupling coefficient matrix comprises near-end crosstalk coefficients, far-end crosstalk coefficients and frequency domain coupling responses.

[0010] Further, based on the electromagnetic interference hotspot distribution map, the FPC sensing module is adaptively designed for electromagnetic shielding to obtain a shielding structure scheme, including the following steps: Based on the electromagnetic interference hotspot distribution map, interference source positioning analysis is performed on the FPC sensing module to obtain an interference source coordinate set, and a three-dimensional electromagnetic field intensity gradient map is constructed based on the interference source coordinate set; Based on the three-dimensional electromagnetic field intensity gradient map, an initial shielding structure layout is designed, and finite element analysis is performed on the initial shielding structure layout to obtain shielding effectiveness evaluation data; Based on the shielding effectiveness evaluation data, the initial shielding structure layout is multi-objective optimized to obtain an improved shielding structure design, and the improved shielding structure design is verified through electromagnetic wave propagation simulation technology to obtain a shielding performance matrix; The shielding performance matrix is subjected to sensitivity analysis to obtain a key design parameter set, and a shielding structure scheme is determined based on the key design parameter set.

[0011] The application also provides a high-order computing power automatic driving FPC sensing module performance optimization device, comprising: The scanning module is used for scanning the flexible circuit board of the FPC sensing module through a multi-spectral imaging system to obtain multi-dimensional spectral image data; The extraction module is used for extracting circuit features from the multi-dimensional spectral image to obtain a circuit wiring feature map; The analysis module is used for performing electromagnetic compatibility analysis on the FPC sensing module based on the circuit wiring feature map under a given working signal condition to obtain an electromagnetic interference hotspot distribution map; The design module is used for adaptively designing electromagnetic shielding for the FPC sensing module based on the electromagnetic interference hotspot distribution map to obtain a shielding structure scheme; The optimization module is used for optimizing the FPC sensing module based on the shielding structure scheme to obtain a target FPC sensing module.

[0012] The application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to realize the steps of the method of any one of the above.

[0013] The application also provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to realize the steps of the method of any one of the above.

[0014] The application provides a performance optimization method of a high-order computing power automatic driving FPC sensing module, including the following steps: scanning a flexible circuit board of the FPC sensing module through a multi-spectral imaging system to obtain multi-dimensional spectral image data; performing circuit feature extraction on the multi-dimensional spectral image to obtain a circuit wiring feature map; under a given working signal condition, performing electromagnetic compatibility analysis on the FPC sensing module based on the circuit wiring feature map to obtain an electromagnetic interference hotspot distribution map; based on the electromagnetic interference hotspot distribution map, performing adaptive electromagnetic shielding design on the FPC sensing module to obtain a shielding structure scheme; and optimizing the FPC sensing module based on the shielding structure scheme to obtain a target FPC sensing module, which solves the technical problem that a traditional design method based on an empirical formula or static shielding is difficult to cope with nonlinear interference in high-frequency signal transmission, resulting in distorted sensing data and rising bit error rate, and realizes the technical effect that in a variable working environment of automatic driving, the method can adjust the shielding strategy in real time according to different signal loads and frequency conditions, ensures that the FPC sensing module can maintain the best performance in various electromagnetic environments, and improves the environmental adaptability of the system. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a step schematic diagram of the performance optimization method of the high-order computing power automatic driving FPC sensing module in an embodiment of the application. Figure 2 is a structural block diagram of a performance optimization device of the high-order computing power automatic driving FPC sensing module in an embodiment of the application. Figure 3 is a structural schematic block diagram of a computer device of an embodiment of the application.

[0016] The purposes, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0017] In order to make the purposes, technical solutions and advantages of the application clearer, further detailed description will be made to the application by combining the embodiments and the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the application and do not limit the application.

[0018] As shown in Figure 1 , Figure 1 is a performance optimization method of a high-order computing power automatic driving FPC sensing module in an embodiment of the application, including the following steps: Step S1, scanning a flexible circuit board of the FPC sensing module through a multi-spectral imaging system to obtain multi-dimensional spectral image data.

[0019] Specifically, the flexible circuit board of the FPC sensing module is scanned by the multispectral imaging system, and the core is to obtain high-precision circuit feature data by using spectral information of different wavelengths, thereby providing a basis for subsequent electromagnetic compatibility analysis and shielding optimization. In specific implementation, the multispectral imaging system can use visible light, infrared, ultraviolet and other multi-band light sources to scan the FPC surface point by point or regionally. Since the conductors, insulating layers and pads of the flexible circuit board are made of different materials, they have different reflection, absorption or transmission characteristics for different wavelengths of light. Therefore, by multispectral imaging, multi-dimensional spectral image data containing circuit layout, material distribution and potential defects can be obtained. For example, in the automatic driving FPC sensing module, the insulating layer near the high-frequency signal line may be micron-level deformed due to electromagnetic heat effect, and the ultraviolet spectrum can sensitively capture such subtle changes, and the infrared spectrum can reflect the temperature distribution of the circuit when it is working. These data together constitute a multi-dimensional spectral image data set. Further, the data not only contains spatial dimension circuit topography information, but also contains multi-dimensional features such as material characteristics and working state, thereby providing comprehensive and high-resolution input for subsequent circuit feature extraction. In actual application, when the FPC module is in a complex electromagnetic environment, the multispectral imaging system can cooperate with the automatic scanning platform to perform rapid global detection on the module. For example, when the vehicle-mounted radar is working at high frequency, the system can dynamically monitor the changes of the electromagnetic sensitive area of the circuit board by real-time acquisition of spectral data of different wavebands, ensure that the subsequent analysis is based on the latest working condition data, and further improve the accuracy of electromagnetic interference hotspot identification. Compared with traditional monocular vision or electrical detection, this multispectral scanning method can detect potential EMI risk points earlier, such as identifying local heating areas caused by high-frequency resonance in millimeter wave radar FPC, thereby providing accurate spatial positioning basis for adaptive shielding design.

[0020] Step S2, circuit feature extraction is performed on the multi-dimensional spectral image to obtain a circuit wiring feature map.

[0021] Specifically, the key to circuit feature extraction from multi-spectral images lies in using image processing algorithms to separate key features directly related to circuit wiring from multi-band data, thereby constructing high-precision circuit wiring feature maps. In specific implementation, first, the images collected in different spectral bands need to be registered and fused to eliminate spatial errors caused by deformation of flexible substrates or device jitter during scanning, ensuring that the data in each band is strictly aligned in space. Subsequently, a semantic segmentation algorithm based on deep learning, such as the U-Net network structure, is used to classify the fused multi-dimensional spectral images at the pixel level, accurately segmenting different functional areas such as conductor tracks, pad areas, and insulating media. At the same time, the potential hot spot areas of high-frequency signal lines are identified by combining the thermal distribution data of the infrared band. Since the automatic driving FPC sensor module usually adopts a high-density interconnection design, special attention needs to be paid to the accurate restoration of micron-level line width and spacing during extraction, such as sub-pixel level edge detection of the differential signal line pair of the 77GHz millimeter wave radar interface circuit to avoid false merging of adjacent conductors in the feature map. After feature extraction is complete, the system generates a vectorized circuit wiring feature map containing geometric parameters, material properties, and electromagnetic characteristic labels. This feature map not only preserves the topological structure of the original wiring, but also labels the working frequency range of different functional blocks, such as differentiating the wide track of the power management module from the impedance control area of the high-speed SerDes interface. Taking the FPC of the automatic driving domain controller as an example, when processing the flexible circuit of the multi-camera data fusion interface, the circuit wiring feature map accurately presents the serpentine structure of the LVDS signal line and labels its spatial position relative to the reference ground layer. These structured data provide accurate geometric and material parameter inputs for subsequent electromagnetic compatibility analysis.

