Chip electromagnetic interference prediction system and method based on adaptive scanning

Abstract

The invention discloses a chip electromagnetic interference prediction system and method based on adaptive scanning, and relates to the technical field of intelligent chips, the system comprises the following components: a two-dimensional automatic scanning platform, a low-frequency magnetic field probe, a signal acquisition module and a data processing and analysis module; the two-dimensional automatic scanning platform comprises an X-axis moving mechanism, a Y-axis moving mechanism, a stepping motor and an Arduino microcontroller, the X-axis moving mechanism and the Y-axis moving mechanism are both provided with a sliding rail system, the stepping motor is in driving connection with the corresponding axis moving mechanism, the Arduino microcontroller is electrically connected with the stepping motor, the Arduino microcontroller outputs a control signal to drive the stepping motor to operate, and the sliding rail system is electrically connected with the Arduino microcontroller. The stepping motor drives the low-frequency magnetic field probe to move in a plane in a chip radiation field; according to the method, electromagnetic interference space distribution reconstruction and prediction are realized based on the radial basis function network (RBFN) and singular value decomposition (SVD), the defect in the aspect of data modeling stability in the prior art is overcome, and the prediction precision is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent chip technology, specifically to a chip electromagnetic interference prediction system and method based on adaptive scanning. Background Technology

[0002] With the rapid development of electronic information technology, chips, as core components of electronic devices, are experiencing increasingly higher integration and operating frequencies. This leads to a gradual increase in the electromagnetic radiation generated during chip operation, making electromagnetic interference (EMI) a more prominent issue. EMI not only affects the performance and stability of the chip itself but can also interfere with surrounding electronic devices, causing malfunctions in the entire electronic system and ultimately impacting the reliability and safety of electronic equipment. Therefore, accurately predicting the near-field magnetic field near the chip has become a crucial prerequisite for chip electromagnetic compatibility assessment and chip design optimization.

[0003] Existing near-field magnetic field prediction technologies for chips primarily employ a fixed-interval uniform scanning method. This method comprehensively scans the chip's radiation field region according to a preset fixed sampling interval. However, this traditional approach has several drawbacks. On one hand, to ensure prediction accuracy, a small sampling interval is required, which significantly increases the number of sampling points, resulting in a longer scanning time and lower efficiency. On the other hand, uniform scanning cannot flexibly adjust the sampling density based on the actual magnetic field distribution. In areas with gentle magnetic fields, there are numerous redundant sampling points, leading to resource waste. In areas with drastic magnetic field changes, the fixed sampling interval can easily result in the omission of key information, leading to uneven measurement data coverage and an inability to accurately reflect the true changes in the magnetic field. Furthermore, existing technologies also have significant shortcomings in terms of test platform control precision, data modeling stability, and EMI assessment visualization. Low test platform control precision affects the accurate positioning of sampling points, thus impacting the accuracy of measurement data. Poor data modeling stability leads to unreliable prediction results, and poor visualization hinders engineers' intuitive understanding of the chip's electromagnetic interference situation. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a chip electromagnetic interference prediction system and method based on adaptive scanning. It achieves efficient and accurate prediction and assessment of chip electromagnetic interference through the collaborative work of a two-dimensional automatic scanning platform, a low-frequency magnetic field probe, a signal acquisition module, and a data processing and analysis module. During the scanning process, an adaptive scanning strategy is adopted to obtain basic data through initial sampling of a uniform regular grid. Then, combined with a sampling adjustment strategy, the positions of subsequent sampling points are dynamically determined, with a focus on supplementing sampling in areas of drastic magnetic field changes, effectively improving scanning efficiency and the accuracy of measurement data. In terms of data processing and analysis, the spatial distribution of electromagnetic interference is reconstructed and predicted based on radial basis function networks and singular value decomposition. By constructing an accurate spatial distribution model of electromagnetic interference, the magnetic field at unmeasured points can be accurately predicted, and the electromagnetic interference distribution trend of the chip under different operating states and environmental conditions can be simulated.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: On the one hand, a chip electromagnetic interference prediction system based on adaptive scanning, the system comprising: a two-dimensional automatic scanning platform, a low-frequency magnetic field probe, a signal acquisition module, and a data processing and analysis module; The two-dimensional automatic scanning platform includes an X-axis moving mechanism, a Y-axis moving mechanism, a stepper motor, and an Arduino microcontroller. Both the X-axis and Y-axis moving mechanisms are equipped with a slide rail system. The stepper motor is driven and connected to the corresponding axis moving mechanism. The Arduino microcontroller is electrically connected to the stepper motor. The Arduino microcontroller outputs control signals to drive the stepper motor to run. The stepper motor drives the low-frequency magnetic field probe to move within the plane where the chip radiation field is located. The low-frequency magnetic field probe is connected to the signal acquisition module. The low-frequency magnetic field probe detects the magnetic field signal in the radiation field region of the chip and acquires the normal component of the magnetic flux density, and transmits the magnetic field signal to the signal acquisition module. The signal acquisition module is connected to the data processing and analysis module. The signal acquisition module receives the magnetic field signal transmitted by the low-frequency magnetic field probe, filters and amplifies the magnetic field signal, and converts the processed magnetic field data into a digital signal and transmits it to the data processing and analysis module. The data processing and analysis module has built-in sampling adjustment logic and electromagnetic interference distribution reconstruction and prediction function unit. It receives magnetic field data in digital signal form transmitted by the signal acquisition module, and calculates the areas with strong magnetic fields that need to be sampled more based on the magnetic field data. It then sends position commands to the Arduino microcontroller of the two-dimensional automatic scanning platform. Based on the acquired magnetic field data, it uses interpolation formulas to interpolate and reconstruct the spatial distribution of electromagnetic interference, predicts the magnetic field of unmeasured points, generates field distribution maps and contour maps, and completes the EMI compliance assessment.

