A test method for stability of electronic products

Abstract

The present application provides a method for testing the stability of an electronic product, comprising: obtaining propagation characteristic data of a transient current in a grounding network, establishing a multi-point grounding network model in combination with the potential distribution data of each node, inputting node potential and transient current data, and predicting the fluctuation trend of ground potential through the multi-point grounding network model; extracting the coupling path of an interference signal in the grounding network when common-mode interference exists according to the predicted ground potential fluctuation trend, determining the main node and propagation direction of the interference signal, and obtaining a sensitive circuit; calculating the node potential distribution data after the node layout of the grounding network, judging whether the potential difference between key nodes is within a preset range, and if it exceeds the preset range, refining the network division, adding equipotential connection, or locally reducing the grounding resistance; determining a matching design scheme for the grounding network and the sensitive circuit according to the anti-interference capability, building a test platform, and testing the stability data of the grounding system under different electromagnetic environments.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method for testing the stability of electronic products. Background Art

[0002] When electronic products are subjected to the instantaneous power impact of the switching power supply, the equipotential balance of the multi-point grounding network is destroyed, and the impedance mismatch of each branch loop causes the loop current to be unable to be quickly established, resulting in a sharp fluctuation in the ground potential. In practical applications, the switching action of the power device will cause a large transient current, causing the ground potential of the grounding network to fluctuate, thereby affecting the working state of the sensitive circuit. Especially in the case of multi-point grounding, the potential difference between different grounding points will further aggravate this impact. Although the sensitive circuit has the ability to suppress common-mode, under the impact of high-frequency transient interference, the common-mode rejection ratio will drop sharply, which greatly reduces the circuit's anti-interference ability. At the same time, due to the impedance imbalance of the multi-point grounding network, the amplitude and phase of the loop current of each branch are quite different, forming a large voltage difference on the pins of the sensitive device, aggravating the risk of device failure and seriously affecting the stability of the system. Therefore, how to improve the suppression ability of the grounding network to the switching transient current, reduce the amplitude of the ground potential fluctuation, and enhance the ability of the sensitive circuit to resist common-mode interference under the premise of ensuring good equipotential characteristics is a key technical problem that needs to be solved to ensure the stable and reliable operation of the system. Summary of the invention

[0003] The present invention provides a method for testing the stability of an electronic product, which mainly includes:

[0004] Acquire the propagation characteristic data of transient current in the grounding network, combine the potential distribution data of each node, establish a multi-point grounding network model, input the node potential and transient current data, and predict the fluctuation trend of the ground potential through the multi-point grounding network model; according to the predicted ground potential fluctuation trend, extract the coupling path of the interference signal in the grounding network when there is common-mode interference, determine the main node and propagation direction of the interference signal, and obtain the sensitive circuit; obtain the quantitative index of the common-mode suppression capability of the sensitive circuit, the quantitative index includes common-mode rejection ratio and common-mode voltage gain, combine the coupling path of the interference signal in the grounding network, analyze the influence of the interference signal on the circuit performance, the influence includes signal distortion and noise increase; preset the common-mode rejection ratio threshold and common-mode voltage gain threshold of the sensitive circuit, if the actual common-mode rejection ratio is lower than the common-mode rejection ratio threshold, or the actual common-mode voltage gain is higher than the common-mode If the voltage gain threshold is exceeded, it is considered that the anti-interference ability does not meet the requirements, and the node layout of the grounding network is adjusted, including increasing the number of connections, optimizing the connection topology, optimizing the impedance matching, or adjusting the grounding resistance of key nodes; the node potential distribution data after the node layout of the grounding network is calculated to determine whether the potential difference between key nodes is within the preset range. If it exceeds the preset range, the network division is refined, equipotential connections are added, or the grounding resistance is partially reduced; through the optimized grounding network design, the input and output signals of the sensitive circuit are extracted, and it is analyzed whether the anti-interference ability of the sensitive circuit is improved before and after optimization. If the common-mode rejection ratio is improved and the common-mode voltage gain is reduced to the target value, it is considered that the target anti-interference ability has been achieved; according to the anti-interference ability, the matching design scheme of the grounding network and the sensitive circuit is determined, and a test platform is built to test the stability data of the grounding system under different electromagnetic environments.

[0005] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0006] The present invention discloses a method for testing the stability of electronic products. The method obtains the propagation characteristics of transient currents in a grounding network and the node potential distribution data, and establishes a multi-point grounding network model to predict the ground potential fluctuation trend. According to the prediction results, the present invention determines the coupling path and sensitive circuit of the interference signal in the grounding network. By analyzing the common mode suppression capability index of the sensitive circuit, the present invention evaluates the impact of the interference signal on the circuit performance. If the anti-interference ability does not meet the requirements, the present invention improves it by adjusting the grounding network node layout, optimizing the connection topology, etc. Finally, the present invention verifies whether the optimized grounding network design achieves the target anti-interference ability, and determines a matching design scheme. This method can effectively improve the stability and anti-interference ability of the circuit system in a complex electromagnetic environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 The present invention is a flow chart of a method for testing the stability of an electronic product.

[0008] Figure 2 It is a schematic diagram of a testing method for electronic product stability of the present invention.

[0009] Figure 3 It is another schematic diagram of a method for testing stability of an electronic product according to the present invention. DETAILED DESCRIPTION

[0010] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0011] like Figure 1-3 In this embodiment, a method for testing the stability of an electronic product may specifically include:

[0012] Step S101, obtain the propagation characteristic data of transient current in the grounding network, combine the potential distribution data of each node, establish a multi-point grounding network model, input the node potential and transient current data, and predict the fluctuation trend of the ground potential through the multi-point grounding network model.

