Anti-interference test method and system for three-phase electric energy meter

Abstract

This invention provides a method and system for anti-interference testing of three-phase electricity meters, relating to the field of power equipment reliability testing technology. The method involves applying an initial excitation strategy to the three-phase electricity meter through a programmable test platform, simultaneously acquiring performance degradation vectors, and quantifying sensitivity evaluation values ​​based on a failure threshold system. K highly sensitive candidate strategies are selected in descending order of sensitivity and subjected to adaptive neighborhood mutation to generate a boundary detection strategy for critical gradient testing, acquiring K sets of boundary performance vectors. Based on these vectors, gradient-driven boundary convergence iteration is performed in a multi-dimensional parameter space, outputting a critical failure boundary test set. This invention addresses the technical problem of existing anti-interference testing techniques for three-phase electricity meters, which only obtain discrete failure points and ignore multi-dimensional interference coupling effects, leading to missed failure modes and weak engineering application value of the test data. It achieves the effect of generating a test set with high engineering practical value.

Description

Technical Field

[0001] This invention relates to the field of power equipment reliability testing technology, specifically to an anti-interference testing method and system for three-phase energy meters. Background Technology

[0002] Current methods for testing the anti-interference capabilities of three-phase electricity meters have significant limitations. Specifically, the industry-standard stepped single-parameter scanning method can only acquire discrete failure points and cannot construct continuous failure boundaries. Actual power grid disturbances are the result of multiple factors such as voltage, frequency, and load imbalance, but existing methods neither consider the interactions between parameters nor are forced to simplify operating condition combinations due to testing cost constraints, such as fixing the load rate and ignoring different harmonic injection paths. This leads to the systematic omission of key failure modes caused by composite interference.

[0003] Ultimately, discrete test data is difficult to transform into effective parameters to guide product optimization and operation and maintenance early warning, which seriously restricts the improvement of the reliability of electricity meters.

[0004] In summary, existing technologies for anti-interference testing of three-phase energy meters can only obtain discrete failure points and ignore multi-dimensional interference coupling effects, resulting in missed failure modes and weak engineering application value of the test data. Summary of the Invention

[0005] This application provides an anti-interference test method and system for three-phase energy meters, which solves the technical problem that the existing anti-interference test of three-phase energy meters can only obtain discrete failure points and ignore multi-dimensional interference coupling effects, resulting in missed failure modes and weak engineering application value of test data.

[0006] In view of the above problems, this application provides an anti-interference test method and system for three-phase energy meters.

[0007] The first aspect of this application provides an interference immunity testing method for a three-phase energy meter. The method includes: simultaneously triggering a monitoring equipment array to perform multi-dimensional test data acquisition while a programmable immunity test platform applies multiple initial excitation strategies to the three-phase energy meter, obtaining multiple performance degradation vectors, wherein the multiple initial excitation strategies originate from a multi-dimensional test parameter space; quantifying the interference sensitivity of the multiple performance degradation vectors based on a failure threshold system to obtain multiple sensitivity evaluation values; selecting and extracting K high-sensitivity candidate strategies from the multiple initial excitation strategies according to the descending order of the multiple sensitivity evaluation values; performing adaptive neighborhood mutation on the K high-sensitivity candidate strategies to obtain K sets of boundary detection strategies; controlling the programmable immunity test platform to perform critical gradient testing on the three-phase energy meter to collect the K sets of boundary performance vectors; and performing gradient-driven boundary convergence iteration in the multi-dimensional test parameter space based on the failure trend measurement of the K sets of boundary performance vectors relative to the failure threshold system until a critical failure boundary test set is output.

[0008] A second aspect of this application provides an anti-interference testing system for three-phase energy meters. The system includes: an anti-interference control module, configured to synchronously trigger a monitoring device array to perform multi-dimensional test data acquisition to obtain multiple performance degradation vectors when a programmable anti-interference test platform applies multiple initial excitation strategies to the three-phase energy meter, wherein the multiple initial excitation strategies originate from a multi-dimensional test parameter space; an anti-interference performance quantification module, configured to quantify the interference sensitivity of the multiple performance degradation vectors based on a failure threshold system to obtain multiple sensitivity evaluation values; and a strategy selection module, configured to select strategies based on the multiple sensitivity... The sensitivity evaluation values ​​are sorted in descending order, and K high-sensitivity candidate strategies are extracted from the multiple initial excitation strategies. The strategy mutation module is used to perform adaptive neighborhood mutation on the K high-sensitivity candidate strategies to obtain K sets of boundary detection strategies. Then, the programmable disturbance rejection test platform is controlled to perform critical gradient tests on the three-phase energy meter to collect K sets of boundary performance vectors. The critical strategy iteration module is used to perform gradient-driven boundary convergence iteration in the multi-dimensional test parameter space according to the failure trend measurement of the K sets of boundary performance vectors relative to the failure threshold system until the critical failure boundary test set is output.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: The method provided in this application, when applying multiple initial excitation strategies to a three-phase energy meter using a controlled programmable immunity test platform, simultaneously triggers a monitoring equipment array to perform multi-dimensional test data acquisition, obtaining multiple performance degradation vectors. These initial excitation strategies originate from a multi-dimensional test parameter space. Based on a failure threshold system, the interference sensitivity of these multiple performance degradation vectors is quantified to obtain multiple sensitivity evaluation values. According to the descending order of these sensitivity evaluation values, K high-sensitivity candidate strategies are selected from the initial excitation strategies. Adaptive neighborhood mutation is performed on these K high-sensitivity candidate strategies to obtain K sets of boundary detection strategies. Then, the controlled programmable immunity test platform is controlled to perform critical gradient testing on the three-phase energy meter to collect the K sets of boundary performance vectors. Based on the failure trend measurement of the K sets of boundary performance vectors relative to the failure threshold system, gradient-driven boundary convergence iteration is performed in the multi-dimensional test parameter space until a critical failure boundary test set is output. This achieves the technical effects of accurately capturing multi-dimensional interference coupling effects, realizing millimeter-level failure boundary location, eliminating missed failure modes, and generating a test set with high engineering practice value. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the anti-interference test method for three-phase energy meters provided in this application.

[0011] Figure 2 A schematic diagram of the anti-interference testing system for a three-phase energy meter provided in this application.

[0012] Figure labeling: Anti-interference control module 10, anti-interference performance quantification module 20, strategy screening module 30, strategy mutation module 40, critical strategy iteration module 50. Detailed Implementation

[0013] This application provides an anti-interference testing method and system for three-phase energy meters, addressing the technical problem that existing anti-interference tests for three-phase energy meters only obtain discrete failure points and ignore multi-dimensional interference coupling effects, leading to missed failure modes and weak engineering application value of the test data. It achieves the technical effects of accurately capturing multi-dimensional interference coupling effects, realizing millimeter-level failure boundary location, eliminating missed failure modes, and generating test sets with high engineering practical value.

[0014] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0015] Example 1, as Figure 1 As shown, this application provides an anti-interference test method for three-phase energy meters, the method comprising: A100: When the control programmable immunity test platform applies multiple initial excitation strategies to the three-phase energy meter, the monitoring equipment array is synchronously triggered to perform multi-dimensional test data acquisition to obtain multiple performance degradation vectors, wherein the multiple initial excitation strategies are derived from the multi-dimensional test parameter space.