[0022] Step S3, under given working signal conditions, performing electromagnetic compatibility analysis on the FPC sensor module based on the circuit wiring feature map to obtain an electromagnetic interference hot spot distribution map.

[0023] Specifically, the process of electromagnetic compatibility analysis based on the circuit layout feature map is realized by establishing an accurate electromagnetic field simulation model, in which the circuit layout feature map provides key geometric parameter and material characteristic input. First, the vectorized circuit layout feature map needs to be imported into the electromagnetic simulation software, and at the same time, the corresponding excitation signal parameters are set according to the actual working conditions of the autonomous driving FPC sensor module, for example, in the millimeter wave radar application, a 77GHz sinusoidal excitation signal needs to be set, and the load impedance characteristics of each functional block are defined. During the simulation process, the finite element method is used to calculate the electromagnetic field distribution in three-dimensional space, focusing on analyzing typical EMI problems such as crosstalk effect between high-frequency signal lines, power plane resonance, and discontinuity of signal return path, and combining with the material conductivity and dielectric constant parameters marked in the circuit layout feature map, the influence of flexible substrate on electromagnetic wave propagation is accurately simulated. In the simulation case of the autonomous driving laser radar control FPC, when simulating the pulse driving signal of a 16-line laser radar, the simulation result will show the standing wave effect caused by impedance mismatch in the long-distance wiring between the control chip and the laser diode, and these areas will be marked as red highlighted areas in the electromagnetic interference hotspot distribution map. In addition, the analysis process also needs to consider the influence of dynamic working conditions, for example, when multiple sensors are enabled simultaneously in the autonomous driving system, the combined interference generated by the mutual modulation of signals of different frequencies needs to be simulated, and the final electromagnetic interference hotspot distribution map not only contains spatial position information, but also marks the interference frequency components and intensity levels corresponding to each hotspot, providing quantitative basis for subsequent adaptive electromagnetic shielding design. Through this detailed simulation based on actual working conditions, potential interference problems that traditional experience design cannot find can be effectively identified, for example, in the 4D imaging radar FPC, the intermediate frequency band resonance phenomenon caused by improper ground plane segmentation is found, thereby significantly improving the EMC reliability of high-order autonomous driving systems.

[0024] Step S4, based on the electromagnetic interference hotspot distribution map, adaptive electromagnetic shielding design is performed on the FPC sensor module to obtain a shielding structure scheme.

[0025] Specifically, the process of adaptive electromagnetic shielding design based on the electromagnetic interference hotspot distribution map is realized through an intelligent material layout algorithm. The algorithm performs multi-objective optimization on the interference intensity, frequency characteristics marked in the electromagnetic interference hotspot distribution map, and the mechanical constraint conditions of the FPC sensor module. In specific implementation, the electromagnetic interference hotspot distribution map needs to be first converted into a three-dimensional shielding requirement matrix, where each voxel contains the shielding effectiveness index required at that spatial location and the tolerable mechanical stress range. Then, combined with the bendable characteristics of the flexible circuit board, a topological optimization algorithm is used to plan the deposition path of the conductive material on the FPC surface, so that the shielding structure meets the electromagnetic protection requirements while not affecting the flexibility of the module. For example, in the design of an automatic driving multi-camera data fusion FPC, for the high-frequency crosstalk hotspots identified between the image processor and the high-speed serial interface, the system will generate a local grid-shaped conductive polymer shielding layer. This structure maintains a 50μm continuous conductive layer in the hotspot area to achieve a near-field shielding effectiveness of more than 30dB, while in the non-critical area, a discrete island-shaped structure is used to retain the bending performance. For interference hotspots with different frequency characteristics, the shielding design scheme will differentially select material combinations. For example, in a millimeter wave radar FPC, nano-silver wire filled dielectric composite material is used for 77GHz interference, while ferrite-silica composite layer is used for power noise to achieve wideband absorption. The feasibility of mass production process also needs to be considered in the design process. By converting the shielding structure scheme into laser direct writing or inkjet printing processing path data, it is ensured that complex shielding patterns can be accurately transferred to the flexible substrate. Taking an automatic driving domain controller FPC as an example, the final shielding structure scheme will contain the spatial arrangement information of multiple heterogeneous materials, which can ensure the integrity of high-speed signals while enabling the module to withstand repeated bending during vehicle door wiring installation. This adaptive design based on accurate electromagnetic analysis can reduce the weight by more than 40% compared to traditional global shielding, while increasing the shielding effectiveness of the key area by 15dB.

[0026] Step S5, optimizing the FPC sensor module based on the shielding structure scheme to obtain a target FPC sensor module.

[0027] Specifically, the final optimization process of the FPC sensing module based on the shielding structure scheme is realized by combining precise manufacturing process with real-time quality verification, and the shielding structure scheme provides accurate material layout and process parameter guidance. In specific implementation, it is necessary to first convert the digitized shielding structure scheme into executable manufacturing instructions, and then use high-precision inkjet printing equipment to deposit conductive silver paste on the surface of the flexible circuit board according to the predetermined pattern, and at the same time, use laser micromachining technology to accurately align and sinter the multi-layer shielding structure, so as to ensure that the nanoscale conductive material can accurately cover the key areas of the electromagnetic interference hotspot distribution map. For example, in the manufacturing process of the automatic driving laser radar control FPC, the system will control the print head to deposit a gradient-changing conductive polymer with a precision of 10 μm, which can maintain a characteristic impedance of 50 Ω while achieving a local shielding effectiveness of 30 dB. The adjacent flexible area is left with sufficient deformation allowance to adapt to the vehicle vibration environment. After manufacturing, the shielding effect needs to be verified by an in-situ detection system. The surface of the optimized module is scanned using a near-field probe, and the measured electromagnetic radiation data is compared with the simulation expectation. If necessary, the performance can be fine-tuned by iterating the process parameters. Taking the 4D imaging radar FPC as an example, the final target FPC sensing module exhibits uniform radiation characteristics in the 77 GHz operating frequency band, with a signal-to-noise ratio of the key signal line improved by more than 12 dB, and the overall weight reduced by 35% compared with the traditional global shielding scheme, fully meeting the dual requirements of "lightweight and high reliability" of the automatic driving system for sensor modules. This optimization method based on precise electromagnetic design-manufacturing closed loop significantly improves the EMC performance of high-order computing power automatic driving systems.