[0013] Transient current sampling data is acquired from a formation current detection device, and the sampling data is processed by an adaptive sampling algorithm to obtain a node current transmission characteristic data set; the node potential value of the grounding network is acquired according to the node current transmission characteristic data set, and the feature of the node potential value is extracted by a random forest classifier to obtain a node feature vector; a grounding network topology unit is generated for the node feature vector, and the grounding network topology unit is connected and configured by a minimum spanning tree algorithm to obtain a network connection relationship matrix; a multi-point grounding network transmission equation group is constructed according to the node feature vector and the network connection relationship matrix, and the multi-point grounding network transmission equation group is solved by an iterative operation to obtain the potential difference distribution data between nodes.

[0014] Specifically, the transient current propagation data is obtained from the formation current detection device and single-point discrete sampling is performed. The adaptive sampling algorithm is used to encrypt the sampling at the point where the current waveform change rate exceeds the set threshold, and the node current transmission characteristic data set is formed according to the sampled data. The potential values ​​of each node of the grounding network are obtained from the distributed potential detection device, and the node potential values ​​are associated with the corresponding node current transmission characteristic data. The node feature vector is constructed by the random forest classifier. The grounding network topology unit is generated according to the node feature vector, and the topology unit is connected and configured according to the underground conductive device layout diagram. The network connection relationship matrix is ​​constructed by the minimum spanning tree algorithm. The soil resistivity distribution data is obtained from the formation detection device, and the hierarchical modeling method is used to generate a three-dimensional grid impedance matrix. The grid impedance value is updated and compensated according to the network connection relationship matrix. The multi-point grounding network transmission equation group is constructed based on the node feature vector and the grid impedance matrix, and the potential difference distribution data between nodes is solved by iterative operation. The impedance characteristic curve is generated according to the potential difference distribution data between nodes and the current transmission characteristic data. The hyperbolic tangent function is used for curve fitting calculation to predict the ground potential fluctuation trend. The formation current detection device detects through the current probe buried in the formation and samples the periodic transient current. When the current fluctuates greatly, the sampling frequency is dynamically adjusted to 100 points per microsecond, and when the waveform is stable, it is reduced to 20 points per microsecond. This adaptive sampling method not only ensures data accuracy but also saves storage space. In a certain grounding network actual application scenario, the probe detected that the transient current peak at point A reached 300 amperes, and the decay period was 50 microseconds. The distributed potential detection device arranges potential sensors at each key node of the grounding network to collect the node potential value to the ground in real time. The node potential is combined with the current data to reflect the node's conductive performance. For example, at node B, when the current flowing through is 100 amperes, the ground potential rises to 80 volts, indicating that the grounding resistance at this location is about 0.8 ohms. The random forest classifier classifies the node characteristics based on the corresponding relationship between current and potential, and identifies node groups with similar conductive properties. The topological structure of the grounding network reflects the spatial layout and connection method of the conductive device. The grounding grid of a certain substation is rectangular, with a side length of 50 meters by 40 meters. There are crisscrossed flat steel grounding bodies inside, with a spacing of 5 meters, forming a total of 80 grid units. The minimum spanning tree algorithm constructs a network connection relationship based on the distance between nodes and the conductive performance, and optimizes the current transmission path. The soil resistivity distribution reflects the conductive characteristics of the stratum. The Wenner quadrupole method was used to measure and found that the soil resistivity at a depth of 0 to 2 meters in the surface layer is 100 ohms, 80 ohms at a depth of 2 to 5 meters, and 50 ohms below 5 meters. When modeling in layers, the stratum is divided into several cubic grids, and each grid is assigned a resistivity value at the corresponding depth to form a three-dimensional impedance distribution model. The transmission equation group of the multi-point grounding network includes node potential equations and network current equations, and the number of equations is equal to the total number of nodes.In the solution process, the reference node potential is set to 0 volts, and the Gaussian elimination method is used to iteratively calculate the potentials of other nodes. The impedance characteristic curve is fitted with a hyperbolic tangent function, and the function parameters are determined by the least squares method. The prediction accuracy reaches more than 90%. After long-term operation verification, the prediction model has a high accuracy rate in judging the ground potential fluctuation trend under transient conditions such as lightning strikes and short circuits.

[0015] Step S102, based on the predicted ground potential fluctuation trend, extract the coupling path of the interference signal in the ground network when common mode interference exists, determine the main node and propagation direction of the interference signal, and obtain the sensitive circuit.

[0016] Fast Fourier transform is used to decompose the ground potential fluctuation data sequence in the frequency domain, and the common mode interference signal characteristic quantity is obtained according to the ratio of the peak value of the spectrum component to the fundamental frequency; the ground potential fluctuation data sequence of adjacent nodes is cross-correlated according to the common mode interference signal characteristic quantity, and the electromagnetic coupling strength between nodes is obtained through the node correlation matrix; the phase difference operation is performed on the electromagnetic coupling strength between the nodes and the ground potential fluctuation data sequence, and the main propagation node coordinates are obtained according to the phase difference value; the gradient iteration method is used to construct a propagation vector field for the main propagation node coordinates, and the sensitive circuit distribution range is obtained according to the impedance change of each node in the propagation vector field.

