Method, apparatus, equipment and storage medium for correcting sampling timing errors in merging units

Abstract

This application discloses a method, apparatus, device, and storage medium for correcting sampling timing errors in merging units, relating to the field of online monitoring technology. The method includes: acquiring the sampling timing of merging units in a smart substation; collecting sampling window data based on the sampling timing; determining multi-dimensional error characteristic timing based on the sampling window data; determining the operating status information of the merging units based on the multi-dimensional error characteristic timing; determining an initial parameter combination based on a preset search strategy; optimizing the initial parameter combination using the preset search strategy and operating status information to obtain a target parameter combination; constructing a target prediction model based on the target parameter combination; determining a synchronization error prediction value based on the multi-dimensional error characteristic timing and the target prediction model; and performing closed-loop correction on the sampling timing based on the synchronization error prediction value to obtain the corrected sampling timing. This application achieves adaptive matching of operating conditions, improving the accuracy of synchronization error prediction under complex operating conditions.

Description

Technical Field

[0001] This application relates to the field of online monitoring technology, and in particular to a method, apparatus, equipment and storage medium for correcting sampling timing errors in merging units. Background Technology

[0002] The merging unit is a key device in smart substations for realizing analog signal acquisition and digital signal output. The accuracy of its sampling timing directly affects the reliability of applications such as relay protection, power metering, and synchronous phasor measurement. Due to factors such as temperature drift, clock source aging, and abnormal timing signals, the sampling timing of the merging unit will shift over time, resulting in synchronization errors in the sampled data.

[0003] To eliminate the aforementioned errors, existing methods typically employ predictive models to estimate the errors and adjust the sampling timing based on the estimation results. The accuracy of the predictive model is highly dependent on the model parameter configuration. If the parameters are improperly configured, there will be a significant deviation between the predicted and actual error values, directly affecting the effectiveness of closed-loop correction. In current practice, there are two main methods for determining model parameters: one is to rely on manual experience to set them, and the other is to use a standard optimization algorithm for offline search and then fix them. However, the actual operating conditions of the merging unit are constantly changing, and neither of the above methods can adaptively adjust the model parameters according to these changes: manual experience cannot cover all operating conditions, and the parameter combinations obtained through offline search cannot continuously adapt to the changed operating conditions. This results in significant differences in the prediction accuracy of the error prediction model under different operating conditions, limiting the overall performance of closed-loop correction.

[0004] Therefore, the parameter optimization process of the combined unit error prediction model in the current scheme is out of touch with the actual operating state of the equipment, resulting in insufficient error prediction accuracy under complex working conditions, which is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] The main objective of this application is to provide a method, apparatus, device, and storage medium for correcting sampling timing errors in merging units, aiming to solve the technical problem that the parameter optimization process of the merging unit error prediction model is disconnected from the actual operating state of the equipment, resulting in insufficient error prediction accuracy under complex working conditions.

[0006] To achieve the above objectives, this application proposes a method for correcting sampling timing errors in merging units, the method comprising: The sampling timing sequence of the merging unit in the intelligent substation is obtained, and sampling window data is obtained by collecting data based on the sampling timing sequence. The multidimensional error feature timing is determined based on the sampling window data, and the operating status information of the merging unit is determined based on the multidimensional error feature timing. An initial parameter combination is determined based on a preset search strategy, and the initial parameter combination is searched and optimized using the preset search strategy and the running status information to obtain a target parameter combination. A target prediction model is constructed based on the target parameter combination, and the synchronization error prediction value is determined based on the multidimensional error feature time series and the target prediction model. The sampling timing is corrected by performing closed-loop correction based on the predicted synchronization error value to obtain the corrected sampling timing.

[0007] In one embodiment, the step of determining an initial parameter combination based on a preset search strategy, and optimizing the initial parameter combination using the preset search strategy and the running status information to obtain a target parameter combination, includes: Generate initial parameter combinations based on a preset search strategy; The preset search strategy is adjusted based on the operating status information to obtain an updated search strategy; Based on the updated search strategy, the initial parameter combination is optimized to obtain the process parameter combination; The performance of the process parameter combination is evaluated to obtain the performance evaluation results; Based on the performance evaluation results, candidate parameter combinations are determined from the process parameter combinations and the initial parameter combinations; The candidate parameter combination is used as the initial parameter combination, and the process returns to the step of adjusting the preset search strategy according to the running status information until the termination condition is met, and the candidate parameter combination is used as the target parameter combination.