[0028] In specific embodiments, the scanning of the flexible circuit board of the FPC sensing module by the multi-spectral imaging system to obtain multi-dimensional spectral image data includes: Positioning and calibration of the flexible circuit board of the FPC sensing module by a pre-set optical alignment device to obtain positioning parameters, and optimization of the scanning path of the multi-spectral imaging system based on the positioning parameters to obtain an adaptive scanning trajectory map; Multi-band synchronous scanning of the flexible circuit board of the FPC sensing module based on the adaptive scanning trajectory map to obtain an original multi-band image data set, and spectral separation of the original multi-band image data set by a spectral unmixing algorithm to obtain separated visible light, near-infrared and thermal imaging data; Multi-scale image registration of the separated visible light, near-infrared and thermal imaging data to obtain aligned multi-dimensional spectral images, and dynamic range compression of the aligned multi-dimensional spectral images by an adaptive histogram equalization technique to obtain multi-dimensional spectral image data.

[0029] Specifically, the process of scanning the flexible circuit board of the FPC sensing module by the multispectral imaging system and obtaining multi-dimensional spectral image data is a highly integrated intelligent detection process. The core technology is to realize omnidirectional feature capture of the flexible circuit board by combining multi-modal optical collaborative scanning and intelligent data processing. In specific implementation, first, the FPC needs to be accurately positioned by a pre-set optical alignment device. The device usually uses a high-precision laser displacement sensor array (resolution up to 1 μm) in combination with a machine vision system to establish six-degree-of-freedom positioning parameters including X / Y / Z axis positions and pitch / yaw angles in a three-dimensional space. These parameters are not only used to compensate for the installation deviation of the FPC due to its flexible nature, but more importantly, they guide the intelligent planning of the subsequent scanning path. For example, in the detection of the FPC of an automatic driving millimeter wave radar, when the system identifies that the circuit board has a local warping of 0.5 mm, it will adjust the Z-axis stroke of the scanning head in real time based on the positioning parameters to ensure that the optical probe maintains a constant working distance of 50 μm from the board surface. At the same time, based on the three-dimensional curved surface topology, the scanning density of the key signal trace area is increased to 200 DPI, while the non-functional area is reduced to 100 DPI. This adaptive scanning density distribution strategy can shorten the overall scanning time by 40% while ensuring data quality. After obtaining the adaptive scanning trajectory map, the multispectral imaging system starts the multi-band synchronous scanning process. The system usually integrates three detectors: a visible light CCD (400-700 nm), a near-infrared InGaAs sensor (900-1700 nm), and a thermal imaging microbolometer (8-14 μm). Through a beam splitter prism, the three-band optical paths are coaxially aligned to ensure that the spatial consistency error of each band image is less than 2 pixels. Taking the control FPC of an automatic driving laser radar as an example, when scanning the 77 GHz radio frequency circuit area, the system synchronously collects three-band data at a line scanning rate of 500 Hz. The visible light band can clearly distinguish 10 μm wide gold traces, the near-infrared band can penetrate the polyimide substrate to detect the oxidation defects of the underlying copper foil, and the thermal imaging data can detect a local temperature rise difference of 0.1 °C under the condition of applying a 10 mA test current. When the original multi-band image data set is processed by a spectral unmixing algorithm, a least squares-based linear unmixing model is used to decompose the mixed spectrum of each pixel into the proportion of three end-member components such as base material, conductive layer, and solder. This process can achieve an identification accuracy of over 95% for silver paste and solder paste materials that cannot be distinguished by traditional RGB images. In the multi-scale image registration link, a pyramid matching algorithm based on feature points is used. First, SIFT feature points (density of about 50 points / mm²) are extracted at the edges of the circuit in the visible light image. Then, the low-resolution features (100 μm / pixel) of the thermal imaging data are spatially aligned with the high-resolution features (5 μm / pixel) of the visible light through affine transformation. The registration error is controlled within 1 thermal imaging pixel.In the detection case of the automatic driving domain controller FPC, when processing a 20-layer high-density interconnection board, the system can accurately map the 3℃ abnormal temperature rise area discovered by thermal imaging to the specific via location of the visible light image, with a positioning accuracy of ±25μm. The multi-dimensional spectral image after registration also needs to undergo adaptive histogram equalization processing. This technology uses differentiated gray scale expansion strategies for different waveband characteristics: the visible light waveband uses the CLAHE algorithm with a limited contrast ratio (2% clipping limit), the near-infrared waveband uses illumination compensation based on the Retinex theory, and the thermal imaging data implements dynamic range compression (mapping 14-bit raw data to an 8-bit display range). After these processes, the 5μm-level micro-cracks that are difficult to discover in ordinary optical detection show a 10:1 signal-to-noise ratio improvement in the enhanced visible light image, the near-infrared data clearly shows 0.1mm² micro-bubble defects in the dielectric layer, and the thermal imaging data can distinguish 0.05℃ temperature gradient changes. The final generated multi-dimensional spectral image data actually constitutes a comprehensive feature database containing geometric topography, material composition, and thermodynamic characteristics. In the quality detection of automatic driving ultrasonic sensor FPC, the system can complete the scanning of a 200mm×150mm-sized circuit board within 30 seconds, outputting a visible light circuit image containing 20 million pixels, a near-infrared material distribution map containing 5 million pixels, and a thermal image containing 200,000 pixels, with a total data volume of about 800MB. These data not only clearly present the 30μm line width of high-frequency signal traces (edge sharpness above 90%), but also mark potential defect areas with dielectric constant fluctuations of ±0.2, and further predict EMI risk points that may cause a 5℃ temperature rise under maximum working current, providing an unprecedented multi-physical field observation dimension for subsequent circuit feature extraction and electromagnetic compatibility analysis. Through this fusion of multi-spectral information three-dimensional digital modeling, the "blind area" problem caused by flexible deformation, material heterogeneity, and electromagnetic-thermal coupling effects in traditional FPC detection is systematically solved, enabling the reliability verification of high-order automatic driving sensor modules to upgrade from traditional sampling detection to full-parameter digital traceability, and improving the early identification rate of potential faults from the industry average of 85% to 99.7%.

[0030] In specific embodiments, the circuit feature extraction on the multi-dimensional spectral image to obtain a circuit wiring feature map comprises: performing multi-scale wavelet decomposition on the multi-dimensional spectral image to obtain image feature subsets in different frequency domains, and performing adaptive threshold segmentation on the multi-dimensional spectral image based on the image feature subsets to obtain a preliminary circuit contour map; performing noise suppression and edge enhancement on the preliminary circuit contour map to obtain an optimized circuit contour map, and performing Hough transform based on the optimized circuit contour map to obtain a circuit line segment parameter set; reconstructing a circuit topology of the flexible circuit board based on the set of circuit segment parameters to obtain a preliminary circuit layout and geometrically correcting the preliminary circuit layout; performing edge detection on the geometrically corrected preliminary circuit layout to obtain circuit edge positioning data and performing circuit width and spacing measurement based on the circuit edge positioning data to obtain a circuit layout feature map.