[0006] On the other hand, a chip electromagnetic interference prediction method based on adaptive scanning, the method comprising: Initial sampling and data acquisition: The Arduino microcontroller sends control commands to the stepper motor of the 2D automatic scanning platform. The stepper motor drives the low-frequency magnetic field probe to scan the test plane according to a regular grid of uniformly distributed sampling points. The low-frequency magnetic field probe detects the magnetic field signal at each sampling point and collects the normal component of the magnetic flux density. The signal acquisition module processes the magnetic field signal and transmits it to the data processing and analysis module. The Arduino microcontroller controls the measurement timing and transmits the collected magnetic field data and corresponding spatial coordinates to the host computer via serial port. The data processing and analysis module stores and analyzes the magnetic field data and spatial coordinates. Sampling adjustment and supplementary sampling: Based on the initial sampling point coordinates and the normal component of the magnetic flux density, the data processing and analysis module calculates the rate of change of the magnetic field between adjacent sampling points. Then, based on the rate of change of the magnetic field on each side of the triangular grid, it determines the magnetic field gradient value of each triangular grid region. Based on the magnetic field gradient value, it identifies regions with drastic magnetic field changes and determines the coordinates of key sampling points in these regions. The data processing and analysis module sends the sampling coordinates to the Arduino controller via serial port. The Arduino controller drives the probe to move to the specified position to perform sampling, iteratively obtaining the coordinates of the new sampling point and the corresponding normal component of the magnetic flux density. Data Modeling and Interference Prediction Analysis: The data processing and analysis module integrates the electromagnetic interference data and coordinates of all sampling points, constructs an electromagnetic interference spatial distribution model, and realizes the interpolation and reconstruction of the electromagnetic interference spatial distribution through a radial basis function network. The data processing and analysis module uses the constructed model to predict the electromagnetic interference distribution trend of the chip under different working states and environments. EMI Assessment and Visualization: The data processing and analysis module calculates the chip's maximum radiation intensity, field distribution map, and contour map at each frequency band based on the prediction model, and assesses the chip's EMI compliance based on the calculation results. Furthermore, in the sampling adjustment and supplementary sampling steps, the data processing and analysis module calculates the rate of change of the magnetic field between adjacent sampling points based on the initial sampling point coordinates and the normal component of the magnetic flux density. The calculation formula is as follows: ,in, Representing the The initial sampling point and the first The rate of change of the normal component of magnetic flux density between the initial sampling points Representing the The normal component of the magnetic flux density at each initial sampling point Representing the The normal component of the magnetic flux density at each initial sampling point Representing the Spatial coordinates of the initial sampling points Representing the Spatial coordinates of the initial sampling points Representing the The and the first The Euclidean distance between the initial sampling points is calculated by taking the square root of the sum of the squares of the differences between the coordinates of the two points, reflecting the spatial interval between the sampling points and providing a distance benchmark for calculating the rate of change of the magnetic field.

[0007] Furthermore, in the sampling adjustment and supplementary sampling steps, the magnetic field gradient value of each triangular grid region is determined based on the rate of change of the magnetic field on each side of the triangular grid. The calculation formula is as follows: ,in, Representing the The magnetic field gradient values ​​of each triangular grid region are used to characterize the overall drastic change in the magnetic field within that grid region. , Representing the first The rate of change of the magnetic field between the three vertices of a triangular grid.