[0017] Specifically, the ground potential fluctuation data sequence is obtained from the ground potential detector, and the fluctuation data is decomposed in the frequency domain by fast Fourier transform. The common mode interference signal is identified according to the ratio of the peak value of the spectrum component to the fundamental frequency, and the interference waveform feature is calculated by the harmonic distortion rate. The ground potential fluctuation data sequence of adjacent nodes is used for cross-correlation operation, and the node correlation matrix is ​​constructed according to the cross-correlation coefficient. The electromagnetic coupling strength between nodes is calculated by the correlation matrix. The phase difference operation is performed according to the electromagnetic coupling strength between nodes and the ground potential fluctuation data sequence. The propagation path network diagram is drawn according to the phase difference value. The node layout in the propagation path network diagram is classified and calculated by the support vector machine to obtain the main propagation node coordinates. The propagation vector field is constructed for the main propagation node coordinates, and the gradient iteration method is used to track the diffusion direction of the common mode interference waveform in the propagation vector field. The interference propagation directional diagram is obtained according to the impedance change law of each node. The interference propagation directional diagram is used to extract the coupling channel characteristic parameters, and the sensitivity evaluation function is constructed according to the coupling channel impedance and transmission delay values. The distribution range of the sensitive circuit is calculated by the sensitivity evaluation function. The common mode interference source in the grounding network usually comes from switching power supplies, inverters and other equipment, and the interference signal generated has periodic characteristics. At a substation site, the ground potential detector collected ground potential fluctuation data with a sampling frequency of 10 kHz and a recording time of 1 second. After fast Fourier transform analysis, it was found that there was a significant spectrum peak in the 20 kHz frequency band, with an amplitude of 0.8 times the fundamental wave and a harmonic distortion rate of 15%, indicating the presence of strong common-mode interference from the switching power supply. The electromagnetic coupling phenomenon between adjacent nodes can be identified by cross-correlation analysis. Two measurement nodes A and B with a distance of 5 meters were selected, and the cross-correlation coefficient of their ground potential fluctuation data was calculated to be 0.85, indicating that there was a strong coupling relationship between the two points. All adjacent node pairs were analyzed by a similar method to obtain a node correlation matrix. The matrix element values ​​were distributed between 0.2 and 0.9, and larger values ​​indicated the presence of a strong coupling channel. The construction of the propagation path network is based on the phase difference characteristics between nodes. Taking the reference node as a reference, the phase delay of the ground potential fluctuation of other nodes was calculated, and it was found that the interference signal propagated from northwest to southeast, and the phase difference between adjacent nodes was about 30 degrees. After the support vector machine classifies the nodes, three main propagation nodes are identified, which are located at coordinates (10,8), (15,12) and (20,15). In the propagation vector field, the diffusion of the interference signal shows obvious directionality. Through gradient iterative calculation, it is found that the diffusion speed of the interference wave around the main propagation node is 8 meters per microsecond, and the diffusion range decays with increasing distance. The node impedance gradually increases from the source point to the outside, increasing by 0.2 ohms within 5 meters, 0.5 ohms within 5 to 10 meters, and 1 ohm after exceeding 10 meters. The sensitivity of the coupling channel is closely related to its impedance characteristics and transmission delay. Set the sensitivity evaluation function, set the node impedance weight coefficient to 0.6, and the transmission delay weight coefficient to 0.4.The calculation results show that sensitive circuits are mainly distributed in three areas: within 5 meters close to the main propagation node, the sensitivity exceeds 0.8; on the line connecting adjacent main propagation nodes, the sensitivity is between 0.6 and 0.8; in areas far away from the main propagation node, the sensitivity is lower than 0.4.

[0018] Step S103, obtaining quantitative indicators of the common-mode suppression capability of the sensitive circuit, the quantitative indicators including common-mode suppression ratio and common-mode voltage gain, combining the coupling path of the interference signal in the grounding network, analyzing the impact of the interference signal on the circuit performance, the impact including signal distortion and noise increase.

[0019] The original input signal and the common-mode interference superposition signal output by the sensitive circuit signal detection device are obtained, and the common-mode interference superposition signal is collected by the sensitive circuit signal detection device; wavelet decomposition is used to perform multi-scale analysis on the common-mode interference superposition signal, and a common-mode voltage gain value is obtained according to the wavelet coefficient amplitude exceeding a preset threshold; a spectrum analysis function is established for the common-mode interference superposition signal, and a common-mode rejection ratio value is obtained through a frequency response curve of the spectrum analysis function; support vector regression is used to perform nonlinear fitting on the common-mode interference superposition signal, and a signal distortion influence factor is calculated according to the nonlinear fitting result; a noise gain function is constructed according to the common-mode voltage gain value and the common-mode rejection ratio value, and a common-mode interference influence index is obtained by weighting the noise gain function through the signal distortion influence factor.

[0020] Specifically, the original input signal and the common-mode interference superposition signal are obtained from the sensitive circuit signal detection device, and the superposition signal is analyzed at multiple scales by wavelet decomposition. The common-mode interference component is identified according to the energy distribution characteristics of the wavelet coefficient amplitude exceeding the preset threshold, and the common-mode voltage gain value is calculated by the ratio of the output signal to the input signal amplitude. A spectrum analysis function is established for the common-mode interference superposition signal, and the signal is decomposed in the frequency domain by discrete cosine transform. The different frequency components are arranged in descending order according to the spectrum amplitude, and the common-mode rejection ratio value is calculated by the frequency response curve of the input signal and the output signal. Support vector regression is used to perform nonlinear fitting on the common-mode interference superposition signal, and the harmonic distortion is calculated according to the signal amplitude and the coupling path impedance. The distortion distribution curve is drawn for the distortion value, and the signal quality parameter set is constructed by the distortion curve. The total harmonic distortion rate parameter is extracted from the signal quality parameter set, and the signal distortion impact factor is calculated by the distortion cumulative distribution function. The signal transmission quality curve is obtained by the mapping relationship between the impact factor and the coupling path length. The noise gain function is constructed according to the common-mode voltage gain value and the common-mode rejection ratio value, and the noise gain function is weighted by the signal distortion influence factor. The influence index of common-mode interference on circuit performance is obtained through the noise gain calculation result. The quantitative evaluation of common-mode interference in sensitive circuits mainly includes three key links: signal identification, index calculation and impact analysis. The signal detection device collects the original input signal and the common-mode interference superposition signal, the sampling rate is set to 100 kHz, and the recording time is 1 second. Through wavelet decomposition, it is found that the energy of the interference signal is mainly concentrated in the frequency band above 50 kHz, and the components with wavelet coefficient amplitude exceeding 0.5 volts are marked as common-mode interference. Comparing the signal amplitudes at the output and input ends, the common-mode voltage gain is calculated to be 1.8. Spectral analysis shows that common-mode interference has obvious high-frequency characteristics. In the discrete cosine transform results, the main frequency components are distributed at 60 kHz, 80 kHz and 100 kHz, and the corresponding amplitudes are 0.8 volts, 0.6 volts and 0.4 volts respectively. By comparing the frequency response curves of the input and output signals, the common mode rejection ratio is -20 dB at the 60 kHz frequency, indicating that the circuit has a certain inhibitory effect on interference in this frequency band. Nonlinear fitting reveals the specific characteristics of signal distortion. The kernel function of support vector regression uses radial basis function, and the penalty factor is set to 0.1. The fitting results show that when the coupling path impedance is 10 ohms, the harmonic distortion reaches 8%, which is much higher than the 2% level in normal operation. The distortion distribution curve shows an upward trend with the increase of coupling path impedance. When the impedance exceeds 20 ohms, the distortion growth rate is significantly accelerated. The total harmonic distortion rate in the signal quality parameter set reflects the degree of waveform distortion. In a certain test scenario, when the fundamental frequency is 10 kHz, the second harmonic distortion is 3%, the third harmonic distortion is 2%, the sum of the fourth and above harmonic distortions is 1%, and the total harmonic distortion rate is 3.7%. The cumulative distribution function shows that the total harmonic distortion rate of 90% of the test points is less than 5%.For every 1 meter increase in the coupling path, the signal transmission quality decreases by about 0.5%. The noise gain function comprehensively considers the effects of common-mode voltage gain and common-mode rejection ratio. When the common-mode voltage gain is 1.8 and the common-mode rejection ratio is -20 dB, the weighted coefficient of the superimposed signal distortion factor of 0.6 is calculated to be 1.5 times the noise gain. This means that common-mode interference causes the circuit noise level to increase by 50%, which has a significant impact on the signal processing accuracy.