[0016] A three-phase energy meter is a metering device used to measure three-phase AC energy. In this embodiment, the three-phase energy meter is an industrial-grade smart energy meter product of an unspecified model, and the anti-interference test method is universal.

[0017] The excitation strategy is essentially a preset combination of interference parameters, including programmable parameters such as voltage / current amplitude, frequency, waveform characteristics and injection path. The programmable immunity test platform is an automated test device that integrates an interference signal generator, a safety protection unit and a data acquisition system, and has the ability to switch parameters at the millisecond level.

[0018] During the process of applying multiple initial excitation strategies to the three-phase energy meter on the programmable immunity test platform, the multi-source synchronous triggering function coordinates the monitoring equipment array to collect multi-dimensional test data in real time and generates multiple performance degradation vectors.

[0019] The initial excitation strategy originates from a multi-dimensional test parameter space, which is constructed based on the failure response characteristics of three-phase energy meters. It covers various interference characteristics such as voltage sags and high-frequency noise, various operating conditions such as load imbalance and power factor changes, and various injection paths such as L1-PE coupling and L2-L3 line-to-line injection.

[0020] Specifically, multiple initial excitation strategies are generated in the multidimensional test parameter space through Latin hypercube sampling to ensure that the sampling points uniformly cover the parameter space. This embodiment will elaborate on the method of obtaining the multiple initial excitation strategies in the subsequent specification section.

[0021] During test operation, the safety isolation function continuously monitors the safety parameters of the electricity meter, such as temperature rise, leakage current harmonic distortion. If the parameters are within the safety threshold, the metering error offset, communication bit error rate, function abnormality flag and abnormal recovery time are collected synchronously. If the safety parameters exceed the limit, the current performance degradation vector is recorded and the test is interrupted; otherwise, it is marked as an empty set.

[0022] The resulting multiple performance degradation vectors, corresponding to the multiple initial excitation strategies, accurately characterize the device performance degradation state under different excitation strategies, providing input for subsequent sensitivity quantization.

[0023] A200: Based on the failure threshold system, the interference sensitivity of the multiple performance degradation vectors is quantified to obtain multiple sensitivity evaluation values.

[0024] The failure threshold system includes metering performance failure threshold, communication performance failure threshold, functional safety failure threshold, and abnormal recovery failure threshold. In this embodiment, different weight coefficients are assigned to the four types of failure thresholds according to the specific equipment type of the three-phase energy meter. The principle of threshold allocation is that metering performance has the highest weight, followed by communication performance.

[0025] The degradation ratio of each performance degradation vector relative to the failure threshold is calculated using a weighted average. Specifically, the measured performance value is compared with the corresponding threshold to obtain the relative deviation, and then a weighted average is calculated according to the weight coefficients. For empty set vectors that did not trigger safety protection and showed no performance degradation during the test, their sensitivity evaluation value is directly set to zero.

[0026] The quantification method in this embodiment transforms complex multidimensional performance states into a unified sensitivity index, objectively reflecting the vulnerability of three-phase energy meters under various interferences.

[0027] A300: Based on the descending order of the multiple sensitivity evaluation values, select and extract K high-sensitivity candidate strategies from the multiple initial excitation strategies.

[0028] The calculated sensitivity evaluation values ​​are sorted in descending order to form a strategy sequence from high sensitivity to low sensitivity. Combined with a preset sensitivity threshold, such as S>0.7, K high-sensitivity candidate strategies are selected and extracted from the initial activation strategies.

[0029] The screening process prioritizes strategies whose sensitivity evaluation values ​​exceed a threshold. If the number of strategies exceeding the threshold is less than K, the remaining slots are filled in descending order. If there are more than K strategies exceeding the threshold, the top K high-sensitivity candidate strategies are selected.

[0030] These K highly sensitive candidate strategies represent the combinations of interference parameters most likely to cause performance degradation in electricity meters. Their sensitivity evaluation values ​​are significantly higher than the average level, providing a crucial starting point for subsequent boundary detection. The screening process strictly follows the principle of threshold priority + sorting and replacement to ensure that the set of highly sensitive candidate strategies simultaneously covers both significant failure risk areas and potentially sensitive areas.

[0031] A400: After performing adaptive neighborhood mutation on the K high-sensitivity candidate strategies to obtain K sets of boundary detection strategies, the programmable immunity test platform is controlled to perform critical gradient tests on the three-phase energy meter to collect K sets of boundary performance vectors.

[0032] This embodiment first classifies the sensitivity levels of the K high-sensitivity candidate strategies based on their corresponding sensitivity evaluation values, and sets K mutation amplitudes. For example, high-sensitivity strategies use small-step fine-grained mutation, while low-sensitivity strategies use large-step exploratory mutation. Then, neighborhood mutation operations are performed on the continuous dimensions, discrete working conditions, and injection path topology of the multidimensional test parameter space to generate K initial detection strategy sets for the K high-sensitivity candidate strategies. Finally, deduplication is performed by comparing multiple historical initial excitation strategies to ensure the innovativeness and effectiveness of the boundary detection strategies, resulting in K sets of boundary detection strategies.

[0033] Finally, the K-group boundary detection strategy was loaded onto the programmable immunity test platform. Under strict safety monitoring, critical gradient tests were performed to collect key performance vector data such as metering error offset and communication bit error rate of the three-phase energy meter at the quasi-failure boundary, and to obtain the K-group boundary performance vector corresponding to the K-group boundary detection strategy.

[0034] A500: Based on the failure trend measurement of the K sets of boundary performance vectors relative to the failure threshold system, perform gradient-driven boundary convergence iteration in the multidimensional test parameter space until the critical failure boundary test set is output.

[0035] The sensitivity evaluation value of the K sets of boundary performance vectors relative to the failure threshold system is calculated. This evaluation value characterizes the trend strength of the energy meter's performance degradation approaching the failure critical point. Using this as a navigation basis, a gradient-driven boundary localization process is executed in the multi-dimensional test parameter space.

[0036] Specifically, a gradient field space covering dimensions such as voltage amplitude, frequency, load conditions, and injection path is first constructed to quantify the influence of parameter changes on sensitivity. The parameter points are then updated along the negative gradient direction of sensitivity increase. An adaptive step size strategy is adopted in the continuous dimension, and structural projection is implemented in the discrete topology dimension.

[0037] The iterative process continuously monitors the parameter displacement, sensitivity change rate, and gradient magnitude. Convergence is determined when the displacement is less than a set threshold, the sensitivity fluctuation stabilizes, and the gradient magnitude drops below the critical value.

[0038] Spatial aggregation analysis is performed on the boundary point set that has reached convergence to identify key failure boundary surfaces and vulnerable topological paths, and finally a structured critical failure boundary test set is output. The critical failure boundary test set is essentially a structured data set labeled with critical parameter coordinates and failure mode types. Here, the parameter coordinates are used to clarify the critical thresholds of parameters such as voltage amplitude, frequency, and load rate that interfere with the normal function of the three-phase energy meter, i.e., the safety red line of anti-interference capability. The failure mode type is the specific failure mode of the three-phase energy meter failure event under the influence of the interference index corresponding to the parameter coordinates.

