A smart meter performance testing method and system

Abstract

The application provides a kind of intelligent electric meter performance test method and system, it is related to data processing technical field, method includes: constructing orthogonal pseudo-random dynamic test voltage signal and current signal;Orthogonal pseudo-random dynamic test voltage signal and current signal are input to the intelligent electric meter to be detected in testing, and dynamic electric energy error signal is counted;Whether dynamic electric energy error signal is lower than preset electric energy error is judged;If yes, determine that the intelligent electric meter to be detected is qualified;Otherwise, determine that the performance of the intelligent electric meter to be detected is questionable, enter next step and carry out fine test;Through deep belief network, signal characteristics in dynamic electric energy error signal are extracted;Signal characteristics extracted by deep belief network are input to multilayer perception machine;Through multilayer perception machine, the performance of the intelligent electric meter to be detected is tested.The application can make that test result can truly reflect the measurement performance of electric meter in real environment, improve the quality of intelligent electric meter factory.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for testing the performance of smart meters. Background Technology

[0002] Smart meters undergo rigorous calibration testing before leaving the factory to ensure their electricity metering accuracy meets national or industry standards and that measurement errors under various typical operating conditions remain within a controllable range. To simulate voltage and current fluctuations and load changes in real-world environments, various test signals under different operating conditions are applied during the test to evaluate the smart meter's metering accuracy, response speed, and anti-interference capabilities. After calibration, the test data is statistically analyzed, and a pass / fail report is issued to ensure that each meter possesses good long-term stability and reliability, thereby guaranteeing fair metering for users, electricity safety, and refined management of the power grid.

[0003] Traditional testing methods often employ a standard power supply to output steady-state sinusoidal voltage and current signals, then superimpose a certain step or ramp disturbance to simulate load changes. These signals are input to the meter under test (DUT), and the dynamic energy error value is obtained by comparing it with a standard energy meter. This error is then used to determine whether it falls within the allowable range of national standards, serving as a basis for factory or metering performance assessments.

[0004] However, real-world electricity usage scenarios do not involve steady-state sinusoidal voltage and current signals. Traditional solutions struggle to accurately simulate real-world load asymmetry, random fluctuations, and periodic jitter, are insensitive to dynamic errors, easily overlook potential problems, and fail to accurately reflect the meter's measurement performance in real-world environments. Furthermore, when the performance of a smart meter is questionable after testing, it is difficult to analyze specific faults, thus failing to provide effective guidance for subsequent smart meter production. Summary of the Invention

[0005] To address the challenges of real-world electricity usage scenarios where voltage and current signals are not steady-state sinusoidal, traditional methods struggle to accurately simulate real-world load asymmetry, random fluctuations, and periodic jitter. They are insensitive to dynamic errors, easily overlooking potential problems and failing to accurately reflect the meter's performance in real-world environments. Furthermore, when performance of a smart meter is questionable after testing, it is difficult to analyze specific faults, hindering effective guidance for subsequent smart meter production. This invention provides a smart meter performance testing method and system.

[0006] The technical solutions provided by the embodiments of the present invention are as follows:

[0007] First aspect:

[0008] This invention provides a method for testing the performance of a smart meter, comprising:

[0009] S1: Construct orthogonal pseudo-random dynamic test voltage and current signals;

[0010] S2: Input the orthogonal pseudo-random dynamic test voltage signal and current signal into the smart meter under test for testing, and statistically calculate the dynamic energy error signal;

[0011] S3: Determine whether the dynamic power error signal is lower than the preset power error; if yes, determine that the smart meter under test is qualified; otherwise, determine that the performance of the smart meter under test is questionable, and proceed to the next step for fine testing;

[0012] S4: Extract signal features from the dynamic power error signal using a deep belief network;

[0013] S5: Input the signal features extracted by the deep belief network into the multilayer perceptron;

[0014] S6: The performance of the smart meter under test is performed using the multilayer sensor.

[0015] The second aspect:

[0016] This invention provides a smart meter performance testing system, comprising:

[0017] processor;

[0018] A memory storing computer-readable instructions, which, when executed by the processor, implement the smart meter performance testing method as described in the first aspect.

[0019] Third aspect:

[0020] The present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the smart meter performance testing method as described in the first aspect.