[0021] Step S104, preset the common mode rejection ratio threshold and common mode voltage gain threshold of the sensitive circuit. If the actual common mode rejection ratio is lower than the common mode rejection ratio threshold, or the actual common mode voltage gain is higher than the common mode voltage gain threshold, it is considered that the anti-interference ability does not meet the requirements, and the node layout of the grounding network is adjusted, including increasing the number of connections, optimizing the connection topology, optimizing the impedance matching, or adjusting the grounding resistance of the key nodes.

[0022] The measured values ​​of the common mode rejection ratio and the common mode voltage gain of the sensitive circuit are obtained from the electrical parameter collector, and the node layout optimization instruction is obtained according to the preset common mode rejection ratio reference value and the preset common mode voltage gain reference value; the hierarchical clustering operation is performed on the sensitive circuit sampling data according to the node layout optimization instruction, and the node electrical correlation matrix is ​​obtained through the hierarchical clustering operation; the Markov random field is used to calculate the transmission probability value of the node electrical correlation matrix, and a key node identification table is constructed through the transmission probability value; the minimum spanning tree algorithm is used to construct the node connection path for the key node identification table, and the network layout adjustment plan is determined through the node connection path.

[0023] Specifically, the measured values ​​of the common mode rejection ratio and common mode voltage gain of the sensitive circuit are obtained from the electrical parameter collector, and the preset common mode rejection ratio reference value and the preset common mode voltage gain reference value are used as the judgment threshold. If the common mode rejection ratio is lower than the reference value or the common mode voltage gain is higher than the reference value, a node layout optimization instruction is generated. The voltage and current values ​​of each node are calculated according to the sensitive circuit sampling data, and the hierarchical clustering algorithm is used to extract the characteristics of the node voltage and current distribution, and the electrical correlation matrix between nodes is constructed through the characteristic data. The Markov random field is used to calculate the transmission probability of electrical correlation between nodes, and the importance of nodes is rated according to the transmission probability value. The key node identification table is constructed through the node rating results. The network connectivity parameters are calculated for the key node identification table data, and the minimum spanning tree algorithm is used to construct the shortest connection path between nodes. The coordinates of the newly added connection position are determined by the path length and the node distribution density. The network layout adjustment plan is generated according to the coordinates of the newly added connection position, and the dynamic programming algorithm is used to calculate the electrical impedance distribution between nodes. The grounding resistance values ​​of key nodes are updated through impedance gradient optimization. The quantitative evaluation and optimization process of the anti-interference ability of the grounding network involves multiple parameter indicators. In actual application scenarios, the common mode rejection ratio reference value is preset to -40 decibels, and the common mode voltage gain reference value is 1.2. These two thresholds are derived based on a large amount of test data statistics. When the electrical parameter collector detects that the common mode rejection ratio of the sensitive circuit is -35 decibels and the common mode voltage gain is 1.4, it indicates that the anti-interference performance does not meet the requirements, triggering the node layout optimization process. The node electrical characteristics analysis is based on real-time sampling data with a sampling frequency of 10 kHz. The sampling data shows that the voltage of node 1 is 20 volts and the current is 2 amperes; the voltage of node 2 is 18 volts and the current is 1.8 amperes; the voltage of node 3 is 15 volts and the current is 1.5 amperes. The hierarchical clustering algorithm divides these nodes into three levels according to the similarity of electrical characteristics, and the constructed electrical correlation matrix reflects the coupling strength between nodes. Markov random field calculations show that the probability of electrical correlation transmission from node 1 to node 2 is 0.8, and the probability of transmission from node 2 to node 3 is 0.6. Based on the transmission probability rating results, node 1 is marked as a first-level key node, node 2 is a second-level key node, and node 3 is a third-level key node. The node rating results directly affect the subsequent network optimization plan. The network connectivity calculation shows that the average connection distance between nodes in the existing layout is 5 meters, and the connectivity coefficient is 0.7. The minimum spanning tree algorithm generates the optimal connection path based on the physical distance and electrical correlation. It is recommended to add a new connection line between node 1 and node 3, and the connection point coordinates are (12, 15) meters. This connection method takes into account the matching of electrical characteristics while maintaining the shortest physical distance. The dynamic programming algorithm calculation shows that in the optimized network structure, the impedance distribution between nodes presents a gradient change characteristic. The impedance value near the key node is small, gradually increasing from 0.1 ohm to 0.5 ohm at the peripheral node.The new impedance distribution scheme enables the grounding current to be dispersed more evenly, reducing the grounding point resistance from the original 2 ohms to 0.8 ohms, while the impedance gradient between nodes is maintained at 0.05 ohms per meter.