[0039] This embodiment achieves millimeter-level precise positioning of the anti-interference failure boundary of a three-phase energy meter through gradient-driven boundary convergence iteration, providing a reproducible and standardized evaluation benchmark for the anti-interference performance of the energy meter.

[0040] Furthermore, the method step A100 provided by the present invention includes: A100-1: Interact to obtain the failure response characteristics of the three-phase energy meter, wherein the failure response characteristics cover multiple interference characteristics, multiple operating conditions and multiple injection paths; A100-2: Construct a multi-dimensional test parameter space based on the failure response characteristics; A100-3: Generate the multiple initial excitation strategies in the multi-dimensional test parameter space through Latin hypercube sampling.

[0041] Furthermore, by generating the multiple initial excitation strategies in the multidimensional test parameter space through Latin hypercube sampling, the method step A100-3 provided by this invention includes: A100-31: Deconstruct the multidimensional test parameter space to obtain multiple continuous interference dimensions corresponding to the multivariate interference characteristics, multiple sets of discrete operating condition settings corresponding to the multiple operating conditions, and multiple port coupling topologies corresponding to the multiple injection paths; A100-32: Perform equal probability interval partitioning sampling on the multiple continuous interference dimensions to obtain multiple parameter hierarchical interval sampling values; A100-33: Perform combined enumeration on the multiple sets of discrete operating condition settings to obtain multiple operating condition combination points; A100-34: Perform combined enumeration on the multiple port coupling topologies to obtain multiple injection path combinations; A100-35: After performing sequential cyclic mapping on the multiple parameter hierarchical interval sampling values, multiple operating condition combination points, and multiple injection path combinations, eliminate parameter correlation based on multidimensional independent random permutation to output the multiple initial excitation strategies.

[0042] Key failure response characteristics of three-phase energy meters are obtained through a human-machine interface. These characteristics cover three core dimensions: multi-dimensional interference characteristics, including electromagnetic interference types such as voltage sags, surges, harmonic distortion, and high-frequency noise; multiple operating conditions, including balanced load, 30% unbalanced load, and capacitive / inductive power factor combinations; and multiple injection paths, including topology connection methods such as L1 phase-to-protective ground injection and L2-L3 line coupling injection. These failure response characteristics are used to establish a target constraint framework for testing, ensuring the completeness and engineering relevance of the parameter space construction.

[0043] Furthermore, the abstract interference characteristics are transformed into continuous parameter dimensions for programmable control, such as voltage amplitude setting range of 0-150% of rated value and frequency adjustment bandwidth of 40Hz-1kHz. Operating conditions are mapped into discrete parameter dimensions, such as load imbalance levels of 10% / 30% / 50% and power factor enumeration values ​​of 0.5L / 0.8C / 1.0. Injection paths are transformed into topology parameter dimensions, such as port coupling matrices [L1,PE], [L2,L3,N], and other connection patterns. This constructs the multidimensional test parameter space, which strictly follows the mathematical expression of failure response characteristics. Each dimension corresponds to the triggering conditions of a specific failure mechanism, forming a parameterized test domain covering all potential failure modes.

[0044] Based on the multi-various interference characteristics, multiple operating conditions, and multiple injection paths covered by the failure response characteristics, this embodiment decouples and reconstructs the multi-dimensional test parameter space into three types of parameter subspaces.

[0045] Specifically, for the continuously changing multi-dimensional interference characteristics such as voltage sags and high-frequency noise, continuously adjustable parameters such as amplitude and frequency are extracted as the continuous dimension of interference. For various operating conditions such as load imbalance and power factor fluctuations, their preset combinations are defined as multiple sets of discrete operating condition settings. For the topological path differences of multiple injection path combinations such as L1-PE coupling and L2-L3 line-to-line injection, the differences are abstracted into discrete connection modes of port-coupled topologies. This deconstruction process establishes a strict mapping relationship between the parameter space and failure response characteristics, ensuring that subsequent sampling fully covers the triggering conditions of all potential failure scenarios of the energy meter.

[0046] By deconstructing the multidimensional test parameter space, multiple continuous interference dimensions corresponding to the multivariate interference characteristics are obtained, as well as multiple sets of discrete operating condition settings corresponding to the multiple operating conditions and multiple port coupling topologies corresponding to the multiple injection paths.

[0047] For the continuous dimension of the decoupled interference, such as voltage 0-150%U nThe core operation of Latin hypercube sampling is performed at a frequency of 40Hz-1kHz. Specifically, based on the probability distribution characteristics of the parameters, the range of values ​​for each continuous dimension is evenly divided into equal-probability sub-intervals with a number equal to the preset number of sampling points. Within each sub-interval, a parameter value is independently and randomly selected as the layered interval sampling value.

[0048] This embodiment ensures that continuous parameters such as voltage and frequency cover the entire value range without omission in a statistical sense through a forced uniform distribution mechanism, completely avoiding parameter clustering or blank areas that may be caused by traditional random sampling, and providing a mathematically complete continuous parameter basis for the initial excitation strategy set.

[0049] For discretized operating condition settings, such as load status: balanced / 10% unbalanced / 30% unbalanced; power factor: 0.5L / 1.0 / 0.8C, perform full combination enumeration. Specifically, generate all possible operating condition combination points through Cartesian product operation. Each combination point represents a specific operating state configuration, such as "30% unbalanced load + 0.8 capacitive power factor".

[0050] This implementation ensures that the response characteristics of three-phase energy meters are fully detected under different load characteristics and power factor conditions through exhaustive combination, eliminating failure detection loopholes caused by operating condition jumps, and giving the initial excitation strategy ergonomic testing capability across the operating condition dimension.

[0051] In terms of injection path topology, for port coupling configurations such as single-phase injection [L1,PE], line-to-line injection [L2,L3], and three-phase injection [L1,L2,L3,N], a complete enumeration of legal combinations is performed to generate all possible injection path combinations.

[0052] Each combination clearly defines the injection port, return path, and coupling phase relationship of the interference signal, such as the differential mode interference path of "L2 injection - L3 return". This forms a complete set of topological scenarios covering various conducted and coupled electromagnetic interferences that a three-phase energy meter may be subjected to in the actual power grid environment, providing a topologically unbiased test basis for the initial strategy.

[0053] For the multiple parameter stratified interval sampling values, multiple operating condition combination points, and multiple injection path combinations generated in the aforementioned steps, a cyclic mapping is first performed to construct a strategy prototype. Specifically, the i-th continuous parameter value, the i-th operating condition point, and the i-th topology combination are bound to the i-th initial strategy in a fixed order, forming a complete parameter configuration covering all dimensions and ensuring that the original parameter set is completely traversed. Subsequently, multi-dimensional independent random permutation optimization is performed. The permutation optimization here is essentially the independent rearrangement of the three types of parameter sequences. Specifically, the continuous parameter sequence is randomly recombined, the operating condition sequence is randomly rearranged, and the topology sequence is randomly perturbed, completely breaking the potential pseudo-correlation between dimensions to avoid high-frequency noise parameters being forcibly associated with specific load conditions. After recombination, the final multiple initial excitation strategies are re-mapped in order.