[0021] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0022] (1) In this embodiment of the invention, orthogonal pseudo-random dynamic test voltage and current signals are constructed to accurately simulate real-life situations such as load asymmetry, random jumps, and periodic jitter. Orthogonal pseudo-random dynamic test voltage and current signals are used to test the dynamic error of the smart meter, thereby improving the smart meter's sensitivity to dynamic error and enabling the test results to truly reflect the meter's measurement performance in the real environment, thus improving the smart meter's factory quality.

[0023] (2) In this embodiment of the invention, the performance test of the smart meter is carried out by deep belief network and multilayer perceptron, which can accurately diagnose the specific fault type of the smart meter, provide effective guidance for subsequent smart meter production, and improve the production efficiency of smart meters. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a flowchart illustrating a smart meter performance testing method provided in an embodiment of the present invention.

[0026] Figure 2 This is a schematic diagram illustrating the formation of a dynamic power error signal, provided in an embodiment of the present invention.

[0027] Figure 3 This is a schematic diagram of the structure of a smart meter performance testing model provided in an embodiment of the present invention.

[0028] Figure 4 This is a schematic diagram of a smart meter performance testing system provided in an embodiment of the present invention. Detailed Implementation

[0029] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0030] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0031] Reference manual attached Figure 1 The diagram shows a flowchart of a smart meter performance testing method provided by an embodiment of the present invention.

[0032] This invention provides a method for testing the performance of smart meters. This method can be implemented using a smart meter performance testing device, which can be a terminal or a server. The processing flow of the smart meter performance testing method may include the following steps:

[0033] Reference manual attached Figure 2 The diagram illustrates the formation of a dynamic power error signal according to an embodiment of the present invention.

[0034] S1: Construct orthogonal pseudo-random dynamic test power signals.

[0035] In one possible implementation, S1 specifically includes:

[0036] S101: Constructing Distorted Steady-State Standard Test Voltage and Current Signals:

[0037]

[0038]

[0039] Among them, u s This represents the distorted steady-state standard test voltage signal, where t represents time, and k = a, b, and c represent phases A, B, and C, respectively. U represents the distorted steady-state standard test voltage signal at time t in phase k. k L represents the fundamental amplitude of the k-th phase voltage signal, n1 represents the harmonic order of the voltage signal, and L represents the fundamental amplitude of the k-th phase voltage signal. k This represents the total harmonic order of the k-th phase voltage signal. A represents the relative amplitude of the n1-th order voltage harmonic component with respect to the fundamental frequency. k1 =1, sin represents the sine function, and ω represents the angular frequency. i represents the phase angle of the n1st order voltage harmonic component. s This represents a distorted steady-state standard test current signal. I represents the distorted steady-state standard test current signal at time t in phase k. k Let nk represent the fundamental amplitude of the k-th phase current signal, and n2 represent the harmonic order of the current signal. k This represents the total harmonic order of the k-th phase current signal. B represents the relative amplitude of the n2nd order current harmonic component with respect to the fundamental frequency. k1 =1, This represents the relative amplitude of the n2nd order current harmonic component relative to the fundamental frequency.

[0040] It should be noted that the distorted steady-state standard test voltage and current signals are the basic input signals used to simulate complex non-ideal electrical energy signals in actual power consumption environments. They are usually composed of multiple harmonics with different amplitudes and phases superimposed to reflect waveform distortion phenomena commonly found in real power grids.

[0041] S102: The distorted steady-state standard test voltage and current signals are truncated using multiple rectangular window functions to obtain the truncated voltage and current signals for each base period, and then combined to form the distorted steady-state test truncated voltage and current signals.

[0042]

[0043]

[0044]

[0045]

[0046]

[0047] Among them, u c This indicates the cutoff voltage signal for the distortion steady-state test. This represents the distortion steady-state test cutoff voltage signal at time t in phase k. This represents the distortion steady-state test cutoff voltage signal at time t in the k-th phase under the n-th base period. Represents the rectangular window function, i c This indicates that the distorted steady-state test cutoff current signal is being used. This represents the distorted steady-state test cutoff current signal at time t in phase k. This represents the distortion steady-state test cutoff current signal at time t in the k-th phase under the n-th base period, where n represents the base period and t0 represents the zero-crossing time of the signal.