[0024] Step S105, calculate the node potential distribution data of the grounding network after the node layout, and determine whether the potential difference between key nodes is within a preset range. If it exceeds the preset range, refine the network division, increase equipotential connection or locally reduce the grounding resistance.

[0025] The grounding network node-to-ground potential data collected by a multi-point potential detection device is obtained, and the node-to-ground potential data is spatially interpolated using a third-order Taylor series expansion to obtain a potential distribution function; the potential difference between key nodes is calculated based on the potential distribution function, and if the potential difference exceeds a preset potential difference range after support vector regression mapping, the potential distribution function is grid-subdivided to obtain a subdivided potential distribution map; the subdivided potential distribution map is divided into regions using a density clustering algorithm, and the boundary coordinates of the high potential difference region are determined based on the regional division results to obtain an equipotential connection path map; an ant colony algorithm is used to perform path optimization search based on the equipotential connection path map, and a variational method is used to discretely solve the current distribution function for the optimized path to obtain the grounding resistance configuration value of the key node.

[0026] Specifically, the ground potential data of each node of the grounding network is obtained from the multi-point potential detection device, and the detection data is spatially interpolated by the third-order Taylor series expansion. The potential distribution function is constructed by three-dimensional surface fitting to obtain the potential distribution diagram at the key node position. The potential difference between key nodes is calculated according to the potential distribution function, and the potential difference is interval mapped by support vector regression. If the mapping result exceeds the preset potential difference range, the potential distribution function is grid-subdivided. The potential contour map is constructed according to the grid subdivision operation result, and the contour map is divided into regions by density clustering algorithm. The boundary coordinates of the high potential difference area are determined by the regional division results. The equipotential connection path map is constructed for the boundary of the high potential difference area, and the ant colony algorithm is used to optimize the search for the connection path. The layout plan of the newly added equipotential connection line is determined by the path optimization results. The current distribution function between nodes is calculated according to the equipotential connection line layout plan, and the current distribution function is discretely solved by the variational method. The grounding resistance values ​​at the key nodes are reconfigured by the solution results. The optimization of the potential distribution of the grounding network nodes involves spatial electric field calculation and equipotential surface reconstruction. The multi-point potential detection device deployed voltage probes at each node of the network, with a sampling frequency of 1 kHz, and recorded 72 hours of potential fluctuation data. When using the third-order Taylor series expansion for spatial interpolation, the expansion center point is taken as the network geometric center, the expansion radius is 50 meters, and the interpolation accuracy reaches 0.1 volt. The three-dimensional potential distribution function obtained by fitting shows that the potential value is non-uniformly distributed in space, and the highest point appears at the coordinate (30, 20) meters, reaching 120 volts. The potential difference between key nodes is calculated based on the support vector regression method. The kernel function is selected as the Gaussian radial basis function, the penalty factor is set to 0.01, and the preset potential difference range value is plus or minus 10 volts. The actual calculation results show that the potential difference between node A and node B is 15 volts, and the potential difference between node B and node C is 12 volts, both of which exceed the preset range. In view of this situation, the original 5m×5m grid is subdivided into 2.5m×2.5m, which improves the spatial resolution. The potential contour map reflects the characteristics of the electric field distribution. The density clustering algorithm divides the space into three regions: the high potential area is greater than 100 volts, the medium potential area is 50 to 100 volts, and the low potential area is less than 50 volts. At the boundary of the high potential difference area, the potential gradient reaches 5 volts per meter, which is much higher than 1 volt per meter in other areas. The boundary coordinate points are mainly distributed in the range of (25, 15) to (35, 25) meters. The ant colony algorithm is used to optimize the path of the equipotential connection line. The pheromone volatility coefficient is set to 0.5, the number of iterations is 1000, and 4 new connection lines are optimized at the boundary of the high potential difference area. The starting and ending points of these connection lines are located at the coordinates (26, 16), (28, 18), (32, 22) and (34, 24) meters, respectively, and the connection direction is perpendicular to the equipotential line. The variational method is used to solve the current distribution function between nodes, and the continuous problem is discretized into 500 grid units.The calculation results show that the addition of equipotential bonding wires changes the direction of current flow, causing the originally concentrated large current (10 amperes) to be divided into multiple small currents (2 to 3 amperes). Based on this, the grounding resistance of key nodes was adjusted from the original 2 ohms to 0.5 ohms, achieving uniform current distribution.

[0027] Step S106, extracting the input and output signals of the sensitive circuit through the optimized ground network design, analyzing whether the anti-interference ability of the sensitive circuit is improved before and after the optimization, if the common mode rejection ratio is improved and the common mode voltage gain is reduced to the target value, it is considered that the target anti-interference ability is achieved.

[0028] The input signal waveform and the output signal waveform of the sensitive circuit signal collector are obtained, and time-frequency analysis is performed using wavelet packet decomposition to obtain a common-mode rejection ratio reference value; a transfer function is constructed for the input signal waveform and the output signal waveform, and the envelope characteristics of the signal waveform are extracted using Hilbert transform, and a common-mode voltage gain reference value is calculated according to the envelope characteristics; a suppression ratio change curve is constructed according to the common-mode rejection ratio reference value, and the common-mode rejection ratio improvement is obtained by curve integral calculation, and the common-mode rejection ratio improvement is compared with a preset increment threshold; a dual-parameter evaluation function is established using the common-mode rejection ratio improvement and the common-mode voltage gain reference value, and the evaluation function is classified using a support vector machine, and if the common-mode rejection ratio improvement is higher than the increment threshold and the gain reduction is greater than the reduction threshold, the optimized grounding network configuration parameters are output.