[0054] The two-stage operation in this embodiment ensures the traversal of the parameter space through cyclic mapping and achieves strict statistical independence between dimensions through random permutation, thereby outputting a mathematically complete set of initial activation strategies. Each strategy satisfies the characteristics of uniform coverage, unbiased enumeration, and zero correlation, thus establishing a high-confidence test input foundation for subsequent sensitivity quantification.

[0055] Furthermore, the method step A100 provided by the present invention includes: A110: Configure the three-phase energy meter to the programmable immunity test platform, wherein the programmable immunity test platform has a safety isolation function, a parameter programmable function, and a multi-source synchronous triggering function; A120: During the test cycle in which the programmable immunity test platform applies the multiple initial excitation strategies to the three-phase energy meter using the parameter programmable function, a dynamic protection strategy is implemented through the safety isolation function, and the multi-source synchronous triggering function is used to coordinate the monitoring equipment array to perform multi-dimensional test data acquisition, thereby obtaining the multiple performance degradation vectors.

[0056] Furthermore, during the first test cycle when the programmable parameter control platform applies the first initial excitation strategy to the three-phase energy meter: A121: A dynamic protection strategy is executed in real time through the safety isolation function to continuously monitor the safety parameters of the electricity meter; A122: When the safety parameters of the electricity meter are at the safe operating threshold, the multi-source synchronous triggering function is used to coordinate the monitoring equipment array to perform multi-dimensional test data acquisition; A123: If the safety parameters of the electricity meter are detected to deviate from the safe operating threshold, the dynamic protection strategy is activated to record the first performance degradation vector, and the test is interrupted. The first performance degradation vector includes metering error offset, communication bit error rate, functional abnormality flag, and abnormality recovery time; A124: If the dynamic protection strategy is not activated within the first test cycle, the first performance degradation vector is recorded as an empty set.

[0057] Specifically, the three-phase energy meter under test is installed on the programmable immunity test platform and the electrical connection and communication configuration are completed. The programmable immunity test platform integrates three core functions. Among them, the safety isolation function protects the equipment safety in real time through hardware protection circuits and software monitoring strategies; the parameter programmable function supports millisecond-level dynamic adjustment of interference parameters such as voltage amplitude, frequency, and waveform characteristics; and the multi-source synchronous triggering function coordinates multiple monitoring devices to achieve nanosecond-level time synchronous acquisition.

[0058] It should be understood that since the methods for obtaining the multiple performance degradation vectors by executing the multiple initial excitation strategies through the programmable immunity test platform are consistent, this embodiment takes obtaining the first performance degradation vector under the first initial excitation strategy as an example to elaborate on the technical solution in detail.

[0059] During the first test cycle when the programmable parameter control platform applies the first initial excitation strategy to the three-phase energy meter, the safety isolation function immediately activates the dynamic protection strategy. The high-speed sensor continuously monitors the key safety parameters of the energy meter, including but not limited to physical quantities such as winding temperature rise rate, effective value of ground leakage current, and current harmonic distortion rate. The monitoring data is refreshed at a microsecond speed. The dynamic protection strategy assesses the risk level in real time according to the preset safety algorithm, providing a basis for safety assurance decisions for subsequent operations.

[0060] If the safety isolation function determines that the safety parameters of the electricity meter are always within the preset safe operating threshold range, the multi-source synchronous triggering function is triggered to send a global clock signal, coordinating the array of monitoring equipment such as the metering error analyzer and communication protocol analyzer connected to the test platform to synchronously start data acquisition.

[0061] Each device in the monitoring equipment array collects the metering error offset, communication message bit error rate, functional status flag bit, and abnormal recovery time of the energy meter under interference conditions according to preset indicators, forming a complete first performance degradation vector.

[0062] If the dynamic protection strategy detects any safety parameter deviating from the safe operating threshold during the test, such as leakage current exceeding the grounding protection limit or winding temperature rise rate exceeding the safety curve, the protection mechanism is immediately activated to record the currently collected performance data as the first performance degradation vector. The vector includes the metering error offset, communication bit error rate, functional abnormality flag and abnormality recovery time obtained before the interruption. At the same time, the test process is forcibly interrupted and interference output is cut off to avoid equipment damage.

[0063] If the dynamic protection strategy is not activated throughout the first test cycle, it indicates that the energy meter has not experienced any safety risks and its performance parameters have not degraded abnormally under this activation strategy. At this time, the first performance degradation vector recorded is an empty set. This empty set state indicates that the current combination of interference parameters has no observable impact on the device performance, providing a basis for zero value determination in subsequent sensitivity measurement.

[0064] Similarly, during the test cycle in which the programmable parameter control platform applies the multiple initial excitation strategies to the three-phase energy meter, a dynamic protection strategy is implemented through the safety isolation function, and the multi-source synchronous triggering function is used to coordinate the monitoring equipment array to perform multi-dimensional test data acquisition, thereby obtaining the multiple performance degradation vectors.

[0065] Furthermore, based on the failure threshold system, the interference sensitivity of the multiple performance degradation vectors is quantified to obtain multiple sensitivity evaluation values. The method step A200 provided by this invention includes: A210: Invoke the failure threshold system according to the device type of the three-phase energy meter, wherein the failure threshold system includes metering performance failure threshold, communication performance failure threshold, functional safety failure threshold, and abnormal recovery failure threshold; A210: Assign vector weight coefficients according to the device type of the three-phase energy meter; A220: Calculate the normalized degradation amount of the failure threshold system and multiple performance degradation vectors based on the vector weight coefficients, and generate the multiple sensitivity evaluation values, wherein when the performance degradation vector is an empty set, the corresponding sensitivity evaluation value is set to 0.

[0066] Specifically, based on the specific device type of the three-phase energy meter under test, the corresponding failure threshold system is called from the preset rule base. The failure threshold system includes four strictly defined failure judgment criteria. Among them, the metering performance failure threshold sets the core accuracy tolerance limit of the energy meter as a metering device, the communication performance failure threshold specifies the collapse boundary of data communication reliability, the functional safety failure threshold clarifies the trigger threshold of safety mechanisms such as equipment insulation protection, and the abnormal recovery failure threshold limits the maximum allowable response time of the self-recovery function. The failure threshold system provides a failure judgment standard that is strictly matched with the device type for subsequent quantification.

[0067] For the three-phase electricity meter types currently being tested, differentiated vector weight coefficients are assigned to four types of failure thresholds. The weight allocation follows the principle of prioritizing equipment functions. Specifically, metering performance, as the core function of the electricity meter, is given the highest weight; communication performance, due to its impact on data upload, is given the second highest weight; functional safety weight reflects the importance of the equipment's protection capabilities; and anomaly recovery weight characterizes fault tolerance. The total weight coefficient is a fixed value of 1, and the adjustments to the ratio reflect the differences in sensitivity of different equipment types to various failure types.

[0068] The measurement error offset, communication bit error rate, functional anomaly flag, and anomaly recovery time of each performance degradation vector are compared with the corresponding failure thresholds: measurement performance failure threshold, communication performance failure threshold, functional safety failure threshold, and anomaly recovery failure threshold, to calculate the relative degradation ratio of each performance indicator. Then, the degradation ratios of all dimensions are aggregated according to the weight coefficients to generate a sensitivity evaluation value.

[0069] It should be understood that for empty set vectors that do not trigger performance degradation during testing, i.e., when the device is completely unaffected, this embodiment directly sets their sensitivity evaluation value to zero in order to reduce unnecessary data processing and waste of computing resources.