[0008] Furthermore, in the sampling adjustment and supplementary sampling steps, the stopping condition for iterative sampling is: the change amplitude of the magnetic field data of the newly added sampling points in two adjacent iterations is less than a preset ratio; after each iteration, the data processing and analysis module recalculates the magnetic field change distribution of the scanning area until the stopping condition is met.

[0009] Furthermore, in the data modeling and interference prediction analysis step, the data processing and analysis module integrates the electromagnetic interference data and coordinates of all sampling points to construct an electromagnetic interference spatial distribution model, the formula of which is: ,in, The normal component of the magnetic flux density at the preset observation point on the measurement surface reflects the intensity of electromagnetic radiation from the chip at that point. The spatial coordinates of the preset observation points on the measurement surface. Represents the first measurement surface Coordinates of the center sampling points of the radial basis functions Representing the The weight values ​​corresponding to each radial basis function. Represents the number of radial basis functions. The larger the value, the more detailed the model's representation of the magnetic field distribution, and the higher the prediction accuracy. Representing the A radial basis function is used to describe the spatial correlation between the observation point and the central sampling point.

[0010] Furthermore, in the data modeling and interference prediction analysis steps, the spatial distribution of electromagnetic interference is interpolated and reconstructed using a radial basis function network. The radial basis function is expressed as: ,in The distance between the observation point and the central sampling point is the Euclidean distance. For shape parameters, The order parameter is denoted by ; the center of the radial basis function is the coordinates of all sampling points, and the shape parameter is determined by trial calculation based on the spatial distribution characteristics of the sampling points.

[0011] Furthermore, the aforementioned Representing the The weights corresponding to the radial basis functions are solved using singular value decomposition (SVD). Specifically, the RBFN output equation is expressed in matrix form. ,in For magnetic field data vectors, The radial basis function matrix, For the weight vector, Perform SVD decomposition, then ,in , It is an orthogonal matrix. It is a diagonal matrix, and then the formula is used. Solve ,in for The pseudo-inverse matrix, , for , The transpose of .

[0012] Furthermore, in the data modeling and interference prediction analysis step, the electromagnetic interference distribution trend of the chip under different operating states and environments is predicted, and the prediction formula is as follows: ,in, Represents unmeasured points Predicted value of the normal component of magnetic flux density at a given location. This represents the spatial coordinates of unmeasured points, i.e., locations where the magnetic field needs to be predicted but no actual sampling has been performed. Representing the The weights of the radial basis functions. Representing the Coordinates of the center sampling points of the radial basis functions Representing the Radial basis functions at unmeasured points The function value at that location.

[0013] Compared with existing technologies, this chip electromagnetic interference prediction system and method based on adaptive scanning has the following advantages: I. This invention reconstructs and predicts the spatial distribution of electromagnetic interference based on radial basis function networks (RBFN) and singular value decomposition (SVD), which solves the shortcomings of existing technologies in terms of data modeling stability and improves prediction accuracy. Furthermore, by generating field distribution maps and contour maps, the prediction results are displayed intuitively, enhancing the visualization effect of EMI assessment and providing a more reliable basis for chip electromagnetic compatibility assessment and optimization.

[0014] Second, the present invention adopts an adaptive scanning method. After initial sampling through a uniform regular grid, the sampling point position is dynamically determined by combining a sampling adjustment strategy. This avoids redundant sampling in areas with gentle magnetic fields, while increasing sampling points in areas with drastic magnetic field changes. This effectively reduces the number of sampling points, improves scanning efficiency, and results in more uniform measurement data coverage and significantly improved data accuracy.

[0015] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0017] Figure 1 This is a block diagram of a chip electromagnetic interference prediction system based on adaptive scanning. Figure 2 This is a flowchart of a chip electromagnetic interference prediction method based on adaptive scanning. Detailed Implementation

[0018] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0019] Example 1 This embodiment provides a chip electromagnetic interference prediction system based on adaptive scanning, such as... Figure 1 As shown, it includes a two-dimensional automatic scanning platform, a low-frequency magnetic field probe, a signal acquisition module, and a data processing and analysis module, enabling efficient scanning of the chip's radiation field area and prediction of electromagnetic interference.