[0029] Specifically, the input signal waveform and the output signal waveform before and after optimization are obtained from the sensitive circuit signal collector, and the waveform data is subjected to multi-scale time-frequency analysis by wavelet packet decomposition, and the common mode rejection ratio reference value is calculated by the high-frequency component energy ratio. A transfer function is constructed for the input signal waveform and the output signal waveform, and the waveform envelope feature is extracted by Hilbert transform, and the common mode voltage gain reference value is calculated according to the envelope curve amplitude ratio. A suppression ratio change curve is constructed according to the common mode rejection ratio reference value, and the common mode rejection ratio improvement is calculated by curve integration, and the improvement value is compared with the preset increment threshold value. A gain change curve is constructed according to the common mode voltage gain reference value, and the common mode voltage gain reduction is calculated by curve integration, and the reduction value is compared with the preset reduction threshold value. A dual-parameter evaluation function is established for the suppression ratio improvement and gain reduction data, and the evaluation parameters are classified and defined by support vector machine. If the suppression ratio improvement is higher than the increment threshold and the gain reduction is greater than the reduction threshold, the optimized grounding network configuration parameters are output. The evaluation of the anti-interference ability of sensitive circuits is completed by waveform comparison. The signal collector records the input and output waveforms at the same time at a sampling frequency of 100 kHz, and the sampling time is 1 second. The input signal amplitude before optimization is 5 volts, and the output has a spike interference of 8 volts. The wavelet packet decomposition results show that the high-frequency component energy accounts for 35%, and the calculated common mode rejection ratio reference value is -15 decibels. Transfer function analysis reveals the signal transmission characteristics. In the envelope curve extracted by Hilbert transform, the output waveform envelope peak is 1.6 times that of the input, indicating that there is a significant common mode interference amplification phenomenon. This 1.6 times is the common mode voltage gain reference value, which intuitively reflects the sensitivity of the circuit to common mode interference. The optimized common mode rejection ratio shows a significant improvement. The rejection ratio change curve always maintains an upward trend in the frequency domain range of 0 to 100 kHz. The curve integral calculation shows that the common mode rejection ratio is improved by 20 decibels, which is much higher than the preset 10 decibel increment threshold. This improvement is mainly reflected in the high frequency band, and the interference of frequencies above 50 kHz is effectively suppressed. The change of common-mode voltage gain is also significant. The gain change curve shows a continuous decline. The curve integration result shows that the common-mode voltage gain is reduced by 0.8, which exceeds the preset 0.5 reduction threshold. The peak value of the output waveform after optimization is reduced to 5.8 volts, and the ratio to the input signal is reduced to 1.16, indicating that the common-mode interference is significantly weakened. The dual-parameter evaluation function comprehensively considers the improvement of the suppression ratio and the gain. The support vector machine uses a two-dimensional feature space, with the horizontal axis representing the suppression ratio improvement and the vertical axis representing the gain reduction. The classification interface is set at (10 dB, 0.5). The measured data point (20 dB, 0.8) is located in the target area, confirming that the optimized grounding network configuration does achieve the expected anti-interference level. The final network configuration parameters include the coordinates of the 25 grounding point locations, and the connection impedance between nodes is distributed in the range of 0.1 to 0.8 ohms.

[0030] Step S107, determining a matching design scheme between the grounding network and the sensitive circuit according to the anti-interference capability, building a test platform, and testing the stability data of the grounding system under different electromagnetic environments.

[0031] A grounding network matching parameter table is generated according to the anti-interference ability index of the sensitive circuit, and a recursive neural network is used to optimize the wire connection impedance and the node spacing parameters to obtain the node layout and wire connection specification data; the electromagnetic field distribution function is used to construct a field strength distribution spectrum, and the interference source strength and frequency parameters are set according to the field strength gradient characteristics in the field strength distribution spectrum to obtain an electromagnetic interference source characteristic data table; multiple electromagnetic interference sources are arranged according to the electromagnetic interference source characteristic data table, and a spatial field strength distribution diagram is obtained by superimposing and calculating the interference source strength and frequency; multiple electromagnetic field detectors are arranged according to the spatial field strength distribution diagram, and the detection data is stored by a synchronous data recorder. If the detection data recording curve calculates the circuit common mode suppression index is less than a preset threshold, it is determined that the stability level of the sensitive circuit meets the standard.