[0070] This embodiment transforms multi-dimensional performance status into a single comparable index, achieving the technical effect of objectively quantifying the overall vulnerability of three-phase energy meters under specific interference, and providing unbiased and standardized input data for subsequent screening of highly sensitive candidate strategies.

[0071] Furthermore, after performing adaptive neighborhood mutation on the K high-sensitivity candidate strategies to obtain K sets of boundary detection strategies, the programmable immunity test platform is controlled to perform critical gradient tests on the three-phase energy meter to collect K sets of boundary performance vectors. The method step A400 provided by this invention includes: A410: Based on the numerical range of the K sensitivity evaluation values ​​of the K high-sensitivity candidate strategies, set K variation amplitudes in a hierarchical manner; A420: Using the K variation amplitudes as the variation radius, perform independent spatial variation of adjacent parameters of the K high-sensitivity candidate strategies by sampling values ​​in multiple parameter hierarchical intervals, multiple operating condition combination points, and multiple injection path combinations, and output K sets of initial detection strategies; A430: Perform strategy deduplication and screening on the multiple initial excitation strategies and the K sets of initial detection strategies to obtain the K sets of boundary detection strategies; A440: Use the K sets of boundary detection strategies to control the programmable disturbance rejection test platform to perform critical gradient tests on the three-phase energy meter, so as to collect the K sets of boundary performance vectors.

[0072] Based on the numerical range of the sensitivity evaluation values ​​corresponding to K high-sensitivity candidate strategies, a hierarchical processing method is adopted. Different variation amplitudes are set for different sensitivity levels, resulting in K variation amplitudes corresponding to the K high-sensitivity candidate strategies. Specifically, for candidate strategies with high sensitivity evaluation values, since they are close to the failure boundary, a smaller variation amplitude is used for fine-tuning to avoid exceeding the failure region; while for candidate strategies with low sensitivity evaluation values, since they are far from the failure boundary, a larger variation amplitude is used to perform exploratory mutations to quickly approach the potential sensitive area. The hierarchical mechanism of this embodiment realizes intelligent control of mutation operations, ensuring a balance between in-depth mining of high-sensitivity areas and efficient exploration of low-sensitivity areas.

[0073] Using the K variation amplitudes as parameter perturbation radii, independent variation operations are performed on three sub-dimensions of the multidimensional test parameter space. Specifically, in the continuous parameter dimensions such as voltage amplitude and frequency, random fine-tuning is performed within the neighborhood of the current value with the variation amplitude as the radius; in the discrete dimensions such as load condition and power factor, a new condition is randomly selected from the adjacent enumerated points of the current condition combination point; and in the injection path topology dimension, a new path is randomly switched from the connection mode similar to the current topology structure.

[0074] This embodiment generates new parameter combinations for each highly sensitive candidate strategy through three-dimensional independent mutation, and outputs K initial detection strategies covering the key regions of the multi-dimensional test parameter space.

[0075] By comparing the multiple initial excitation strategies and K sets of initial detection strategies obtained from history, strategy deduplication is performed to eliminate duplicate strategies with the same parameters as historical strategies, and to filter out strategy combinations that have different parameter values ​​but equivalent actual functions. Strategies with innovative parameters and clear testing significance are retained to form the K sets of boundary detection strategies for the final boundary detection.

[0076] The same method is used to obtain the performance degradation vector by driving the programmable immunity test platform to perform critical gradient tests on the three-phase energy meter using the K-group boundary detection strategy, so as to collect the K-group boundary performance vectors and provide accurate boundary state input for subsequent gradient-driven iterations.

[0077] Furthermore, based on the failure trend measurement of the K sets of boundary performance vectors relative to the failure threshold system, gradient-driven boundary convergence iteration is performed in the multidimensional test parameter space until a critical failure boundary test set is output. The method step A500 includes: A510: Calculate the K-group sensitivity evaluation values ​​of the K-group boundary performance vectors relative to the failure threshold system; A520: Compare the K-group sensitivity evaluation values ​​with a preset sensitivity threshold to select K-group updated high-sensitivity strategies from the K-group boundary performance vectors; A530: Perform spatial interpolation of the K-group updated high-sensitivity strategies in the multi-dimensional test parameter space to construct a gradient field space; A540: Quantize the K-group gradient vectors according to the K-group updated evaluation values ​​corresponding to the K-group updated high-sensitivity strategies, and define the negative gradient direction in the gradient field space; A550: Perform gradient-driven boundary convergence iteration along the negative gradient direction in the gradient field space until the critical failure boundary test set is output.

[0078] Furthermore, in the gradient field space, gradient-driven boundary convergence iteration is performed along the negative gradient direction until the critical failure boundary test set is output. The method step A550 includes: A551: setting K iteration amplitudes according to the K sets of update evaluation values; A552: starting from the first update high-sensitivity strategy, in the gradient field space along the negative gradient direction, performing boundary convergence iterations of the first iteration amplitude, and outputting the first critical boundary point, wherein, at the first critical boundary point, the change in sensitivity evaluation value is <0.005 and the gradient magnitude is <0.05 in W consecutive iterations; A553: similarly, starting from the K sets of update high-sensitivity strategies, in the gradient field space along the negative gradient direction, performing boundary convergence iterations of the K iteration amplitudes, and outputting K sets of critical boundary points; A554: performing multimodal space aggregation on the K sets of critical boundary points in the gradient field space, and outputting the critical failure boundary test set.

[0079] Using the refined sensitivity evaluation value calculation method in step A200, the K sets of boundary performance vectors are calculated as K sets of sensitivity evaluation values ​​relative to the failure threshold system. The sensitivity evaluation values ​​obtained in this step accurately quantify the degree of failure approach of the three-phase energy meter under the boundary state, and focus more on the vulnerability characteristics of the critical region than the initial sensitivity calculation, providing high-precision input for gradient field construction.

[0080] The screening process inherits the threshold priority principle of A300. It uses a preset sensitivity threshold to compare the K groups of sensitivity evaluation values ​​to select K groups of updated high-sensitivity strategies from the K groups of boundary performance vectors. It should be understood that the K groups of updated high-sensitivity strategies selected here represent the parameter combinations that are closest to the actual failure boundary. This screening mechanism ensures that subsequent gradient calculations are only based on high-threat parameter points, effectively filtering out pseudo-boundary points caused by measurement noise or random fluctuations, and improving the purity and reliability of gradient field modeling.