[0020] The core of the 2D automatic scanning platform consists of an X-axis moving mechanism, a Y-axis moving mechanism, two two-phase hybrid stepper motors, and an Arduino Mega2560 microcontroller. Both the X-axis and Y-axis moving mechanisms use a ball screw slide rail system, with the slide rail length adapted to the maximum size of the chip under test (covering the common chip package size range). The X-axis slide rail is horizontally fixed on the base, and the Y-axis slide rail is vertically mounted on the X-axis moving slider, forming a cross structure. The two stepper motors are connected to one end of the X-axis and Y-axis ball screws respectively via couplings. The motor drivers are electrically connected to the Arduino microcontroller's digital I / O ports (pins D2-D5 for the X-axis motor and pins D6-D9 for the Y-axis motor) via shielded cables to achieve control signal transmission. The low-frequency magnetic field probe is fixed to the lower support of the Y-axis moving slider using a special fixture, with the probe's detection end facing downwards, maintaining a preset vertical distance from the surface of the chip under test (ensuring that the probe does not contact the chip and is within the near-field detection range).

[0021] The low-frequency magnetic field probe uses a Hall effect miniature probe. The probe body is packaged in a size that is compatible with the fixture for installation. The diameter of the probe end is miniaturized (to reduce disturbance to the chip's radiation field). The probe signal output line is a twisted pair cable with a tinned copper mesh shielding layer. One end of the shielding layer is grounded (the grounding end is connected to the system grounding terminal to avoid external electromagnetic interference coupling), and the other end is detachably connected to the signal input terminal (BNC interface) of the signal acquisition module through a connector to achieve stable transmission of the magnetic field signal.

[0022] The signal acquisition module hardware includes a signal conditioning board, a 16-bit resolution analog-to-digital converter (ADC), an STM32F103 microcontroller, and a USB data interface. The signal conditioning board integrates a low-pass filter (the cutoff frequency can be adjusted via an onboard DIP switch), an instrumentation amplifier (gain configured via the system), and a voltage follower. The input of the low-pass filter is connected to the signal output of the magnetic field probe, and the output of the voltage follower is connected to the input of the ADC via a ribbon cable. The ADC communicates with the microcontroller via an SPI interface, and the STM32 microcontroller is connected to the data processing and analysis module via a USB interface. At the same time, the STM32 microcontroller is connected to the D10 pin of the Arduino microcontroller via a shielded cable to receive the sampling trigger signal sent by the Arduino and synchronize the measurement timing.

[0023] The data processing and analysis module uses an industrial computer equipped with a quad-core processor, 16GB of memory, and a 1TB solid-state drive. It is equipped with a dedicated control and analysis system (including data storage, sampling adjustment, modeling prediction, and visualization evaluation modules). The industrial computer is connected to the Arduino microcontroller's hardware serial port via a USB-to-serial module to enable the issuance of sampling coordinate commands and data feedback. The system establishes communication with the signal acquisition module through a driver program to receive magnetic field data in real time. The built-in sampling adjustment logic program, radial basis function network (RBFN) modeling program, and singular value decomposition (SVD) calculation program are all integrated in the system background and can be triggered and executed through the operation interface.

[0024] The industrial computer and dedicated system are started. The system automatically detects the connection status of the two-dimensional automatic scanning platform and signal acquisition module. If the connection is normal, the system prompts the input of the chip parameters to be tested (such as chip size and operating frequency band) and automatically recommends the initial sampling grid spacing based on the parameters. At the same time, the Arduino microcontroller receives the initialization command issued by the system, drives the X-axis and Y-axis stepper motors to move the probe to the starting position of the scanning area, and calibrates the initial position through the position feedback signal (limit switch signal of the stepper motor driver) to ensure positioning accuracy.

[0025] The system sends an initial scan command to the Arduino microcontroller, including sampling grid parameters (spacing, scanning area range) and scanning path (grid path). After parsing the command, the Arduino microcontroller sends direction signals and pulse signals to the X-axis and Y-axis stepper motor drivers according to a preset timing sequence: for example, first drive the X-axis motor to move the probe along the X-axis by one grid spacing, and stop the X-axis motor after reaching the X coordinate of the first sampling point; then drive the Y-axis motor to move the probe along the Y-axis by one grid spacing, and reach the Y coordinate of the first sampling point to complete the sampling point positioning. After positioning is completed, the Arduino microcontroller sends a sampling trigger signal to the signal acquisition module through the D10 pin.

[0026] After receiving the trigger signal, the signal acquisition module uses an STM32 microcontroller to control the ADC to start sampling. First, the analog magnetic field signal transmitted by the probe is processed by the signal conditioning board (a low-pass filter removes high-frequency noise, and an instrumentation amplifier amplifies the μV-level signal to the ADC's range). Then, the ADC converts the processed analog signal into a digital signal. The STM32 microcontroller reads the digital signal output by the ADC, calculates the average value of a preset number of sampling points (to reduce the influence of random noise), and obtains the normal component of the magnetic flux density at that sampling point. Subsequently, the STM32 microcontroller packages the magnetic field data with the sampling point coordinates synchronously transmitted by the Arduino microcontroller and sends it to the industrial computer system via a USB interface. The system stores the data in a database and displays the sampling point location and corresponding magnetic field data in real time on the operation interface.