[0031] Specifically, the process of circuit feature extraction and final generation of circuit layout feature map from multi-dimensional spectral images is a multi-level intelligent image processing flow, and its core technology is to realize the accurate digital reconstruction of flexible circuit board layout features through the combination of multi-scale feature analysis and geometric modeling. In specific implementation, first of all, multi-scale wavelet decomposition processing needs to be performed on the multi-dimensional spectral images. This process adopts 5-layer two-dimensional discrete wavelet transform to decompose the original image (for example, 5000×4000 pixel visible light data) into 16 sub-bands containing high-frequency details and low-frequency contours. Among them, the first layer high-frequency sub-band (scale 1.2 μm) can capture the microscopic burrs of the wire edge, the third layer medium-frequency sub-band (scale 6.5 μm) is suitable for extracting standard wire features, and the fifth layer low-frequency sub-band (scale 32 μm) is used for analyzing the large-area copper foil area. In the detection case of automatic driving millimeter wave radar FPC, this multi-scale analysis can simultaneously identify 10 μm level edge defects of 77 GHz radio frequency lines and millimeter level corrosion spots of power planes. Through an adaptive threshold algorithm (window size 15×15 pixels), each sub-band is locally binarized, and the finally synthesized preliminary circuit contour map can retain 99.2% of the effective circuit features while reducing the background noise to less than 3% of the original image. After obtaining the preliminary circuit contour map, noise suppression process based on morphological processing is needed. This step uses 3 times of opening operation (structure element 3×3 circle) to eliminate isolated noise pixels, and cooperates with 2 times of closing operation to fill in the broken line gaps below 5 μm, so that the continuity of the wire edge is improved to 98.5%. In the edge enhancement link, the strategy of combining Sobel operator and Canny detection is used. In the processing of 12-layer high-density interconnection board of automatic driving laser radar control FPC, this scheme can improve the edge sharpness of 30 μm line width high-speed differential signal line from 0.65 of the original image to 0.92 (normalized index). The optimized circuit contour map then enters the Hough transform processing stage. This algorithm sets the θ step size to 1° and the ρ step size to 0.5 pixels, and can accurately detect straight line segments with a length of more than 500 μm in a 200 mm×150 mm FPC area. The detection of automatic driving domain controller FPC shows that this process can extract more than 20,000 circuit segment parameters with a position accuracy of ±2 μm and an angle accuracy of ±0.5°, including the start / end point coordinates (resolution 0.5 μm) and the strike angle of each wire. When reconstructing the circuit topology structure based on the circuit segment parameter set, the system will use a connectivity analysis algorithm based on graph theory. First, an adjacency graph between line segment endpoints is established (search radius 10 μm), and then the missing connection nodes are filled in through Delaunay triangulation. In the processing of automatic driving multi-camera interface FPC, this algorithm can automatically splice 2000 scattered line segments into 500 complete circuit traces, with a reconstruction accuracy of 99.8%.The geometric correction link uses the three-dimensional curved surface data obtained by the early multispectral scanning to compensate the two-dimensional projection deformation to the original curved surface through bilinear interpolation. For example, when processing a curved FPC with a curvature radius of 50 mm, the correction can reduce the actual measurement error of a wire with a length of 100 mm in three-dimensional space from ±200 μm without correction to ±15 μm. The edge detection and size measurement stage adopts a sub-pixel level edge positioning technology, and the image resolution is virtually increased by 4 times through cubic spline interpolation, so that a repeat measurement accuracy of ±0.1 μm can be realized under the condition of 100 times optical magnification. When measuring the width of the circuit, the system will densely sample along the normal direction of the wire at a sampling rate of 50 points / μm, and automatically count the size difference between the narrowest and widest places. For example, in the detection of the FPC of an automatic driving ultrasonic sensor, it is found that the actual width of the power line with a nominal value of 35 μm fluctuates between 32.5-37.8 μm. Such nanometer-level precision measurement provides key parameters for subsequent electromagnetic analysis. The inter-wire spacing analysis adopts the Voronoi diagram algorithm to automatically divide the influence area of each wire, and can identify potential short-circuit risk points with a minimum spacing of 8 μm in a dense bus area with a spacing of 0.2 mm. The finally generated circuit layout feature map is actually a structured database containing vector graphics and attribute data. In a typical 200 mm x 150 mm automatic driving sensor FPC, the map contains about 15000 graphic objects, records the width value (accuracy 0.1 μm) of each trace, the spacing between adjacent wires (accuracy 1 μm), the curvature radius at the turning point (accuracy 5 μm), and more than 300 characteristic parameters, with a total data amount of about 120 MB. Such high-precision digital expression enables subsequent electromagnetic compatibility analysis to be based on actual physical dimensions rather than design values, reducing the ±15% parameter uncertainty in traditional design to within ±1.5%, providing unprecedented circuit feature fidelity for high-order automatic driving systems. Through such all-around feature extraction from macro layout to micro size, the performance prediction accuracy of flexible circuit boards in complex electromagnetic environments has been qualitatively improved, and the signal integrity simulation error of a 77 GHz vehicle-mounted radar FPC has been reduced from the industry typical 20% to below 3%, significantly improving the reliability threshold of the automatic driving system under extreme working conditions.

[0032] In specific embodiments, the adaptive threshold segmentation of the multi-dimensional spectral image based on the image feature subset to obtain a preliminary circuit contour map comprises: The local entropy value of the image feature subset is calculated to obtain a circuit region gray scale distribution feature map, and a multi-level clustering is performed based on the circuit region gray scale distribution feature map to obtain a circuit region hierarchical data set; The circuit region hierarchical data set is regionally marked to obtain a circuit contour region marking map; performing two-dimensional curvature analysis on the circuit contour region label map to obtain circuit surface curvature feature data, and performing local optimal threshold calculation based on the circuit surface curvature feature data to obtain a dynamic threshold mapping matrix; segmenting the circuit contour region label map based on the dynamic threshold mapping matrix to obtain a circuit region binary result, and performing morphological post-processing on the circuit region binary result to obtain a preliminary circuit contour map.