[0048] It should be noted that the distortion steady-state test truncates the voltage and current signals based on this, and then uses a rectangular window function to truncate the entire cycle, retaining the waveform segments of each base cycle, and combining them in chronological order to more accurately restore the meter's response behavior in each cycle, thus laying the foundation for subsequent dynamic modulation and error analysis.

[0049] S103: Construct an orthogonal pseudo-random measurement matrix using a pseudo-random sequence diagonal matrix, an orthogonal Hadamard matrix, and a downsampling matrix.

[0050] The pseudo-random sequence diagonal matrix is ​​a diagonal matrix composed of pseudo-randomly generated binary sequences as its main diagonal elements, with all off-diagonal elements being 0. This matrix is ​​used to pseudo-randomly modulate the input signal, breaking its periodicity and predictability, thereby enhancing the system's ability to simulate non-stationary disturbances. It is suitable for compressed sensing and dynamic testing scenarios.

[0051] An orthogonal Hadamard matrix is ​​a square matrix composed of 1s and -1s with orthogonal rows and columns, exhibiting good orthogonality and fast transformation properties. In dynamic testing, introducing an orthogonal Hadamard matrix can achieve orthogonal transformation of the signal, preserving its energy structure while avoiding interference between dimensions, thus facilitating the construction of an uncorrelated measurement basis.

[0052] The downsampling matrix is ​​a sparse diagonal matrix with diagonal elements of 0 or 1, where only M diagonal elements are 1 and the rest are 0. It is used to select a subset of sampling points in the signal. The introduction of this matrix enables compressed sampling of high-dimensional signals, reducing data dimensionality while retaining key information, thus satisfying the compressed sensing theory.

[0053] Below we will introduce how to construct orthogonal pseudo-random measurement matrices.

[0054] Optionally, S103 specifically includes:

[0055] S1031: Select the first and second random sequences as the preferred sequence pair to generate the Gold sequence family:

[0056]

[0057] Where R1 represents the first random sequence and R2 represents the second random sequence. D represents bitwise XOR. l This indicates a right circular shift operation of 1 bit. , Indicates the length of the Gold sequence family.

[0058] Among them, the Gold sequence family is a set of pseudo-random binary sequences generated by combining two preferred pairs of m-sequences (maximum length sequences) through XOR and shift. It has good autocorrelation and cross-correlation properties. The Gold sequence family has been widely used in communication, radar, spread spectrum and testing fields, and will not be elaborated further in this invention.

[0059] S1032: Select the Gold sequence with the best balance from the Gold sequence family.

[0060] The balance refers to the even distribution of the digits 1 and 0 in the sequence. Specifically, the balance can be evaluated by the ratio of the number of digits 1 to digits 0 in the sequence.

[0061] S1033: Construct a pseudo-random sequence diagonal matrix based on the optimally balanced Gold sequence:

[0062]

[0063]

[0064]

[0065] Where G represents the diagonal matrix of the pseudo-random sequence, g ij G represents the value of the element in the i-th row and j-th column of the diagonal matrix of the pseudo-random sequence. iLet Z represent the middle vector of the i-th row of the diagonal matrix of the pseudo-random sequence, and let δ represent the Gold sequence with optimal balance. ij The parameter represents the value assigned to the i-th row and j-th column. N represents the size of the pseudo-random sequence diagonal matrix and the number of columns in the final generated orthogonal pseudo-random measurement matrix.

[0066] In this embodiment of the invention, selecting the best-balanced sequence from the Gold sequence family to construct a pseudo-random sequence diagonal matrix ensures that the modulated signal has good statistical balance in time, i.e., the distribution of "1"s and "0"s is as uniform as possible, thereby avoiding problems such as uneven energy distribution and excessive bias in signal modulation. This guarantee of balance helps improve the representativeness and stability of subsequent test signals, making the constructed orthogonal pseudo-random measurement matrix more robust and accurate in compressed sensing or fault testing, thereby improving the overall performance and reliability of dynamic error detection and smart meter performance testing.

[0067] S1034: Generate an orthogonal Hadamard matrix through Walsh-Hadamard transformation.