[0027] After initial sampling, the system automatically calls the sampling adjustment logic program: based on the initial sampling data, it constructs a triangular mesh and calculates the magnetic field gradient value of each mesh to determine areas with drastic magnetic field changes; based on the determination results, it generates a list of coordinates for newly added sampling points and sends it to the Arduino microcontroller via serial port. The Arduino microcontroller drives a stepper motor to move the probe to each newly added sampling point, repeating the above data acquisition process to collect magnetic field data from the newly added sampling points; after the newly added sampling data is transmitted back to the system in real time, the system updates the magnetic field distribution scatter plot to visually display the supplementary sampling situation in areas with drastic magnetic field changes. If multiple rounds of iterative supplementary sampling are required, the system automatically repeats the gradient calculation-region determination-new sampling process until the data converges.

[0028] After sampling is completed, the system prompts the operator to start the modeling and prediction function. After the operator confirms, the system reads all sampled data (initial sampling and supplementary sampling) from the database and constructs a modeling dataset. It automatically divides the training set and validation set, calls the RBFN modeling program to construct an electromagnetic interference spatial distribution model, and solves the model weights through the SVD program. After the model is constructed, the system uses the validation set data to calculate the prediction error. If the error is less than the preset threshold, the model is valid. If the error exceeds the threshold, the system automatically adjusts the RBFN shape parameters and remodels until the error meets the standard.

[0029] Once the model is valid, the system can receive prediction commands input by the operator (such as predicting the electromagnetic interference distribution of the chip under different loads), calculate the magnetic field data of unmeasured points based on the model, and call the visualization module to generate field distribution maps and contour maps. Simultaneously, it calculates the maximum radiation intensity of each frequency band against preset standards and compares it with standard limits, generating an evaluation report and completing the entire system's workflow.

[0030] Example 2 This embodiment tests the electromagnetic interference of a consumer MCU chip. Before starting the prediction method, hardware deployment and system configuration must be completed: The consumer MCU chip to be tested is fixed on a dedicated test fixture, ensuring precise connection between the chip pins and the circuit interface of the test board. The test board is connected to an external power supply and signal source to provide the stable voltage and control signals required for chip operation. Then, the test fixture is placed in the center of the base of the two-dimensional automatic scanning platform. The position of the fixture is adjusted using the platform's built-in positioning scale to align the chip's geometric center with the preset center of the scanning area. Simultaneously, a micrometer is used to calibrate the vertical distance between the low-frequency magnetic field probe and the chip surface, ensuring the distance is within the near-field detection range. Furthermore, the probe does not contact the chip surface to avoid affecting the normal operation of the chip or damaging the probe. The dedicated control and analysis system on the industrial computer is activated to automatically perform equipment connection testing. By sending test commands to the two-dimensional automatic scanning platform and signal acquisition module and receiving response signals, the normal connection of each module is confirmed. After the connection is correct, the operator inputs the key parameters of the chip to be tested (chip size is 10mm×10mm, operating frequency band is 100kHz-1MHz). Based on the parameters, the system automatically recommends an initial sampling grid spacing of 1mm and sets the scanning area to 12mm×12mm to cover the chip and the surrounding radiation area. The operator can adjust the parameters according to actual needs.

[0031] Initial sampling phase, such as Figure 2 As shown, an initial scanning command containing sampling grid parameters (spacing 1mm, scanning area X-axis 0-12mm, Y-axis 0-12mm) and grid scanning path is sent to the Arduino microcontroller of the 2D automatic scanning platform. After parsing the command, the Arduino microcontroller sends direction signals and pulse signals to the X-axis and Y-axis stepper motor drivers according to a preset timing sequence. First, the X-axis motor is driven to move the probe along the X-axis to the X coordinate of the first sampling point, and then the Y-axis motor is driven to move the probe along the Y-axis to the corresponding Y coordinate, completing the sampling point positioning. After positioning, the Arduino microcontroller sends a sampling trigger signal to the signal acquisition module through a dedicated pin. After receiving the trigger signal, the signal acquisition module first filters (removes high-frequency noise) and amplifies (amplifies the μV level signal to the ADC adaptation range) the analog magnetic field signal transmitted by the probe through the signal conditioning board. Then, the ADC converts the processed analog signal into a digital signal. The STM32 microcontroller reads the digital signal output by the ADC, calculates the average value of a preset number of sampling points to reduce the influence of random noise, and obtains the normal component of the magnetic flux density at that sampling point. Subsequently, the signal acquisition module converts the magnetic field data... Sampling point coordinates transmitted synchronously with the Arduino microcontroller (The coordinates are calculated based on the correspondence between the stepper motor step distance and the number of pulses.) The data is packaged and sent to the industrial computer system via USB interface. The data is stored in the database and the sampling point location and corresponding magnetic field data are displayed in real time on the operation interface. Following the above process, the two-dimensional automatic scanning platform drives the probe to complete the measurement of 169 initial sampling points in 13×13 in sequence. Finally, the initial sampling data table and magnetic field distribution scatter plot are generated, providing basic data for subsequent formula calculations.