[0032] Specifically, a grounding network matching parameter table is generated according to the anti-interference ability index of the sensitive circuit, and a recursive neural network is used to optimize the wire connection impedance and node spacing parameters. The fitness of the optimized parameter combination is evaluated by the genetic algorithm to obtain the node layout and wire connection specification parameters. The electromagnetic field distribution function is used to construct the field strength distribution map, and the interference source strength and frequency parameters are set according to the field strength gradient characteristics. Different electromagnetic field strength matrices are generated by combining multiple sets of parameters to obtain the electromagnetic interference source characteristic data table. According to the electromagnetic interference source characteristic data table, an experimental test bench is built, and multi-point electromagnetic interference sources are arranged around the sensitive circuit. The spatial field strength distribution map is obtained by superposition calculation of the interference source strength and frequency. Multi-point electromagnetic field detectors are arranged according to the spatial field strength distribution map, and a synchronous data recorder is used to store the detection data. The circuit common mode suppression index is calculated according to the data recording curve, and the stability level of the sensitive circuit is judged by the common mode suppression index. The grounding network matching parameters are adjusted according to the stability level of the sensitive circuit, and the node layout and wire connection scheme are updated by the automatic parameter configurator. The circuit operation data under various electromagnetic environments is recorded through repeated experimental tests. The test and verification process of the grounding network matching design involves multiple links. In the parameter optimization stage, the recursive neural network adopts a three-layer structure. The input layer receives the node spacing and wire impedance data, the number of hidden layer neurons is 128, and the output layer generates the optimized parameter combination. The genetic algorithm sets the population size to 200, the crossover probability to 0.8, and the mutation probability to 0.1. After 50 generations of iterations, the optimal solution is obtained, and the node spacing is finally determined to be 3 meters and the wire connection impedance is 0.2 ohms. The electromagnetic field distribution characteristic test adopts a multi-point excitation method. The interference sources are arranged around the test area, with an intensity range from 1 volt per meter to 10 volts per meter, and a frequency range of 10 kilohertz to 1 megahertz. The field intensity distribution map shows that the field strength reaches a maximum of 8 volts per meter at 1 meter away from the interference source, and decays exponentially with increasing distance, dropping to 2 volts per meter at 5 meters. The experimental test bench adopts a standardized configuration. The four interference sources are located in the four corners of the test area. The frequency interval of each interference source is set to 10 kilohertz to avoid frequency overlap. After the superposition of multiple interference points, the maximum field strength at the center of the test area reached 15 volts per meter, and the minimum field strength was 5 volts per meter, forming a large electromagnetic field strength gradient. The electromagnetic field detection adopts 8-channel synchronous acquisition. The detectors are arranged at key positions around the sensitive circuits. The sampling frequency is set to 10 MHz and the recording time is 10 seconds. The test data shows that under an interference field strength of 15 volts per meter, the circuit common mode rejection ratio is maintained above -40 decibels, and the common mode voltage gain is less than 1.2, indicating that the grounding network has good anti-interference performance. The automatic parameter configuration process reflects the adaptability of the test system. When the common mode rejection ratio is detected to be lower than -35 decibels, the automatic configurator will reduce the distance between the nearest nodes to 2 meters and the wire connection impedance to 0.15 ohms.Repeated test results show that the optimized grounding network can still maintain stable operation in a strong interference environment of 20 volts per meter, and the common mode rejection ratio is improved to -45 decibels.

[0033] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for testing the stability of electronic products, characterized in that: The method comprises: Acquire the propagation characteristic data of transient current in the grounding network, combine it with the potential distribution data of each node, establish a multi-point grounding network model, input the node potential and transient current data, and predict the fluctuation trend of ground potential through the multi-point grounding network model; According to the predicted ground potential fluctuation trend, the coupling path of the interference signal in the ground network when common mode interference exists is extracted, the main node and propagation direction of the interference signal are determined, and the sensitive circuit is obtained, including: Fast Fourier transform is used to decompose the ground potential fluctuation data sequence in the frequency domain, and the common mode interference signal characteristic quantity is obtained according to the ratio of the peak value of the spectrum component to the fundamental frequency. According to the common mode interference signal characteristic quantity, a cross-correlation operation is performed on the ground potential fluctuation data sequence of adjacent nodes, and the electromagnetic coupling strength between nodes is obtained through the node correlation matrix; Performing phase difference calculation on the electromagnetic coupling strength between the nodes and the ground potential fluctuation data sequence, and obtaining the coordinates of the main propagation node according to the phase difference value; A propagation vector field is constructed for the coordinates of the main propagation nodes using a gradient iteration method, and a distribution range of sensitive circuits is obtained according to impedance changes of each node in the propagation vector field; Obtaining quantitative indicators of the common-mode rejection capability of sensitive circuits, including common-mode rejection ratio and common-mode voltage gain, and analyzing the impact of interference signals on circuit performance in combination with the coupling path of interference signals in the grounding network, including signal distortion and noise increase; The common mode rejection ratio threshold and common mode voltage gain threshold of the sensitive circuit are preset. If the actual common mode rejection ratio is lower than the common mode rejection ratio threshold, or the actual common mode voltage gain is higher than the common mode voltage gain threshold, it is considered that the anti-interference ability does not meet the requirements, and the node layout of the grounding network is adjusted, including increasing the number of connections, optimizing the connection topology, optimizing the impedance matching, or adjusting the grounding resistance of key nodes; Calculate the node potential distribution data of the grounding network after the node layout, and determine whether the potential difference between key nodes is within the preset range. If it exceeds the preset range, refine the network division, increase equipotential connection or locally reduce the grounding resistance; Through the optimized grounding network design, the input and output signals of the sensitive circuit are extracted to analyze whether the anti-interference ability of the sensitive circuit is improved before and after optimization. If the common-mode rejection ratio is improved and the common-mode voltage gain is reduced to the target value, it is considered that the target anti-interference ability has been achieved; Determine the matching design scheme between the grounding network and the sensitive circuit based on the anti-interference capability, build a test platform, and test the stability data of the grounding system in different electromagnetic environments.

2. The method according to claim 1, characterized in that The method comprises: obtaining the propagation characteristic data of transient current in the grounding network, combining the potential distribution data of each node, establishing a multi-point grounding network model, inputting the node potential and transient current data, and predicting the fluctuation trend of the ground potential through the multi-point grounding network model, including: Acquire transient current sampling data from a formation current detection device, and process the sampling data using an adaptive sampling algorithm to obtain a node current transmission characteristic data set; Acquire the grounding network node potential value according to the node current transmission characteristic data set, and use a random forest classifier to extract features of the node potential value to obtain a node feature vector; Generate a grounding network topology unit according to the node feature vector, and use a minimum spanning tree algorithm to connect and configure the grounding network topology unit to obtain a network connection relationship matrix; A multi-point grounding network transmission equation group is constructed according to the node characteristic vector and the network connection relationship matrix, and the multi-point grounding network transmission equation group is solved by iterative operation to obtain the potential difference distribution data between nodes.