[0081] Using the selected K groups of highly sensitive update strategies as spatial anchors, local fine interpolation operations are performed within the complete framework of the multidimensional test parameter space to construct a gradient field space whose physical range is precisely shrunk to the key region. The gradient field space achieves focused coverage through a triple mechanism. First, the boundary range of continuous dimensions is dynamically defined based on the spatial distribution of high-sensitivity strategy points. For example, the original range of the voltage dimension is 0-150% of the rated value. If all high-sensitivity points are concentrated in the 112% to 138% range, the voltage range of the gradient field space is shrunk to this sub-interval. Second, only the actual combination of operating conditions and topology paths involved in the high-sensitivity strategy are retained in the discrete dimension. For example, if the original operating conditions include 10% / 30% / 50% imbalance levels, and the high-sensitivity strategy only triggers the 30% imbalance condition, the gradient field space locks the configuration of that condition. Finally, super-resolution interpolation is performed in the shrunk continuous parameter sub-interval to generate dense grid points with a preset precision step size. For example, a millimeter-level grid is constructed in the voltage dimension with a step size of 0.1V. At the same time, the discrete dimension is fixed as a specific configuration combination of the high-sensitivity strategy. The sensitivity values ​​of the grid points are calculated through bilinear interpolation to form a high-resolution response surface. The resulting gradient field space strictly inherits the dimensional structure and physical meaning of the original parameter space, but by shrinking the range and refining the grid, computational resources are focused on high-threat areas, achieving a leap in the accuracy of failure boundary detection.

[0082] Based on the discrete operating condition configuration and topology path locked in the gradient field space, gradient vector quantization is performed focusing on two continuous dimensions: voltage and frequency. Specifically, for each updated high-sensitivity strategy point, the partial derivative of the sensitivity evaluation value with respect to voltage is calculated using the central difference method in the voltage dimension. That is, a small positive or negative perturbation is applied near the current voltage value and the rate of change of sensitivity obtained by interpolation in the gradient field space is observed. The same operation is performed simultaneously in the frequency dimension to obtain the partial derivative with respect to frequency. The two partial derivatives are combined to form a two-dimensional gradient vector as a set of gradient vectors, which accurately represents the direction of the fastest increase in sensitivity in the voltage-frequency plane.

[0083] By analogy, the K-group gradient vectors are quantized based on the K-group update evaluation values ​​corresponding to the K-group update high-sensitivity strategies. Then, based on the K-group gradient vectors, the negative gradient direction is defined as the sensitivity reduction path, that is, the parameter adjustment direction that makes the energy meter performance tend to be safe under the current locked operating conditions and topology. This embodiment transforms the abstract failure trend into a directional navigation signal in the parameter space through mathematical analysis, providing an executable optimization path for boundary convergence iteration.

[0084] Using the method of setting the same variation range in step A410, K iteration ranges are set according to the K groups of updated evaluation values.

[0085] Starting from the current parameter coordinates of any updated high-sensitivity strategy, a boundary convergence iteration operation is performed along the negative gradient direction in the gradient field space. In each iteration, the voltage and frequency coordinate values ​​are updated according to the current gradient vector. The step distance is adjusted according to the preset first iteration amplitude in the continuous parameter dimension, while keeping the discrete working condition and topology path dimensions locked. The change in sensitivity evaluation value and gradient vector magnitude are continuously monitored during the iteration process. When the sensitivity fluctuation amplitude is less than 0.005 and the gradient magnitude is less than 0.05 in W consecutive iterations, it is determined that the current parameter point has converged to the critical failure boundary. This point is output as the first critical boundary point. The voltage and frequency coordinate values ​​of the first critical boundary point are the accurate failure threshold under the current working condition topology combination.

[0086] In other words, under the operating topology combination of the first updated high-sensitivity strategy, the three-phase energy meter will reach the absolute limit threshold of the device's anti-interference capability when subjected to voltage and frequency interference at the first critical boundary point. If the interference intensity increases by any minute amount at this point, it will immediately trigger irreversible permanent functional failure or hardware damage.

[0087] By analogy, starting with the K sets of updated high-sensitivity strategies, the boundary convergence iterations of the K iteration magnitudes are performed along the negative gradient direction in the gradient field space, outputting K sets of critical boundary points. Each set of points precisely corresponds to the failure boundary parameter coordinates under a specific working condition and topology combination, realizing boundary detection with full coverage of the multi-dimensional parameter space.

[0088] A multimodal spatial aggregation operation is performed on K groups of critical boundary points. In the voltage-frequency continuous dimension, a smooth family of critical failure boundary curves is generated through a surface fitting algorithm. Each curve is labeled with the corresponding operating condition configuration and topology path information. In the discrete dimension, a structured mapping relationship is established, and boundary points with the same operating condition topology combination are clustered into machine-readable data units. Each unit contains critical voltage value, critical frequency value, load condition code, injection path code, and failure mode type. Finally, a critical failure boundary test set that conforms to the automated testing specifications is output.

[0089] In the production and factory testing of three-phase energy meters, the critical failure boundary test set is used to conduct anti-interference condition testing. If the energy meter does not trigger permanent failure when the programmable anti-interference test platform applies the critical interference of the critical failure boundary test set, it is judged to be a qualified product.

[0090] In operational monitoring scenarios, the critical failure boundary test set can serve as the core parameter of the intelligent early warning system. When the actual disturbance of the power grid approaches the critical threshold of the load rate-topology combination recorded in the critical failure boundary test set, the load reduction protection or the switching of the backup path is automatically triggered, thereby implementing proactive protection before equipment damage.

[0091] The critical failure boundary test set precisely quantifies the anti-interference limit threshold of three-phase energy meters under various operating conditions and topology combinations through mathematically converged critical boundary points. The recorded voltage, frequency and other parameter coordinates directly correspond to the strict inflection point of the equipment from recoverable abnormality to irreversible damage. It provides a benchmark for calculating the tolerance of protection circuits for product design, establishes automated judgment standards for factory testing, configures real-time early warning thresholds for operation monitoring, and generates legal test basis for type certification. Finally, it transforms the abstract anti-interference performance into executable, verifiable and traceable engineering control parameters in the form of structured data, realizing fine-grained and precise control of anti-interference management throughout the entire life cycle of the energy meter.

[0092] Example 2, based on the same inventive concept as the anti-interference test method for three-phase energy meters in the foregoing examples, such as... Figure 2 As shown, this application provides an anti-interference testing system for three-phase energy meters, wherein the system includes: The anti-interference control module 10 is used to synchronously trigger the monitoring equipment array to perform multi-dimensional test data acquisition when the control programmable anti-interference test platform applies multiple initial excitation strategies to the three-phase energy meter, and obtain multiple performance degradation vectors, wherein the multiple initial excitation strategies are derived from the multi-dimensional test parameter space.

[0093] The anti-interference performance quantification module 20 is used to quantify the interference sensitivity of the multiple performance degradation vectors based on the failure threshold system, and obtain multiple sensitivity evaluation values.

[0094] The strategy screening module 30 is used to screen and extract K high-sensitivity candidate strategies from the multiple initial excitation strategies based on the descending order of the multiple sensitivity evaluation values.

[0095] The strategy mutation module 40 is used to perform adaptive neighborhood mutation on the K highly sensitive candidate strategies to obtain K sets of boundary detection strategies, and then control the programmable disturbance rejection test platform to perform critical gradient tests on the three-phase energy meter to collect K sets of boundary performance vectors.

[0096] The critical strategy iteration module 50 is used to perform gradient-driven boundary convergence iteration in the multidimensional test parameter space based on the failure trend measurement of the K sets of boundary performance vectors relative to the failure threshold system, until the critical failure boundary test set is output.