[0032] After the initial sampling is completed, supplementary sampling is automatically started and the logic program is adjusted: first, based on the coordinates of 169 initial sampling points... The Delaunay triangulation algorithm was used to construct 240 triangular meshes covering the entire scan area; then, for each triangular mesh, three vertices were selected (corresponding to three initial sampling points, with indices of: ...). , , ) magnetic field data , Through formula Calculate the rate of change of the magnetic field between adjacent vertices, for example, a certain grid vertex. , ( ), ), calculated , , Then through the formula The maximum value of the rate of change of the three edges of each grid is taken as the magnetic field gradient value of that grid. (the above grid) According to the accuracy requirements of the chip in IEC 61967-3 standard, a magnetic field gradient threshold is set. ,Will The 32 grids were identified as regions of drastic magnetic field changes. Within these regions, new sampling points were set at three times the initial grid density (with an additional sampling interval of 0.33 mm). The specific coordinates of the 128 new sampling points were determined using a spatial interpolation algorithm. The coordinate list is then sent to the Arduino microcontroller via serial port. The Arduino microcontroller drives the probe to move sequentially according to the coordinates of the newly added sampling points, and the signal acquisition module synchronously acquires the magnetic field data of each newly added sampling point. It also transmits data back in real time, updating the scatter plot of the magnetic field distribution and marking the location of newly added sampling points during the acquisition process. After the new sampling is completed, the magnetic field gradient distribution of all 297 sampling points (169 initial points + 128 new points) is recalculated, and the change range of adjacent sampling data is judged (the percentage of the magnetic field difference between the new sampling point and the surrounding initial sampling points is calculated). At this time, the change range is 3%, which is less than the preset 5% convergence threshold, and the iterative supplementary sampling is automatically stopped.

[0033] After the sampling process is completed, complete data (coordinates) of 297 sampling points are read from the database. With corresponding magnetic field data The dataset is automatically divided into a training set of 208 data points and a validation set of 89 data points in a 7:3 ratio. The operator clicks "Build Model" to begin radial basis function network modeling: using the coordinates of all sampling points in the training set... As the center of RBFN The shape parameters of the RBFN were determined through trial calculations. To minimize the model's fitting error to the training set data, a radial basis function is used, the expression of which is: Then construct the output relationship of RBFN. The number of sampling points in the training set. (as weight values), and represent the relationship. In matrix form To train the magnetic field data vector, The radial basis function matrix, (as a weight vector), then the Singular Value Decomposition (SVD) procedure is called to... Decompose to obtain ( , It is an orthogonal matrix. (For a diagonal matrix), through the formula for The pseudo-inverse matrix, , for , Solving for the weight vector using the transpose matrix After completing model training, the system uses validation set data to verify accuracy, and then calculates the coordinates of the validation set sampling points. Input the model, through formula Calculate the predicted magnetic field value and compare it with the actual measured values ​​on the validation set. By comparison, the average error is obtained. If the error is less than the preset 5% error threshold, the model is deemed valid. The operator then selects either light load or full load operating mode for the chip, clicks "Start Prediction," and uses the formula... Calculate all unmeasured points within the scanned area, divided at 0.1mm intervals. Magnetic field data, such as under full load conditions, can be used to accurately calculate the magnetic field data of the chip's central region. Normal component distribution of magnetic flux density at surrounding unmeasured points And generate the corresponding prediction data table.