[0033] Specifically, the process of adaptive threshold segmentation of multi-dimensional spectral images based on image feature subsets to obtain the preliminary circuit outline is a meticulous and multi-level processing flow, and its purpose is to accurately extract the circuit outline on the circuit board. First, local entropy value calculation is performed on the image feature subsets. This process aims to evaluate the gray scale distribution in each local region, thereby generating a circuit region gray scale distribution feature map. In this step, by analyzing the subbands in different frequency domains, the micro details of the wire edges to various features of the large-area copper foil region can be identified. For example, in a 5000x4000 pixel visible light data, the gray scale variation features of the wire edges can be described in detail using local entropy value calculation, which is crucial for subsequent steps. Next, multi-level clustering is performed based on the obtained circuit region gray scale distribution feature map, with the goal of classifying regions with similar gray scale distribution features into one category to form a circuit region hierarchical data set. In this process, by setting different clustering parameters and conditions, different parts of the circuit can be effectively separated. For example, in the case of detecting the FPC of an autonomous driving millimeter wave radar, this multi-level clustering method can distinguish between the tiny defects of the 77GHz radio frequency line and the large-area corrosion spots of the power plane, ensuring effective classification of the circuit region. Subsequently, region labeling is performed on the circuit region hierarchical data set, which generates a circuit outline region label map. In this map, each labeled region represents a circuit part with specific gray scale distribution features. For example, in a 200mmx150mm FPC region, after region labeling, each straight line segment with a length of more than 500μm can be clearly labeled, which is particularly important for subsequent analysis. Next, two-dimensional curvature analysis is performed to obtain the curvature feature data of the circuit surface, and based on this, the local optimal threshold is calculated to obtain a dynamic threshold mapping matrix. The dynamic threshold mapping matrix mentioned here not only contains the curvature measurement value of the circuit, the stress distribution parameter, but also considers the deformation compensation coefficient. For example, when processing a curved FPC with a curvature radius of 50mm, through two-dimensional curvature analysis, the actual shape of a 100mm long wire in three-dimensional space can be accurately measured, and the threshold is adjusted accordingly, making the final binary result more accurate. Finally, based on the dynamic threshold mapping matrix, the circuit outline region label map is segmented to obtain the circuit region binary result, and morphological post-processing is performed to remove noise and enhance edges, ultimately forming a preliminary circuit outline map. For example, when processing a 12-layer high-density interconnection board of an autonomous driving laser radar control FPC, through the above series of processing, not only can the 30μm line width of the high-speed differential signal line be clearly outlined, but also the edge sharpness can be improved from 0.65 of the original image to 0.92 (normalized index), greatly improving the clarity and accuracy of the circuit outline.The whole process realizes the accurate extraction of the circuit profile in the multi-dimensional spectral image by combining local entropy calculation, multi-level clustering, region marking, two-dimensional curvature analysis and morphological post-processing. Especially in the high-performance flexible circuit board in the complex electromagnetic environment (such as the application in the automatic driving system), this method can significantly improve the accuracy and reliability of the circuit characteristic analysis, so that the signal integrity simulation error of the 77GHz vehicle-mounted radar FPC is reduced from the industry typical 20% to below 3%, greatly improving the stability and performance of the system. In this way, even in extremely complex circuit design, the width, spacing and actual position of each wire in three-dimensional space can be accurately measured and recorded, providing a solid data foundation for subsequent electromagnetic compatibility analysis.

[0034] In specific embodiments, the electromagnetic compatibility analysis of the FPC sensing module based on the circuit wiring feature map under the given working signal condition to obtain an electromagnetic interference hotspot distribution map includes: Perform S-parameter measurement on the FPC sensing module under the given working signal condition by a preset vector network analyzer to obtain a frequency domain scattering parameter matrix, and perform time domain gating transformation on the frequency domain scattering parameter matrix to obtain time domain reflection characteristic data; Extract the wire geometric parameters of the circuit wiring feature map to obtain a three-dimensional wire structure parameter set, and perform dipole current distribution modeling based on the three-dimensional wire structure parameter set to obtain an equivalent current source distribution map; Perform near-field coupling calculation on the equivalent current source distribution map by the method of moments to obtain spatial electromagnetic field intensity distribution data; Perform transmission line crosstalk analysis on the FPC sensing module based on the time domain reflection characteristic data to obtain an inter-line coupling coefficient matrix, and perform far-field radiation transformation on the spatial electromagnetic field intensity distribution data to obtain a radiation emission power spectral density map; Decompose the eigenvalues of the inter-line coupling coefficient matrix to obtain main crosstalk mode parameters, and perform ground plane common mode current inversion calculation based on the radiation emission power spectral density map to obtain a common mode current distribution vector, and perform weighted fusion of the main crosstalk mode parameters and the common mode current distribution vector by the superposition principle to obtain an electromagnetic interference hotspot distribution map.

[0035] Specifically, the entire process of analyzing the electromagnetic compatibility of the FPC sensor module under given operating signal conditions to obtain the electromagnetic interference hotspot distribution map is complex and meticulous. It involves the acquisition of the frequency-domain scattering parameter matrix, the extraction of the three-dimensional wire structure parameter set, the establishment of the equivalent current source distribution map, the calculation of the spatial electromagnetic field intensity distribution data, and the generation of the final electromagnetic interference hotspot distribution map. First, by using a pre-set vector network analyzer, we can perform S-parameter measurements on the FPC sensor module under specific operating signal conditions. This step allows us to obtain the frequency-domain scattering parameter matrix, which contains information about the signal transmission characteristics between various points in the circuit. For example, in a 77GHz radar system for an autonomous vehicle, this measurement may need to cover a frequency range from 1GHz to 80GHz to ensure that all potential electromagnetic interferences are captured. Then, by performing time-domain gating transformation on the frequency-domain scattering parameter matrix, we obtain time-domain reflection characteristic data, which provides specific information about impedance mismatches and reflection phenomena on the signal path. Next, we extract the wire geometry parameters from the circuit layout feature map and construct a three-dimensional wire structure parameter set. This is a critical step because it directly affects the accuracy of the subsequent dipole current distribution modeling. For example, when dealing with a gold-plated signal layer with a thickness of 18μm, accurate width and spacing (e.g., 50μm) are crucial for accurately simulating current density. Based on these three-dimensional wire structure parameter sets, we can model the dipole current distribution and generate an equivalent current source distribution map containing information about the wire segment current density, phase distribution, and polarization direction. This process not only helps us understand the flow of current in the circuit but also provides basic data for subsequent near-field coupling calculations. Using the method of moments, we can perform near-field coupling calculations on the equivalent current source distribution map to obtain spatial electromagnetic field intensity distribution data, which is one of the core steps in evaluating the electromagnetic compatibility of the FPC sensor module. In practical operations, if we are considering a 200mm x 150mm size autonomous driving domain controller FPC, we need to analyze each fine wire and its surrounding electromagnetic field in detail. At the same time, based on the time-domain reflection characteristic data, we can perform transmission line crosstalk analysis on the FPC sensor module to obtain an inter-line coupling coefficient matrix, which reveals the degree of interaction between different signal lines. In addition, by performing far-field radiation transformation on the spatial electromagnetic field intensity distribution data, we can generate a radiation emission power spectral density map, which is important for predicting the electromagnetic radiation level of the FPC in actual use environment. By decomposing the eigenvalues of the inter-line coupling coefficient matrix to determine the main crosstalk mode parameters, and based on the radiation emission power spectral density map, we can perform ground plane common mode current inversion calculation to obtain the common mode current distribution vector, which is an important step in realizing electromagnetic interference quantification. For example, when dealing with a complex FPC design containing multiple signal lines and power planes, this method can help identify the areas that are most susceptible to crosstalk effects.Finally, by superimposing the main crosstalk mode parameters and the common-mode current distribution vector, the electromagnetic interference hotspot distribution map can be obtained. This map not only shows the signal crosstalk intensity level and the radiation interference spectrum distribution, but also accurately locates the hotspot area of the ground plane common-mode current, which plays an irreplaceable role in improving FPC design and improving its electromagnetic compatibility. For example, in a specific scenario, suppose we are analyzing an FPC used in an automatic driving vehicle laser radar control system, which contains multiple differential signal line pairs with a spacing of 50 μm. Through the above series of electromagnetic compatibility analysis steps for the FPC, not only can the potential fault points caused by signal crosstalk be effectively located, but also the design can be optimized to reduce the impact of electromagnetic interference, ensuring that the laser radar system can operate stably in a complex electromagnetic environment. This comprehensive analysis method enables engineers to more accurately understand and solve electromagnetic compatibility problems in FPC design, greatly improving the reliability and safety of vehicle-mounted electronic devices.