3. The method according to claim 1, characterized in that The quantitative indicators of the common mode rejection capability of the sensitive circuit are obtained, and the quantitative indicators include common mode rejection ratio and common mode voltage gain. Combined with the coupling path of the interference signal in the ground network, the influence of the interference signal on the circuit performance is analyzed, and the influence includes signal distortion and noise increase, including: Acquire an original input signal and a common-mode interference superposition signal output by a sensitive circuit signal detection device, wherein the common-mode interference superposition signal is acquired by the sensitive circuit signal detection device; Using wavelet decomposition to perform multi-scale analysis on the common-mode interference superposition signal, and obtaining a common-mode voltage gain value according to the amplitude of the wavelet coefficient exceeding a preset threshold; Establishing a spectrum analysis function for the common-mode interference superposition signal, and obtaining a common-mode rejection ratio value through a frequency response curve of the spectrum analysis function; Performing nonlinear fitting on the common-mode interference superposition signal using support vector regression, and calculating the signal distortion influence factor according to the nonlinear fitting result; A noise gain function is constructed according to the common-mode voltage gain value and the common-mode rejection ratio value, and a common-mode interference influence index is obtained by performing a weighted operation on the noise gain function through a signal distortion influence factor.

4. The method according to claim 1, characterized in that: The common mode rejection ratio threshold and the common mode voltage gain threshold of the preset sensitive circuit are set. If the actual common mode rejection ratio is lower than the common mode rejection ratio threshold, or the actual common mode voltage gain is higher than the common mode voltage gain threshold, it is considered that the anti-interference ability does not meet the requirements, and the node layout of the grounding network is adjusted, including increasing the number of connections, optimizing the connection topology, optimizing the impedance matching, or adjusting the grounding resistance of the key nodes, including: Obtain the measured values ​​of the common mode rejection ratio and the common mode voltage gain of the sensitive circuit from the electrical parameter collector, and obtain the node layout optimization instruction according to the preset common mode rejection ratio reference value and the preset common mode voltage gain reference value; Performing a hierarchical clustering operation on the sensitive circuit sampling data according to the node layout optimization instruction, and obtaining a node electrical correlation matrix through the hierarchical clustering operation; Using a Markov random field to calculate the transmission probability value of the node electrical correlation matrix, and constructing a key node identification table according to the transmission probability value; A minimum spanning tree algorithm is used to construct a node connection path for the key node identification table, and a network layout adjustment plan is determined through the node connection path.

5. The method according to claim 1, characterized in that The node potential distribution data after calculating the node layout of the grounding network is judged whether the potential difference between key nodes is within a preset range. If it exceeds the preset range, the network division is refined, the equipotential connection is increased, or the grounding resistance is partially reduced, including: Acquire the ground potential data of the grounding network nodes collected by the multi-point potential detection device, perform spatial interpolation operation on the ground potential data of the nodes using a third-order Taylor series expansion, and obtain a potential distribution function; Calculate the potential difference between key nodes according to the potential distribution function, and if the potential difference exceeds the preset potential difference range after support vector regression mapping, perform grid subdivision operation on the potential distribution function to obtain a subdivided potential distribution diagram; Using a density clustering algorithm to divide the subdivided potential distribution map into regions, determining the boundary coordinates of the high potential difference region through the regional division results, and obtaining an equipotential connection path map; According to the equipotential connection path diagram, an ant colony algorithm is used to perform path optimization search, and a variational method is used to discretely solve the current distribution function for the optimized path to obtain the grounding resistance configuration value of the key node.

6. The method according to claim 1, characterized in that The optimized ground network design is used to extract the input and output signals of the sensitive circuit, and analyze whether the anti-interference ability of the sensitive circuit is improved before and after the optimization. If the common mode rejection ratio is improved and the common mode voltage gain is reduced to the target value, it is considered that the target anti-interference ability has been achieved, including: Obtain the input and output signal waveforms of the sensitive circuit signal collector, perform time-frequency analysis using wavelet packet decomposition, and obtain a common mode rejection ratio reference value; Constructing a transfer function for the input-end signal waveform and the output-end signal waveform, extracting envelope characteristics of the signal waveform using Hilbert transform, and calculating a common-mode voltage gain reference value according to the envelope characteristics; Constructing a rejection ratio variation curve according to the common mode rejection ratio reference value, obtaining a common mode rejection ratio improvement amount by curve integral calculation, and comparing the common mode rejection ratio improvement amount with a preset increment threshold; A dual-parameter evaluation function is established by using the common-mode rejection ratio improvement and the common-mode voltage gain reference value, and the evaluation function is classified by a support vector machine. If the common-mode rejection ratio improvement is higher than the increment threshold and the gain reduction is greater than the reduction threshold, the optimized grounding network configuration parameters are output.

7. The method according to claim 1, characterized in that The matching design scheme of the grounding network and the sensitive circuit is determined according to the anti-interference capability, and a test platform is built to test the stability data of the grounding system under different electromagnetic environments, including: Generate a ground network matching parameter table based on the anti-interference capability index of the sensitive circuit, use a recursive neural network to optimize the wire connection impedance and node spacing parameters, and obtain node layout and wire connection specification data; The electromagnetic field distribution function is used to construct a field intensity distribution spectrum, and the interference source intensity and frequency parameters are set according to the field intensity gradient characteristics in the field intensity distribution spectrum to obtain an electromagnetic interference source characteristic data table; Arrange multiple electromagnetic interference sources according to the electromagnetic interference source characteristic data table, and obtain a spatial field strength distribution diagram by superimposing and calculating the interference source strength and frequency; Multi-point electromagnetic field detectors are arranged according to the spatial field strength distribution diagram, and the detection data is stored by a synchronous data recorder. If the circuit common mode suppression index calculated by the detection data recording curve is less than a preset threshold, it is determined that the stability level of the sensitive circuit meets the standard.

Images