[0034] After the prediction is completed, the system automatically calls the EMI evaluation module to read the prediction data of all unmeasured points. The chip was divided into segments according to its operating frequency band (100kHz-1MHz), and the maximum radiation intensity within each segment was calculated. The 100kHz band 500kHz band None of them exceeded the radiation limit of 6nT for this type of chip in the frequency band below 1MHz in the relevant standards. Then, the visualization module was called to analyze the radiation levels based on the results. The system generates a field distribution map under full load (with red gradients representing high magnetic field areas and blue gradients representing low magnetic field areas, clearly showing the central area of ​​the chip as a high magnetic field concentration area), contour maps, and a three-dimensional magnetic field distribution model. Operators can rotate the model through interface controls to view the spatial distribution of the chip's electromagnetic radiation from different angles. Finally, the system automatically integrates chip parameters, sampling settings, model accuracy, comparison results of maximum radiation intensity in each frequency band, and visualization graphics to generate an EMI assessment report in PDF format. Operators can directly export the report, thus completing the entire workflow of the adaptive scanning-based chip electromagnetic interference prediction method.

[0035] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A chip electromagnetic interference prediction system based on adaptive scanning, characterized in that, The system includes: a two-dimensional automatic scanning platform, a low-frequency magnetic field probe, a signal acquisition module, and a data processing and analysis module; The two-dimensional automatic scanning platform includes an X-axis moving mechanism, a Y-axis moving mechanism, a stepper motor, and an Arduino microcontroller. Both the X-axis and Y-axis moving mechanisms are equipped with a slide rail system. The stepper motor is driven and connected to the corresponding axis moving mechanism. The Arduino microcontroller is electrically connected to the stepper motor. The Arduino microcontroller outputs control signals to drive the stepper motor to run. The stepper motor drives the low-frequency magnetic field probe to move within the plane where the chip radiation field is located. The low-frequency magnetic field probe is connected to the signal acquisition module. The low-frequency magnetic field probe detects the magnetic field signal in the radiation field region of the chip and acquires the normal component of the magnetic flux density, and transmits the magnetic field signal to the signal acquisition module. The signal acquisition module is connected to the data processing and analysis module. The signal acquisition module receives the magnetic field signal transmitted by the low-frequency magnetic field probe, filters and amplifies the magnetic field signal, converts the processed magnetic field data into a digital signal, and transmits it to the data processing and analysis module. The data processing and analysis module has built-in sampling adjustment logic and electromagnetic interference distribution reconstruction and prediction function unit. It receives magnetic field data in digital signal form transmitted by the signal acquisition module, and calculates the areas with strong magnetic fields that need to be sampled more based on the magnetic field data. It then sends position commands to the Arduino microcontroller of the two-dimensional automatic scanning platform. Based on the acquired magnetic field data, it uses interpolation formulas to interpolate and reconstruct the spatial distribution of electromagnetic interference, predicts the magnetic field of unmeasured points, generates field distribution maps and contour maps, and completes the EMI compliance assessment.

2. A chip electromagnetic interference prediction method based on adaptive scanning, applicable to the chip electromagnetic interference prediction system based on adaptive scanning as described in claim 1, characterized in that, The method includes: Initial sampling and data acquisition: The Arduino microcontroller sends control commands to the stepper motor of the 2D automatic scanning platform. The stepper motor drives the low-frequency magnetic field probe to scan the test plane according to a regular grid of uniformly distributed sampling points. The low-frequency magnetic field probe detects the magnetic field signal at each sampling point and collects the normal component of the magnetic flux density. The signal acquisition module processes the magnetic field signal and transmits it to the data processing and analysis module. The Arduino microcontroller controls the measurement timing and transmits the collected magnetic field data and corresponding spatial coordinates to the host computer via serial port. The data processing and analysis module stores and analyzes the magnetic field data and spatial coordinates. Sampling adjustment and supplementary sampling: Based on the initial sampling point coordinates and the normal component of the magnetic flux density, the data processing and analysis module calculates the rate of change of the magnetic field between adjacent sampling points. Then, based on the rate of change of the magnetic field on each side of the triangular grid, it determines the magnetic field gradient value of each triangular grid region. Based on the magnetic field gradient value, it identifies regions with drastic magnetic field changes and determines the coordinates of key sampling points in these regions. The data processing and analysis module sends the sampling coordinates to the Arduino controller via serial port. The Arduino controller drives the probe to move to the specified position to perform sampling, iteratively obtaining the coordinates of the new sampling point and the corresponding normal component of the magnetic flux density. Data Modeling and Interference Prediction Analysis: The data processing and analysis module integrates the electromagnetic interference data and coordinates of all sampling points, constructs an electromagnetic interference spatial distribution model, and realizes the interpolation and reconstruction of the electromagnetic interference spatial distribution through a radial basis function network. The data processing and analysis module uses the constructed model to predict the electromagnetic interference distribution trend of the chip under different working states and environments. EMI Assessment and Visualization: The data processing and analysis module calculates the chip's maximum radiation intensity, field distribution map, and contour map in each frequency band based on the prediction model, and assesses the chip's EMI compliance based on the calculation results.