[0068] The Walsh-Hadamard Transform (WHT) is an orthogonal transform method that transforms signals using a Hadamard matrix consisting only of +1 and -1, eliminating the need for multiplication and resulting in high computational efficiency. Orthogonal Hadamard matrices can be generated using a recursive construction method based on the Walsh-Hadamard Transform. This is a well-established and mature technology, and will not be elaborated upon further in this invention.

[0069] S1035: Construct the downsampling matrix:

[0070]

[0071]

[0072]

[0073] Among them, D down Represents the downsampling matrix. Let M represent the value of the element in the i-th row and j-th column of the downsampling matrix. The sum of the elements that are 1 on the diagonal of the downsampling matrix is ​​M, and M also represents the number of columns in the final generated orthogonal pseudo-random measurement matrix. This indicates that it is much smaller than.

[0074] It should be noted that the downsampling matrix can be constructed by randomly setting 1s or 0s on the diagonal.

[0075] S1036: Combine the pseudo-random sequence diagonal matrix, the orthogonal Hadamard matrix, and the downsampling matrix to form the orthogonal pseudo-random measurement matrix:

[0076]

[0077] in, H represents an orthogonal pseudorandom measurement matrix. U This represents an orthogonal Hadamard matrix.

[0078] It should be noted that the principle behind constructing the orthogonal pseudo-random measurement matrix using the above method lies in achieving efficient sampling under sparse signal representation through the combination of three matrix modules. The pseudo-random sequence diagonal matrix is ​​used to perturb and weight the original signal. The orthogonal Hadamard matrix provides a sparse orthogonal transform basis, projecting the signal into the frequency domain or orthogonal space, enhancing separability and feature saliency. The downsampling matrix is ​​a sparse diagonal matrix used to select M observations from the result of the Hadamard transform, achieving dimensionality reduction in compressed sensing, reducing data dimensionality, and improving computational efficiency.

[0079] In this embodiment of the invention, three major mechanisms are integrated: pseudo-random perturbation, orthogonal transformation, and sparse sampling. This not only preserves the main features of the signal but also significantly reduces data redundancy and improves the complexity and diversity of the test signal. Compared with traditional methods, it can more realistically simulate current / voltage fluctuations under non-ideal power conditions, effectively improving the accuracy and coverage of dynamic error detection in smart meters and providing high-quality input features for subsequent performance testing.

[0080] S1037: Based on the incoherence criterion, determine the optimal orthogonal pseudorandom measurement matrix from the generated multiple orthogonal pseudorandom measurement matrices:

[0081]

[0082] Where μ represents the coherence measure, Describes an orthogonal pseudo-random measurement matrix. This represents a sparse matrix, and max indicates taking the maximum value. This represents a row vector consisting of the elements of the i-th row of the measurement matrix. T This represents the matrix transpose operation. This represents the column vector consisting of the elements of the j-th row of a sparse matrix. This indicates the inner product operation.

[0083] It's important to note that in signal processing and compressed sensing, the incoherence criterion is used to ensure there is no strong correlation between the measurement matrix and the sparse matrix, thereby improving signal recovery capabilities. The coherence metric is the maximum value of the inner product between each row of the measurement matrix and each column of the sparse matrix. The smaller this maximum value, the weaker the correlation between the measurement matrix and the sparse matrix, resulting in less interference and information loss during signal recovery.

[0084] Furthermore, the orthogonal pseudorandom measurement matrix with the smallest coherence metric can be selected as the optimal orthogonal pseudorandom measurement matrix.

[0085] In this embodiment of the invention, the optimal orthogonal pseudo-random measurement matrix is ​​selected using the incoherence criterion, with the goal of minimizing the inner product between the measurement matrix and the sparse matrix. This ensures that the compressed sampling of the signal does not lose critical information and helps improve the accuracy of subsequent signal recovery.

[0086] S104: Modulate the distorted steady-state test cutoff voltage and current signals using an orthogonal pseudo-random measurement matrix to generate orthogonal pseudo-random dynamic test voltage, current, and power signals.

[0087] Optionally, S104 specifically includes:

[0088] S1041: Perform a vectorization operation on the orthogonal pseudo-random measurement matrix to form an orthogonal pseudo-random dynamic sequence:

[0089]

[0090] Where Vec represents vectorization operation. Describes the k-th phase orthogonal pseudo-random dynamic sequence. Let L represent the Lth sequence value in the k-th phase orthogonal pseudo-random dynamic sequence, where L = MN.