[0036] In specific embodiments, the transmission line crosstalk analysis of the FPC sensing module based on the time-domain reflection characteristic data obtains an inter-wire coupling coefficient matrix, including: The time-domain reflection characteristic data is differentiated to obtain a reflection waveform slope graph, and the impedance discontinuity of the flexible circuit board of the FPC sensing module is detected based on the reflection waveform slope graph to obtain an impedance change node distribution graph; The physical structure of the impedance change node distribution graph is reconstructed by geometric inversion technology to generate a conductor geometric deformation distribution graph, and the electric field distribution is calculated based on the conductor geometric deformation distribution graph combined with the circuit layout feature map to obtain an inter-wire electric field coupling strength graph; The inter-wire electric field coupling strength graph is integrated by Green's function to obtain a mutual inductance coefficient distribution matrix, and the crosstalk transmission path of the FPC sensing module is analyzed based on the mutual inductance coefficient distribution matrix to obtain a crosstalk propagation characteristic vector; The inter-wire coupling coefficient matrix is obtained by matrix eigenvalue decomposition based on the crosstalk propagation characteristic vector.

[0037] Specifically, in the step, the process of transmission line crosstalk analysis of the FPC sensing module based on time domain reflection characteristic data to obtain the inter-line coupling coefficient matrix is a highly systematic, multi-stage electromagnetic modeling and signal processing procedure. The process first relies on the S parameters obtained by the vector network analyzer in advance and the time domain reflection characteristic data obtained after the time domain gating transformation. These data essentially reflect the signal reflection behavior on the FPC sensing module internal wire path caused by impedance discontinuity. By differentiating these time domain reflection characteristic data, the reflection waveform slope graph reflecting the transient change characteristics of the signal path can be obtained. This slope graph can effectively highlight the local signal mismatch area caused by physical structure mutation. Then, based on this, the impedance discontinuity of the flexible circuit board of the FPC sensing module is detected using the reflection waveform slope graph, thereby generating an impedance change node distribution graph. The distribution graph essentially identifies all possible positions inside the FPC that can cause signal integrity problems in the form of nodes. For example, in the FPC of the 77GHz radar system of an autonomous vehicle, if a 50μm wide differential signal line is locally reduced to 40μm in width due to manufacturing errors, an obvious impedance jump will be formed at this position, thereby becoming an interference hotspot. Subsequently, the physical structure of each node in the impedance change node distribution graph is reconstructed by geometric inversion technology, thereby generating a wire geometry deformation distribution graph. The key of this step is to restore the electrical measurement results to actual physical size changes, so that the subsequent electromagnetic field simulation is closer to the real situation. For example, in the above radar system FPC, if the wire locally bends or deviates due to the deviation of the pressing process, the geometric distortion degree of the region can be accurately identified by the inversion technology, and an accurate three-dimensional structure model is established accordingly. Further, based on the wire geometry deformation distribution graph and the original circuit layout feature map, the electric field distribution is calculated, and finally the electric field coupling strength graph between the wires is obtained. This graph reveals the capacitive coupling strength between different wires due to spatial proximity effect, and is especially suitable for crosstalk prediction in high-density wiring scenarios. For example, in a 200mm×150mm size FPC of an autonomous vehicle domain controller, if the spacing between adjacent signal lines is only 50μm, the electric field coupling strength between them will be significantly enhanced, which is easy to cause bit error or functional abnormality. Next, the obtained inter-wire electric field coupling strength graph is subjected to Green function integral processing to construct a mutual inductance coefficient distribution matrix. The matrix describes the magnetic coupling characteristics between the wires and is the basis for evaluating the near-end and far-end crosstalk levels. For example, in an FPC design containing multiple high-speed differential line pairs, if there is an asymmetric layout between one pair of signal lines and the power layer, the mutual inductance coefficient at a certain frequency band will be significantly increased, thereby exacerbating the common mode interference.Finally, based on the mutual inductance coefficient distribution matrix, the crosstalk transmission path of the FPC sensing module is analyzed, the crosstalk propagation characteristic vector is extracted, and the matrix eigenvalue decomposition is performed on the characteristic vector, so as to obtain the inter-wire coupling coefficient matrix. The matrix not only includes the near-end crosstalk coefficient (NEXT) and the far-end crosstalk coefficient (FEXT) in the traditional sense, but also covers the frequency domain coupling response information, thereby comprehensively representing the electromagnetic coupling relationship between the signal lines in the FPC. For example, in an FPC used in a laser radar control system, it can be identified through the above analysis that there is strong low-frequency common-mode coupling between a pair of differential lines and their adjacent ground planes, which will directly affect the EMC performance of the system. In summary, the above series of operations gradually deduce the inter-wire coupling coefficient matrix from the original time-domain reflection data, realize the complete mapping from the physical structure to the electromagnetic interference characteristics, and provide key input basis for the subsequent construction of the electromagnetic interference hotspot distribution map.

[0038] In specific embodiments, the adaptive electromagnetic shielding design of the FPC sensing module based on the electromagnetic interference hotspot distribution map obtains a shielding structure scheme, including the following steps: Based on the electromagnetic interference hotspot distribution map, interference source positioning analysis is performed on the FPC sensing module to obtain an interference source coordinate set, and a three-dimensional electromagnetic field intensity gradient map is constructed based on the interference source coordinate set; Based on the three-dimensional electromagnetic field intensity gradient map, an initial shielding structure layout is designed, and finite element analysis is performed on the initial shielding structure layout to obtain shielding effectiveness evaluation data; Based on the shielding effectiveness evaluation data, the initial shielding structure layout is multi-objective optimized to obtain an improved shielding structure design, and the improved shielding structure design is verified through electromagnetic wave propagation simulation technology to obtain a shielding performance matrix; The shielding performance matrix is subjected to sensitivity analysis to obtain a key design parameter set, and a shielding structure scheme is determined based on the key design parameter set.