[0097] Furthermore, the critical strategy iteration module 50 is used to perform the following operation steps: Calculate the K-group sensitivity evaluation values ​​of the K-group boundary performance vectors relative to the failure threshold system; compare the K-group sensitivity evaluation values ​​with a preset sensitivity threshold to select K-group updated high-sensitivity strategies from the K-group boundary performance vectors; perform spatial interpolation of the K-group updated high-sensitivity strategies in the multi-dimensional test parameter space to construct a gradient field space; quantize the K-group gradient vectors according to the K-group updated evaluation values ​​corresponding to the K-group updated high-sensitivity strategies, and define the negative gradient direction in the gradient field space; perform gradient-driven boundary convergence iteration along the negative gradient direction in the gradient field space until the critical failure boundary test set is output.

[0098] Furthermore, the critical strategy iteration module 50 is used to perform the following operation steps: K iteration amplitudes are set according to the K sets of update evaluation values; starting from the first update high-sensitivity strategy, boundary convergence iterations of the first iteration amplitude are performed along the negative gradient direction in the gradient field space, and the first critical boundary point is output, wherein the change in sensitivity evaluation value and the gradient magnitude are both less than 0.005 and less than 0.05 in W consecutive iterations at the first critical boundary point; similarly, starting from the K sets of update high-sensitivity strategies, boundary convergence iterations of the K iteration amplitudes are performed along the negative gradient direction in the gradient field space, and K sets of critical boundary points are output; multimodal space aggregation is performed on the K sets of critical boundary points in the gradient field space, and the critical failure boundary test set is output.

[0099] Furthermore, the anti-interference control module 10 is used to perform the following operation steps: The failure response characteristics of the three-phase energy meter are obtained interactively, wherein the failure response characteristics cover multiple interference characteristics, multiple operating conditions and multiple injection paths; a multidimensional test parameter space is constructed based on the failure response characteristics; and multiple initial excitation strategies are generated in the multidimensional test parameter space through Latin hypercube sampling.

[0100] Furthermore, the anti-interference control module 10 is used to perform the following operation steps: The multidimensional test parameter space is deconstructed to obtain multiple continuous interference dimensions corresponding to the multivariate interference characteristics, multiple sets of discrete operating condition settings corresponding to the multiple operating conditions, and multiple port coupling topologies corresponding to the multiple injection paths. The multiple continuous interference dimensions are sampled by equal-probability interval division to obtain multiple parameter hierarchical interval sampling values. The multiple sets of discrete operating condition settings are combined and enumerated to obtain multiple operating condition combination points. The multiple port coupling topologies are combined and enumerated to obtain multiple injection path combinations. After sequentially cyclically mapping the multiple parameter hierarchical interval sampling values, multiple operating condition combination points, and multiple injection path combinations, parameter correlation is eliminated based on multidimensional independent random permutation, and the multiple initial excitation strategies are output.

[0101] Furthermore, the strategy mutation module 40 is used to perform the following operational steps: Based on the numerical range of the K sensitivity evaluation values ​​of the K high-sensitivity candidate strategies, K variation amplitudes are set in a hierarchical manner. Using the K variation amplitudes as variation radii, adjacent parameter spaces of the K high-sensitivity candidate strategies are independently mutated by sampling values ​​in multiple parameter hierarchical intervals, multiple operating condition combinations, and multiple injection path combinations, resulting in K sets of initial detection strategies. The multiple initial excitation strategies and the K sets of initial detection strategies are deduplicated to obtain the K sets of boundary detection strategies. The K sets of boundary detection strategies are used to control the programmable disturbance rejection test platform to perform critical gradient tests on the three-phase energy meter to collect the K sets of boundary performance vectors.

[0102] Furthermore, the anti-interference control module 10 is used to perform the following operation steps: The three-phase energy meter is configured on the programmable immunity test platform, which has a safety isolation function, a parameter programmable function, and a multi-source synchronous triggering function. During the test cycle in which the programmable immunity test platform applies the multiple initial excitation strategies to the three-phase energy meter using the parameter programmable function, a dynamic protection strategy is implemented through the safety isolation function, and the multi-source synchronous triggering function coordinates the monitoring equipment array to perform multi-dimensional test data acquisition to obtain the multiple performance degradation vectors.

[0103] Furthermore, the anti-interference control module 10 is used to perform the following operation steps: During the first test cycle in which the programmable parameter control platform applies a first initial excitation strategy to the three-phase energy meter: a dynamic protection strategy is executed in real time through the safety isolation function to continuously monitor the energy meter's safety parameters; when the energy meter's safety parameters are at the safe operating threshold, the multi-source synchronous triggering function coordinates the monitoring equipment array to perform multi-dimensional test data acquisition; if the energy meter's safety parameters are detected to deviate from the safe operating threshold, the dynamic protection strategy is activated to record a first performance degradation vector, and the test is interrupted, wherein the first performance degradation vector includes metering error offset, communication bit error rate, functional abnormality flag, and abnormality recovery time; if the dynamic protection strategy is not activated during the first test cycle, the first performance degradation vector is recorded as an empty set.

[0104] Furthermore, the anti-interference performance quantification module 20 is used to perform the following operation steps: The failure threshold system is invoked according to the device type of the three-phase energy meter, wherein the failure threshold system includes metering performance failure threshold, communication performance failure threshold, functional safety failure threshold, and abnormal recovery failure threshold; vector weight coefficients are assigned according to the device type of the three-phase energy meter; the normalized degradation amount of the failure threshold system and multiple performance degradation vectors is calculated based on the vector weight coefficients to generate the multiple sensitivity evaluation values, wherein when the performance degradation vector is an empty set, the corresponding sensitivity evaluation value is set to 0.

[0105] In summary, any of the methods or steps described above can be stored as computer instructions or programs in various types of computer memory, and the computer instructions or programs can be recognized by various types of computer processors to implement any of the above methods or steps.

[0106] Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by those skilled in the art without departing from the principle of the present invention shall fall within the patent protection scope of the present invention.

Claims

1. A method for immunity test of a three-phase electric energy meter, characterized in that, The method includes: When the programmable immunity test platform applies multiple initial excitation strategies to the three-phase energy meter, the monitoring equipment array is simultaneously triggered to perform multi-dimensional test data acquisition to obtain multiple performance degradation vectors. The multiple initial excitation strategies are derived from the multi-dimensional test parameter space. Based on the failure threshold system, the interference sensitivity of the multiple performance degradation vectors is quantified to obtain multiple sensitivity evaluation values; Based on the descending order of the multiple sensitivity evaluation values, K high-sensitivity candidate strategies are selected and extracted from the multiple initial activation strategies; After performing adaptive neighborhood mutation on the K high-sensitivity candidate strategies to obtain K sets of boundary detection strategies, the programmable disturbance rejection test platform is controlled to perform critical gradient tests on the three-phase energy meter to collect K sets of boundary performance vectors. Based on the failure trend measurement of the K sets of boundary performance vectors relative to the failure threshold system, gradient-driven boundary convergence iteration is performed in the multidimensional test parameter space until the critical failure boundary test set is output. The method includes, based on the failure trend measurement of the K sets of boundary performance vectors relative to the failure threshold system, performing gradient-driven boundary convergence iteration in the multidimensional test parameter space until a critical failure boundary test set is output. Calculate the K sets of sensitivity evaluation values ​​of the boundary performance vectors relative to the failure threshold system; A preset sensitivity threshold is used to compare the K groups of sensitivity evaluation values, so as to select K groups of updated high-sensitivity strategies from the K groups of boundary performance vectors; Spatial interpolation of the K sets of high-sensitivity update strategies is performed in the multidimensional test parameter space to construct a gradient field space; The K-group gradient vectors are quantized based on the K-group update evaluation values ​​corresponding to the K-group update high-sensitivity strategies, and the negative gradient direction is defined in the gradient field space. Gradient-driven boundary convergence iterations are performed in the gradient field space along the negative gradient direction until the critical failure boundary test set is output. The method includes performing gradient-driven boundary convergence iterations along the negative gradient direction in the gradient field space until the critical failure boundary test set is output. K iteration ranges are set based on the K sets of updated evaluation values; Starting with the first updated high-sensitivity strategy, a boundary convergence iteration of the first iteration magnitude is performed along the negative gradient direction in the gradient field space, and a first critical boundary point is output, wherein the change in sensitivity evaluation value and the gradient magnitude are both less than 0.005 in W consecutive iterations at the first critical boundary point. Similarly, starting with the K sets of high-sensitivity update strategies, the boundary convergence iterations of the K iteration magnitudes are performed along the negative gradient direction in the gradient field space, and the K sets of critical boundary points are output. Multimodal spatial aggregation is performed on the K sets of critical boundary points in the gradient field space to output the critical failure boundary test set.