[0091] It should be noted that converting the two-dimensional matrix into a one-dimensional modulation sequence facilitates subsequent element-by-element modulation with voltage and current signals. Simultaneously, the orthogonal pseudo-randomness of the original matrix is ​​preserved, resulting in a modulated signal with excellent structure and resolution.

[0092] S1042: The distorted steady-state test truncated voltage and current signals are modulated using the orthogonal pseudo-random dynamic sequence to obtain the orthogonal pseudo-random dynamic test voltage and current signals:

[0093]

[0094]

[0095] Among them, ud This represents an orthogonal pseudo-random dynamic test voltage signal. Let i represent the orthogonal pseudo-random dynamic test voltage signal at time t in phase k. d This represents an orthogonal pseudo-random dynamic test current signal. This represents the orthogonal pseudo-random dynamic test current signal at time t in phase k. This represents Hadamard multiplication.

[0096] It should be noted that converting the static truncation signal into a dynamic signal enables "on / off" control of power transmission. Combined with subsequent pseudo-random modulation, it can simulate complex real-world operating conditions such as sudden changes, jumps, and jitter in the power grid.

[0097] S1043: Perform Hadamard multiplication on the orthogonal pseudo-random dynamic test voltage signal and current signal to obtain the orthogonal pseudo-random dynamic test power signal:

[0098]

[0099] Where p represents the orthogonal pseudo-random dynamic test power signal. This represents the orthogonal pseudo-random dynamic test power signal at time t of phase k.

[0100] It should be noted that the orthogonal pseudo-random dynamic test power signal can reflect the transmission intensity of electrical energy at each moment in real time, forming a dynamic power input. Consistent with the internal energy accumulation logic of the meter, this helps to accurately simulate real power fluctuations. Ultimately, this results in a highly realistic "power drive signal" with rich perturbations, facilitating high-precision dynamic error detection.

[0101] S2: Input the orthogonal pseudo-random dynamic test power signal into the smart meter under test for testing, and statistically analyze the dynamic power error signal.

[0102] In one possible implementation, S2 specifically includes:

[0103] S201: Input the orthogonal pseudo-random dynamic test power signal into the smart meter under test for testing.

[0104] S202: Generate a gating signal to control the test interval, each of the test intervals containing the base periods of multiple orthogonal pseudo-random dynamic test power signals.

[0105] S203: Determine the reference dynamic energy sequence generated by inputting the orthogonal pseudo-random dynamic test power signal into the smart meter under test:

[0106]

[0107] Among them, En This represents the reference dynamic electrical energy in the nth base period. This represents the nth sequence value in the k-th phase orthogonal pseudo-random dynamic sequence, taking the value 0 or 1, which determines whether electrical energy is transmitted in the nth base period. At that time, the dynamic electrical energy is E0. At that time, the dynamic electrical energy is 0, N t This indicates the total number of base periods contained in each test interval.

[0108] It should be noted that, based on orthogonal pseudo-random sequences, a 0 / 1 switching control is introduced to theoretically construct the desired energy transfer mode, thereby establishing a comparative benchmark and providing a "gold standard" for subsequent error assessment.

[0109] It should be noted that the dynamic electrical energy specifically refers to:

[0110]

[0111] Where t0 represents the start time of a certain base period, T represents the base period, and p(t) represents the power at time t.

[0112] S204: Calculate the reference dynamic cumulative energy based on the aforementioned reference dynamic energy sequence:

[0113]

[0114] Among them, E ref This indicates the reference dynamic cumulative electrical energy.

[0115] S205: Measure the actual dynamic cumulative electrical energy generated when the orthogonal pseudo-random dynamic test power signal is input to the smart meter under test.

[0116] S206: Calculate the dynamic energy error between the reference dynamic cumulative energy and the actual dynamic cumulative energy.

[0117]

[0118] Where e represents the dynamic energy error, E act The measured actual dynamic cumulative electrical energy.

[0119] It should be noted that dynamic energy error provides a precise quantitative error indicator, which can be used to determine whether the meter is qualified and serve as a basis for further diagnosis.