[0039] Specifically, the process of adaptive electromagnetic shielding design for FPC sensor modules based on the electromagnetic interference hotspot distribution map is a complex process that integrates precise analysis, optimization design, and verification evaluation. First, the electromagnetic interference hotspot distribution map is used to perform interference source positioning analysis on the FPC sensor module. This process identifies the locations that produce significant electromagnetic radiation during high-frequency signal transmission and records their coordinates to form a set of interference source coordinates. For example, in an FPC used in an autonomous vehicle radar system, if a segment of differential signal line has local impedance mismatch due to design or manufacturing defects, that location may become a major interference source, and its coordinates can be accurately determined. Subsequently, a three-dimensional electromagnetic field intensity gradient map is constructed based on these interference source coordinates. This step not only visualizes the specific impact range of each interference source but also reveals the trend of electromagnetic field intensity at different frequencies. For example, at 24 GHz, the electromagnetic field intensity in a certain region may increase sharply. Next, an initial shielding structure layout is designed based on the obtained three-dimensional electromagnetic field intensity gradient map. This step aims to minimize the impact of electromagnetic interference by reasonably arranging shielding materials. For example, in the aforementioned radar system FPC, if it is found that the electromagnetic field intensity in a certain region is particularly high, a 0.5mm thick copper foil can be added around that region as a preliminary shielding measure. To ensure the effectiveness of the design, finite element analysis is then performed on the initial shielding structure layout to obtain detailed shielding effectiveness evaluation data. This includes calculating the changes in electromagnetic wave penetration rate, reflectivity, and other parameters before and after shielding. Assuming that the electromagnetic wave penetration rate in that region is 80% before shielding and decreases to 30% after adding the preliminary shielding, it indicates that the design has some effect but still has room for improvement. Based on the above shielding effectiveness evaluation data, the initial shielding structure layout is further optimized to obtain a more perfect improved shielding structure design. The multi-objective optimization mentioned here usually involves the trade-off between multiple performance indicators, such as balancing shielding effectiveness and factors such as cost and weight. For example, while ensuring that the shielding effectiveness is not less than 90%, the amount of shielding material used is minimized to make the entire FPC module meet EMC requirements while maintaining lightweight design. Then, the improved shielding structure design is verified through electromagnetic wave propagation simulation technology to generate a shielding performance matrix. This matrix contains actual performance data of the shielding structure under various working conditions, such as changes in shielding efficiency under temperature ranges from -40°C to +85°C and different humidity conditions, which helps to comprehensively evaluate the reliability of the design scheme. Finally, sensitivity analysis is performed on the shielding performance matrix to extract a set of key design parameters that have a crucial impact on shielding performance. For example, it is found that the shielding layer thickness, material conductivity, and dielectric constant are key factors that determine shielding performance, so the final design scheme can be adjusted based on these parameters.Based on the key set of design parameters, the final shielding structure scheme is determined to ensure that it can effectively suppress electromagnetic interference and meet other constraint conditions in actual application. For example, in the FPC design of an autonomous vehicle, the shielding scheme after a series of optimizations can still maintain a shielding efficiency of more than 95% in extreme environments, while controlling the additional weight to be no more than 10 g, fully embodying the great potential of the method in improving the electromagnetic compatibility and reliability of products. The entire process is closely centered around the electromagnetic interference hotspot distribution map, and through a series of refined design and verification links, the whole process coverage from interference source positioning to efficient shielding scheme development is realized.

[0040] The performance optimization method of the high-order computing power autonomous driving FPC sensing module in the embodiment of the application is described above, and the performance optimization device of the high-order computing power autonomous driving FPC sensing module in the embodiment of the application is described below. Please refer to Figure 2 An embodiment of the performance optimization device of the high-order computing power autonomous driving FPC sensing module in the embodiment of the application includes: The scanning module 21 is configured to scan the flexible circuit board of the FPC sensing module through the multispectral imaging system to obtain multi-dimensional spectral image data. The extraction module 22 is configured to extract circuit features from the multi-dimensional spectral image to obtain a circuit wiring feature map. The analysis module 23 is configured to perform electromagnetic compatibility analysis on the FPC sensing module based on the circuit wiring feature map under a given working signal condition to obtain an electromagnetic interference hotspot distribution map. The design module 24 is configured to perform adaptive electromagnetic shielding design on the FPC sensing module based on the electromagnetic interference hotspot distribution map to obtain a shielding structure scheme. The optimization module 25 is configured to optimize the FPC sensing module based on the shielding structure scheme to obtain a target FPC sensing module.

[0041] In this embodiment, the specific implementation of each unit in the above device embodiment is described above in the method embodiment, and will not be described here.

[0042] Referring to Figure 3 In the embodiment of the application, a computer device is also provided, and the internal structure of the computer device can be as Figure 3The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store corresponding data in the embodiment. The network interface of the computer device is configured to communicate with an external terminal through a network connection. The computer program is executed by the processor to implement the above method.

[0043] Those skilled in the art can understand that Figure 3 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied.

[0044] The embodiment of the present application also provides a computer readable storage medium, which stores a computer program. The computer program is executed by the processor to implement the above method. It can be understood that the computer readable storage medium in the embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0045] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. Any reference to the memory, storage, database or other medium provided by the present application and used in the embodiments can include non-volatile and / or volatile memory. The non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. The volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM, etc.

[0046] It is to be understood that the terminology "including", "comprising", or any other variation thereof, is intended to cover a non-exclusive inclusion such that process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises a... " does not, without more constraints, exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0047] The preferred embodiments of the present application have been described above with the specific language and / or illustrative examples, but it should be understood that the patent protection is not limited to the specific embodiments and / or examples described herein, but it should be understood that all equivalents developed by one skilled in the art that are within the scope of the present application are to be included as further embodiments of the present application.

Claims

1. A method for performance optimization of a high-order computing power autonomous driving FPC sensor module, characterized in that, The method comprises the following steps: scanning the flexible circuit board of the FPC sensing module by a multi-spectral imaging system to obtain multi-dimensional spectral image data; extracting circuit features from the multi-dimensional spectral image to obtain a circuit wiring feature map; under a given working signal condition, performing electromagnetic compatibility analysis on the FPC sensing module based on the circuit wiring feature map to obtain an electromagnetic interference hotspot distribution map; based on the electromagnetic interference hotspot distribution map, performing adaptive electromagnetic shielding design on the FPC sensing module to obtain a shielding structure scheme; optimizing the FPC sensing module based on the shielding structure scheme to obtain a target FPC sensing module.

2. The performance optimization method of the high-order computing power automatic driving FPC sensing module according to claim 1, wherein The scanning of the flexible circuit board of the FPC sensing module by the multi-spectral imaging system to obtain multi-dimensional spectral image data comprises: positioning and calibrating the flexible circuit board of the FPC sensing module by a preset optical alignment device to obtain positioning parameters, and optimizing the scanning path of the multi-spectral imaging system based on the positioning parameters to obtain an adaptive scanning trajectory map; based on the adaptive scanning trajectory map, performing multi-band synchronous scanning on the flexible circuit board of the FPC sensing module to obtain an original multi-band image data set, and performing spectral separation on the original multi-band image data set by a spectral unmixing algorithm to obtain separated visible light, near-infrared and thermal imaging data; performing multi-scale image registration on the separated visible light, near-infrared and thermal imaging data to obtain aligned multi-dimensional spectral images, and performing dynamic range compression on the aligned multi-dimensional spectral images by an adaptive histogram equalization technique to obtain multi-dimensional spectral image data.