3. The chip electromagnetic interference prediction method based on adaptive scanning according to claim 2, characterized in that, In the sampling adjustment and supplementary sampling steps, the data processing and analysis module calculates the rate of change of the magnetic field between adjacent sampling points based on the initial sampling point coordinates and the normal component of the magnetic flux density. The calculation formula is as follows: ,in, Representing the The initial sampling point and the first The rate of change of the normal component of magnetic flux density between the initial sampling points Representing the The normal component of the magnetic flux density at each initial sampling point Representing the The normal component of the magnetic flux density at each initial sampling point Representing the The spatial coordinates of the initial sampling points Representing the The spatial coordinates of the initial sampling points Representing the The and the first The Euclidean distance between the initial sampling points is calculated by taking the square root of the sum of the squares of the differences between the coordinates of the two points, reflecting the spatial interval between the sampling points and providing a distance benchmark for calculating the rate of change of the magnetic field.

4. The chip electromagnetic interference prediction method based on adaptive scanning according to claim 2, characterized in that, In the sampling adjustment and supplementary sampling steps, the magnetic field gradient value of each triangular grid region is determined based on the rate of change of the magnetic field on each side of the triangular grid. The calculation formula is as follows: ,in, Representing the The magnetic field gradient values ​​of each triangular grid region are used to characterize the overall drastic change in the magnetic field within that grid region. , Representing the first The rate of change of the magnetic field between the three vertices of a triangular grid.

5. The chip electromagnetic interference prediction method based on adaptive scanning according to claim 2, characterized in that, In the sampling adjustment and supplementary sampling steps, the stopping condition for iterative sampling is: the change amplitude of the magnetic field data of the newly added sampling points in two adjacent iterations is less than a preset ratio; after each iteration, the data processing and analysis module recalculates the magnetic field change distribution of the scanned area until the stopping condition is met.

6. The chip electromagnetic interference prediction method based on adaptive scanning according to claim 2, characterized in that, In the data modeling and interference prediction analysis step, the data processing and analysis module integrates the electromagnetic interference data and coordinates of all sampling points to construct an electromagnetic interference spatial distribution model, the formula of which is: ,in, The normal component of the magnetic flux density at the preset observation point on the measurement surface reflects the intensity of electromagnetic radiation from the chip at that point. The spatial coordinates of the preset observation points on the measurement surface. Represents the first measurement surface Coordinates of the center sampling points of the radial basis functions Representing the The weight values ​​corresponding to each radial basis function. Represents the number of radial basis functions. The larger the value, the more detailed the model's representation of the magnetic field distribution, and the higher the prediction accuracy. Representing the A radial basis function is used to describe the spatial correlation between the observation point and the central sampling point.

7. The chip electromagnetic interference prediction method based on adaptive scanning according to claim 6, characterized in that, In the data modeling and interference prediction analysis steps, the spatial distribution of electromagnetic interference is interpolated and reconstructed using a radial basis function network. The radial basis function is expressed as: ,in The distance between the observation point and the central sampling point is the Euclidean distance. For shape parameters, The order parameter is denoted by ; the center of the radial basis function is the coordinates of all sampling points, and the shape parameter is determined by trial calculation based on the spatial distribution characteristics of the sampling points.

8. The chip electromagnetic interference prediction method based on adaptive scanning according to claim 6, characterized in that, The Representing the The weights corresponding to the radial basis functions are solved using singular value decomposition (SVD). Specifically, the RBFN output equation is expressed in matrix form. ,in For magnetic field data vectors, The radial basis function matrix, For the weight vector, Perform SVD decomposition, then ,in , It is an orthogonal matrix. It is a diagonal matrix, and then the formula is used. Solve ,in for The pseudo-inverse matrix, , for , The transpose of .

9. The chip electromagnetic interference prediction method based on adaptive scanning according to claim 8, characterized in that, In the data modeling and interference prediction analysis step, the electromagnetic interference distribution trend of the chip under different operating states and environments is predicted, and the prediction formula is as follows: ,in, Represents unmeasured points Predicted value of the normal component of magnetic flux density at a given location. This represents the spatial coordinates of unmeasured points, i.e., locations where the magnetic field needs to be predicted but no actual sampling has been performed. Representing the The weights of the radial basis functions. Representing the Coordinates of the center sampling points of the radial basis functions Representing the Radial basis functions at unmeasured points The function value at that location.

Images