[0120] S207: Record the dynamic power error under each test interval in chronological order as the dynamic power error signal.

[0121] In this embodiment of the invention, orthogonal pseudo-random dynamic test voltage and current signals are constructed to accurately simulate real-life situations such as load asymmetry, random jumps, and periodic jitter. The orthogonal pseudo-random dynamic test voltage and current signals are then used to test the dynamic error of the smart meter, thereby improving the smart meter's sensitivity to dynamic errors. This ensures that the test results can truly reflect the meter's measurement performance in real-world environments, thus improving the smart meter's factory quality.

[0122] S3: Determine whether the dynamic power error signal is lower than the preset power error. If yes, determine that the smart meter under test is qualified. Otherwise, determine that the performance of the smart meter under test is questionable, and proceed to the next step for detailed testing.

[0123] Those skilled in the art can set the magnitude of the preset power error according to the actual situation, and the present invention does not limit it.

[0124] In this embodiment of the invention, by setting a preset power error threshold, rapid screening and preliminary judgment of smart meters can be achieved. When the dynamic power error signal is below the threshold, it indicates that the meter still has good metering accuracy under simulated complex load conditions and can be directly judged as qualified; if it exceeds the threshold, it automatically enters a more in-depth feature analysis and performance testing stage to avoid misjudgment and missed judgment. This hierarchical detection mechanism not only improves testing efficiency and reduces resource waste, but also enables targeted and refined processing of abnormal meters, enhancing the accuracy and intelligence level of the entire quality inspection process.

[0125] Reference manual attached Figure 3 The diagram shows a structural schematic of a smart meter performance testing model provided in an embodiment of the present invention.

[0126] S4: Extract signal features from the dynamic power error signal using a deep belief network.

[0127] The deep belief network consists of multiple layers of restricted Boltzmann machines, each layer of which comprises a visible layer and a hidden layer. Deep belief networks are a very mature existing technology, and will not be described in detail here.

[0128] S5: Input the signal features extracted by the deep belief network into the multilayer perceptron.

[0129] In one possible implementation, S5 specifically includes:

[0130] S501: Evaluate the feature dispersion of the signal features extracted by the deep belief network:

[0131]

[0132] Where J represents the feature dispersion, SB S represents the inter-class eigenvalue scatter matrix. W tr represents the within-class feature scatter matrix, and tr represents the trace of the matrix.

[0133] The larger the inter-class scatter matrix and the smaller the inter-class scatter matrix, the stronger the feature's ability to distinguish between different categories, and the higher the feature dispersion value. Using this metric as an evaluation criterion helps quantify the feature extraction effect and serves as a basis for determining whether to trigger the feature enhancement mechanism, thereby improving the accuracy and reliability of performance testing and avoiding misclassification caused by feature aliasing.

[0134] S502: Determine whether the feature dispersion is greater than a preset dispersion. If yes, proceed directly to S6. Otherwise, trigger the feature enhancement mechanism.

[0135] Those skilled in the art can set the preset dispersion value according to the actual situation, and the present invention does not limit it.

[0136] In this embodiment of the invention, by determining whether the feature dispersion is greater than a preset threshold, the ability to dynamically evaluate whether the currently extracted features have good class discrimination ability can be assessed. If the feature dispersion is high, it indicates that the existing features are sufficient for accurate classification, and the model can directly enter the performance testing stage to improve efficiency; otherwise, a feature enhancement mechanism is triggered to introduce more dimensions or more discriminative features, thereby improving the model's ability to recognize complex or ambiguous samples. This strategy balances diagnostic accuracy and computational efficiency, realizing an adaptive and refined smart meter fault identification process.

[0137] Optionally, the feature enhancement mechanism specifically includes: introducing wavelet transform to extract the time-frequency domain features of the dynamic power error signal, and using the time-frequency domain features as input to the deep belief network.

[0138] In this embodiment of the invention, wavelet transform is introduced to extract the time-frequency domain features of dynamic power error signals. This effectively captures transient changes and frequency components that appear at different time scales in the signal, thereby revealing richer and more discriminative feature information. Using these time-frequency domain features as supplementary inputs to deep belief networks helps improve the model's ability to understand complex, non-stationary signals, enhances the accuracy and robustness of performance testing, and is particularly effective in dealing with scenarios where dynamic error fluctuations are severe or local features are significant.