3. The performance optimization method of the high-order computing power automatic driving FPC sensing module according to claim 1, wherein The extraction of circuit features from the multi-dimensional spectral image to obtain a circuit wiring feature map comprises: performing multi-scale wavelet decomposition on the multi-dimensional spectral image to obtain image feature subsets in different frequency domains, and performing adaptive threshold segmentation on the multi-dimensional spectral image based on the image feature subsets to obtain a preliminary circuit outline map; performing noise suppression and edge enhancement on the preliminary circuit outline map to obtain an optimized circuit outline map, and performing Hough transform based on the optimized circuit outline map to obtain a circuit segment parameter set; based on the circuit segment parameter set, reconstructing the circuit topology structure of the flexible circuit board to obtain a preliminary circuit wiring map, and geometrically correcting the preliminary circuit wiring map; performing edge detection on the geometrically corrected preliminary circuit wiring map to obtain circuit edge positioning data, and performing circuit width and spacing measurement based on the circuit edge positioning data to obtain a circuit wiring feature map.

4. The performance optimization method of the high-order computing power automatic driving FPC sensing module according to claim 3, characterized in that, The adaptive threshold segmentation of the multi-dimensional spectral image based on the image feature subsets to obtain a preliminary circuit outline map comprises: performing local entropy value calculation on the image feature subsets to obtain a circuit region gray scale distribution feature map, and performing multi-level clustering based on the circuit region gray scale distribution feature map to obtain a circuit region hierarchical data set; performing region labeling on the circuit region hierarchical data set to obtain a circuit outline region label map; Perform two-dimensional curvature analysis on the circuit contour region label map to obtain circuit surface curvature feature data, and perform local optimal threshold calculation based on the circuit surface curvature feature data to obtain a dynamic threshold mapping matrix; Segment the circuit contour region label map based on the dynamic threshold mapping matrix to obtain a circuit region binary result, and perform morphological post-processing on the circuit region binary result to obtain a preliminary circuit contour map.

5. The performance optimization method of the high-order computing power automatic driving FPC sensing module according to claim 1, wherein Under the given working signal condition, based on the circuit wiring feature map, the electromagnetic compatibility of the FPC sensing module is analyzed to obtain an electromagnetic interference hotspot distribution map, including: Perform S-parameter measurement on the FPC sensing module under the given working signal condition through a preset vector network analyzer to obtain a frequency domain scattering parameter matrix, and perform time domain gating transformation on the frequency domain scattering parameter matrix to obtain time domain reflection feature data; Extract the conductor geometric parameters of the circuit wiring feature map to obtain a three-dimensional conductor structure parameter set, and perform dipole current distribution modeling based on the three-dimensional conductor structure parameter set to obtain an equivalent current source distribution map; Perform near-field coupling calculation on the equivalent current source distribution map through the method of moments to obtain spatial electromagnetic field intensity distribution data; Based on the time domain reflection feature data, transmission line crosstalk analysis is performed on the FPC sensing module to obtain an inter-line coupling coefficient matrix, and far-field radiation transformation is performed on the spatial electromagnetic field intensity distribution data to obtain a radiation emission power spectral density map; Decompose the eigenvalues of the inter-line coupling coefficient matrix to obtain main crosstalk mode parameters, and based on the radiation emission power spectral density map, perform common mode current inversion calculation to obtain a common mode current distribution vector, and through the superposition principle, weight and fuse the main crosstalk mode parameters and the common mode current distribution vector to obtain the electromagnetic interference hotspot distribution map.

6. The performance optimization method of the high-order computing power automatic driving FPC sensing module according to claim 5, characterized in that, The transmission line crosstalk analysis of the FPC sensing module based on the time domain reflection feature data to obtain the inter-line coupling coefficient matrix includes: Differential processing is performed on the time domain reflection feature data to obtain a reflection waveform slope map, and based on the reflection waveform slope map, impedance discontinuity detection is performed on the flexible circuit board of the FPC sensing module to obtain an impedance change node distribution map; Through geometric inversion technology, the physical structure of the impedance change node distribution map is reconstructed to generate a conductor geometric deformation distribution map, and based on the conductor geometric deformation distribution map, electric field distribution calculation is performed in combination with the circuit wiring feature map to obtain an inter-conductor electric field coupling strength map; Green function integration is performed on the inter-conductor electric field coupling strength map to obtain a mutual inductance coefficient distribution matrix, and based on the mutual inductance coefficient distribution matrix, crosstalk analysis is performed on the crosstalk transmission path of the transmission line in the FPC sensing module to obtain a crosstalk propagation feature vector; Based on the crosstalk propagation feature vector, matrix eigenvalue decomposition is performed to obtain the inter-line coupling coefficient matrix.

7. The performance optimization method of the high-order computing power automatic driving FPC sensing module according to claim 1, wherein Based on the electromagnetic interference hotspot distribution map, adaptive electromagnetic shielding design is performed on the FPC sensing module to obtain a shielding structure scheme, including the following steps: Based on the electromagnetic interference hotspot distribution map, interference source positioning analysis is performed on the FPC sensing module to obtain an interference source coordinate set, and a three-dimensional electromagnetic field strength gradient map is constructed based on the interference source coordinate set; Based on the three-dimensional electromagnetic field strength gradient map, an initial shielding structure layout is designed, and finite element analysis is performed on the initial shielding structure layout to obtain shielding effectiveness evaluation data; Based on the shielding effectiveness evaluation data, the initial shielding structure layout is optimized in multiple targets to obtain an improved shielding structure design, and the improved shielding structure design is verified through electromagnetic wave propagation simulation technology to obtain a shielding performance matrix; Sensitivity analysis is performed on the shielding performance matrix to obtain a key design parameter set, and a shielding structure scheme is determined based on the key design parameter set.

8. A performance optimization device for a high-order computing power autonomous driving FPC sensor module, characterized in that, Comprise: A scanning module for scanning the flexible circuit board of the FPC sensing module through a multispectral imaging system to obtain multidimensional spectral image data; An extraction module for circuit feature extraction on the multidimensional spectral image to obtain a circuit wiring feature map; An analysis module for electromagnetic compatibility analysis of the FPC sensing module based on the circuit wiring feature map under a given working signal condition to obtain an electromagnetic interference hotspot distribution map; A design module for adaptive electromagnetic shielding design of the FPC sensing module based on the electromagnetic interference hotspot distribution map to obtain a shielding structure scheme; An optimization module for optimizing the FPC sensing module based on the shielding structure scheme to obtain a target FPC sensing module. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8. The processor executes the computer program to realize the steps of the method of any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method of any one of claims 1 to 7.

Images