2. The immunity test method for a three-phase electric energy meter according to claim 1, wherein, The method further includes: The failure response characteristics of the three-phase energy meter are obtained interactively, wherein the failure response characteristics cover multiple interference characteristics, multiple operating conditions and multiple injection paths; Based on the failure response characteristics, a multidimensional test parameter space is constructed; The multiple initial excitation strategies are generated in the multidimensional test parameter space by Latin hypercube sampling.

3. The immunity test method for a three-phase electric energy meter according to claim 2, wherein The method for generating the multiple initial excitation strategies in the multidimensional test parameter space via Latin hypercube sampling includes: By deconstructing the multidimensional test parameter space, multiple continuous interference dimensions corresponding to the multivariate interference characteristics are obtained, as well as multiple sets of discrete operating condition settings corresponding to the multiple operating conditions and multiple port coupling topologies corresponding to the multiple injection paths. The multiple continuous dimensions of interference are divided into equal probability intervals for sampling to obtain multiple parameter hierarchical interval sampling values; The multiple sets of discrete working conditions are combined and enumerated to obtain multiple working condition combination points; Multiple injection path combinations are obtained by combining and enumerating the various port coupling topologies; After performing sequential cyclic mapping on the sampled values ​​of the multiple parameter hierarchical intervals, multiple working condition combination points, and multiple injection path combinations, the parameter correlation is eliminated based on multidimensional independent random permutation, and the multiple initial excitation strategies are output.

4. The anti-interference test method for three-phase energy meters as described in claim 3, characterized in that, After performing adaptive neighborhood mutation on the K high-sensitivity candidate strategies to obtain K sets of boundary detection strategies, the programmable disturbance rejection test platform is controlled to perform critical gradient tests on the three-phase energy meter to collect K sets of boundary performance vectors. The method includes: K variation ranges are set according to the numerical intervals of the K sensitivity evaluation values ​​of the K high-sensitivity candidate strategies. Using the K variation amplitudes as the variation radius, the adjacent parameter spaces of the K high-sensitivity candidate strategies are independently mutated by sampling values ​​in multiple parameter stratification intervals, multiple working condition combination points, and multiple injection path combinations, and K initial detection strategies are output. The multiple initial excitation strategies and the K sets of initial detection strategies are deduplicated to obtain the K sets of boundary detection strategies. The K-group boundary detection strategy is used to control the programmable immunity test platform to perform critical gradient tests on the three-phase energy meter in order to collect the K-group boundary performance vectors.

5. The anti-interference test method for three-phase energy meters as described in claim 1, characterized in that, The method further includes: The three-phase energy meter is configured onto the programmable immunity test platform, wherein the programmable immunity test platform has a safety isolation function, a parameter programmable function, and a multi-source synchronous triggering function; During the test cycle in which the programmable parameter control platform applies the multiple initial excitation strategies to the three-phase energy meter, a dynamic protection strategy is implemented through the safety isolation function, and the multi-source synchronous triggering function is used to coordinate the monitoring equipment array to perform multi-dimensional test data acquisition, thereby obtaining the multiple performance degradation vectors.

6. The anti-interference test method for three-phase energy meters as described in claim 5, characterized in that, During the first test cycle when the programmable parameter function is used to control the programmable immunity test platform to apply the first initial excitation strategy to the three-phase energy meter: Dynamic protection strategies are implemented in real time through the safety isolation function, and the safety parameters of the electricity meter are continuously monitored. When the safety parameters of the electricity meter are at the safe operating threshold, the multi-source synchronous triggering function is used to coordinate the monitoring equipment array to perform multi-dimensional test data acquisition. If the safety parameters of the electricity meter are detected to deviate from the safe operation threshold, the dynamic protection strategy is activated to record the first performance degradation vector and the test is interrupted. The first performance degradation vector includes metering error offset, communication bit error rate, functional abnormality flag and abnormality recovery time. If the dynamic protection strategy is not activated during the first test period, the first performance degradation vector is recorded as an empty set.

7. The anti-interference test method for three-phase energy meters as described in claim 6, characterized in that, The interference sensitivity of the multiple performance degradation vectors is quantified based on a failure threshold system to obtain multiple sensitivity evaluation values. The method includes: The failure threshold system is invoked according to the device type of the three-phase energy meter, wherein the failure threshold system includes metering performance failure threshold, communication performance failure threshold, functional safety failure threshold, and abnormal recovery failure threshold; Assign vector weight coefficients according to the equipment type of the three-phase energy meter; Based on the vector weight coefficients, the normalized degradation amount of the failure threshold system and multiple performance degradation vectors is calculated to generate the multiple sensitivity evaluation values. When the performance degradation vector is an empty set, the corresponding sensitivity evaluation value is set to 0.

8. An anti-interference testing system for three-phase energy meters, characterized in that, The steps for implementing the method according to any one of claims 1 to 7 include: An anti-interference control module is used to synchronously trigger the monitoring equipment array to perform multi-dimensional test data acquisition when the control programmable anti-interference test platform applies multiple initial excitation strategies to the three-phase energy meter, thereby obtaining multiple performance degradation vectors. The multiple initial excitation strategies are derived from the multi-dimensional test parameter space. The anti-interference performance quantification module is used to quantify the interference sensitivity of the multiple performance degradation vectors based on the failure threshold system, and obtain multiple sensitivity evaluation values. The strategy screening module is used to select and extract K high-sensitivity candidate strategies from the multiple initial activation strategies based on the descending order of the multiple sensitivity evaluation values. The strategy mutation module is used to perform adaptive neighborhood mutation on the K high-sensitivity candidate strategies to obtain K sets of boundary detection strategies, and then control the programmable disturbance rejection test platform to perform critical gradient test on the three-phase energy meter to collect K sets of boundary performance vectors. The critical strategy iteration module is used to perform gradient-driven boundary convergence iteration in the multidimensional test parameter space based on the failure trend measurement of the K sets of boundary performance vectors relative to the failure threshold system, until the critical failure boundary test set is output.

Images