[0139] Optionally, the feature enhancement mechanism specifically includes: introducing a dynamic response voltage signal, a dynamic response voltage signal, and / or a dynamic response power signal, which, together with the dynamic power error signal, are used as inputs to the deep belief network.

[0140] The dynamic response voltage signal refers to the signal generated by the smart meter under test when the orthogonal pseudo-random dynamic test voltage signal is input to the smart meter under test for testing, and the dynamic response current signal refers to the signal generated by the smart meter under test when the orthogonal pseudo-random dynamic test current signal is input to the smart meter under test for testing.

[0141] In this embodiment of the invention, dynamic response voltage signals, current signals, and / or power signals are introduced as supplementary inputs to comprehensively reflect the true response behavior of smart meters under orthogonal pseudo-random dynamic test signal excitation. These response signals contain dynamic characteristics of the meter's internal measurement and processing mechanisms, which helps to uncover potential abnormal patterns or minor fault features. Inputting these signals, along with dynamic energy error signals, into a deep belief network not only enriches the feature dimensions but also improves the model's ability to identify complex fault types, enhancing the accuracy and robustness of performance testing.

[0142] S6: The performance of the smart meter under test is performed using the multilayer sensor.

[0143] In one possible implementation, S6 specifically includes sub-steps S601 to S603:

[0144] S601: Input multiple signal features extracted by the deep belief network into the input layer of the multilayer perceptron.

[0145] S602: Extract the hidden states from the signal features in the hidden layer of the multilayer perceptron:

[0146] S603: In the output layer of the multilayer sensor, the performance of the smart meter to be tested is performed according to the hidden state.

[0147] Multilayer perceptrons are a very mature existing technology, and the specific usage process and principles of multilayer perceptrons will not be elaborated in this invention.

[0148] In this embodiment of the invention, by using deep belief networks and multilayer perceptrons to perform performance testing on smart meters, the specific fault types of smart meters can be accurately diagnosed, providing effective guidance for subsequent smart meter production and improving the production efficiency of smart meters.

[0149] Reference manual attached Figure 4 The diagram shows a structural schematic of a smart meter performance testing system provided by the present invention.

[0150] The present invention also provides a smart meter performance testing system 20, applied to the above-mentioned smart meter performance testing method, comprising:

[0151] Processor 201.

[0152] The memory 202 stores computer-readable instructions, which, when executed by the processor 201, implement the smart meter performance testing method as described in the method embodiment.

[0153] The smart meter performance testing system 20 provided by the present invention can perform the above-described smart meter performance testing method and achieve the same or similar technical effects. To avoid duplication, the present invention will not elaborate further.

[0154] It should be understood that the processor in the embodiments of the present invention can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0155] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0156] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0157] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0158] This invention provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the smart meter performance testing method as described in the method embodiment.

[0159] The present invention provides a computer-readable storage medium that can implement the steps and effects of the smart meter performance testing method of the above-described method embodiments. To avoid repetition, the present invention will not repeat them.

Claims

1. A method for testing the performance of a smart meter, characterized in that, include: S1: Construct orthogonal pseudo-random dynamic test power signals; S2: Input the orthogonal pseudo-random dynamic test power signal into the smart meter under test for testing, and statistically calculate the dynamic energy error signal; S3: Determine whether the dynamic power error signal is lower than the preset power error; if so, determine that the smart meter to be tested is qualified. Otherwise, if the performance of the smart meter under test is deemed questionable, proceed to the next step for detailed testing; S4: Extract signal features from the dynamic power error signal using a deep belief network; S5: Input the signal features extracted by the deep belief network into the multilayer perceptron; S6: The performance of the smart meter under test is performed using the multilayer sensor; Specifically, S1 includes: S101: Construct distorted steady-state standard test voltage and current signals; S102: The distorted steady-state standard test voltage and current signals are truncated using multiple rectangular window functions to obtain the truncated voltage and current signals for each base period, and then combined to form the distorted steady-state test truncated voltage and current signals. S103: Construct an orthogonal pseudo-random measurement matrix using a pseudo-random sequence diagonal matrix, an orthogonal Hadamard matrix, and a downsampling matrix; S104: Modulate the distorted steady-state test cutoff voltage signal and current signal through the orthogonal pseudo-random measurement matrix to generate orthogonal pseudo-random dynamic test voltage signal, current signal and power signal; Specifically, S2 includes: S201: Input the orthogonal pseudo-random dynamic test power signal into the smart meter under test for testing; S202: Generate a gating signal to control the test interval, each of the test intervals containing the base periods of multiple orthogonal pseudo-random dynamic test power signals; S203: Determine the reference dynamic energy sequence generated by inputting the orthogonal pseudo-random dynamic test power signal into the smart meter under test; S204: Calculate the reference dynamic cumulative energy based on the aforementioned reference dynamic energy sequence; S205: Measure the actual dynamic cumulative electrical energy generated by inputting the orthogonal pseudo-random dynamic test power signal into the smart meter under test; S206: Calculate the dynamic energy error between the reference dynamic cumulative energy and the actual dynamic cumulative energy; S207: Record the dynamic power error under each of the test intervals in chronological order as the dynamic power error signal.

2. The smart meter performance testing method according to claim 1, characterized in that, S103 specifically includes: S1031: Select the first random sequence and the second random sequence as the preferred sequence pair to generate the Gold sequence family; S1032: Select the Gold sequence with the best balance from the Gold sequence family; S1033: Construct the diagonal matrix of the pseudo-random sequence based on the Gold sequence with the best balance; S1034: Generate the orthogonal Hadamard matrix through Walsh-Hadamard transformation; S1035: Construct the downsampling matrix; S1036: The orthogonal pseudo-random measurement matrix is ​​formed by combining the pseudo-random sequence diagonal matrix, the orthogonal Hadamard matrix, and the downsampling matrix; S1037: Based on the incoherence criterion, determine the optimal orthogonal pseudorandom measurement matrix from the various generated orthogonal pseudorandom measurement matrices.

3. The smart meter performance testing method according to claim 1, characterized in that, S104 specifically includes: S1041: Perform a vectorization operation on the orthogonal pseudo-random measurement matrix to form an orthogonal pseudo-random dynamic sequence; S1042: Modulate the distorted steady-state test cutoff voltage signal and current signal using the orthogonal pseudo-random dynamic sequence to obtain the orthogonal pseudo-random dynamic test voltage signal and current signal; S1043: Perform Hadamard multiplication on the orthogonal pseudo-random dynamic test voltage signal and current signal to obtain the orthogonal pseudo-random dynamic test power signal.

4. The smart meter performance testing method according to claim 1, characterized in that, S5 specifically includes: S501: Evaluate the feature dispersion of the signal features extracted by the deep belief network; S502: Determine whether the feature dispersion is greater than the preset dispersion; if yes, proceed directly to S6; otherwise, trigger the feature enhancement mechanism.

5. The smart meter performance testing method according to claim 4, characterized in that, The feature enhancement mechanism specifically includes: Wavelet transform is introduced to extract the time-frequency domain features of the dynamic power error signal, and these time-frequency domain features are used as input to the deep belief network.

6. The smart meter performance testing method according to claim 4, characterized in that, The feature enhancement mechanism specifically includes: Dynamic response voltage signal, dynamic response current signal and / or dynamic response power signal are introduced and used together with the dynamic energy error signal as input to the deep belief network; The dynamic response voltage signal refers to the signal generated by the smart meter under test when the orthogonal pseudo-random dynamic test voltage signal is input to the smart meter under test for testing, and the dynamic response current signal refers to the signal generated by the smart meter under test when the orthogonal pseudo-random dynamic test current signal is input to the smart meter under test for testing.

7. The smart meter performance testing method according to claim 4, characterized in that, S6 specifically includes: S601: Input multiple signal features extracted by the deep belief network into the input layer of the multilayer perceptron; S602: Extract the hidden state from the signal features in the hidden layer of the multilayer perceptron; S603: In the output layer of the multilayer sensor, the performance of the smart meter to be tested is performed according to the hidden state.

8. A smart meter performance testing system, characterized in that, include: processor; A memory storing computer-readable instructions, which, when executed by the processor, implement the smart meter performance testing method as described in any one of claims 1 to 7.

Images