[0008] In one embodiment, adjusting the preset search strategy based on the running status information to obtain an updated search strategy includes: The current operating condition and signal non-stationarity in the operating status information are obtained, and the search direction, search step size and search mode of the preset search strategy are obtained. Adjust the search direction based on the current operating conditions; Adjust the search step size and the search mode according to the non-stationarity of the signal; An updated search strategy is obtained based on the adjusted search direction, search step size, and search pattern.

[0009] In one embodiment, determining the multidimensional error feature timing based on the sampling window data, and determining the operating status information of the merging unit based on the multidimensional error feature timing, includes: Based on the sampling window data, the fundamental frequency characteristics and broadband characteristics are determined; Based on the fundamental wave characteristics and the broadband characteristics, a multidimensional error feature time series is generated. Based on the multidimensional error characteristic time series, the current operating condition and signal non-stationarity of the merging unit are determined; The current operating conditions and the signal non-stationarity are used as the operating status information of the merging unit.

[0010] In one embodiment, determining the synchronization error prediction value based on the multidimensional error feature time series and the target prediction model includes: The multidimensional error features are input into the target prediction model in time series for feature extraction to obtain time-dependent features; Based on the time-dependent characteristics, the predicted value of the synchronization error is determined.

[0011] In one embodiment, the step of performing closed-loop correction on the sampling timing based on the predicted synchronization error value to obtain the corrected sampling timing includes: Based on the predicted synchronization error value, determine the sampling timing correction amount; Based on the sampling timing correction amount, the frequency and / or phase of the sampling timing are adjusted to obtain the corrected sampling timing.

[0012] In one embodiment, obtaining the sampling timing sequence of the merging unit in the smart substation, and collecting sampling window data using the sampling timing sequence, includes: Obtain the sampling timing, preset sampling window length, and window overlap of the merging unit in the smart substation; The voltage and current signals output by the merging unit are sampled according to the sampling timing to obtain a continuous sampling sequence; Based on the sampling window length and the window overlap, the continuous sampling sequence is divided into multiple partially overlapping sampling window data.

[0013] Furthermore, to achieve the above objectives, this application also proposes a sampling timing error correction device for a merging unit, the sampling timing error correction device comprising: The data sampling module is used to obtain the sampling timing of the merging unit in the smart substation, and to collect sampling window data based on the sampling timing. The feature extraction module is used to determine the multidimensional error feature time series based on the sampling window data, and to determine the running status information of the merging unit based on the multidimensional error feature time series; The parameter optimization module is used to determine the initial parameter combination based on a preset search strategy, and to optimize the initial parameter combination by the preset search strategy and the running status information to obtain the target parameter combination. The error prediction module is used to construct a target prediction model based on the target parameter combination, and to determine the synchronous error prediction value based on the multidimensional error feature time series and the target prediction model. The timing correction module is used to perform closed-loop correction on the sampling timing based on the predicted synchronization error value to obtain the corrected sampling timing.

[0014] In addition, to achieve the above objectives, this application also proposes a merging unit sampling timing error correction device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the merging unit sampling timing error correction method described above.

[0015] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the merging unit sampling timing error correction method described above.

[0016] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the merging unit sampling timing error correction method described above.

[0017] One or more technical solutions proposed in this application have at least the following technical effects: This application extracts multi-dimensional error feature time series from sampled data to identify the current operating conditions of the merging unit and signal non-stationarity, thus incorporating the actual operating state of the equipment into the parameter optimization basis. Based on this, the operating state information drives the adjustment of the preset search strategy, enabling the parameter optimization process to dynamically adjust the search direction and search step size according to changes in operating conditions and non-stationarity, thereby obtaining a target parameter combination that matches the current operating conditions. Since the key parameters of the target prediction model are specifically optimized based on the current operating state of the merging unit, rather than using fixed configurations detached from actual operating conditions, the model can maintain high synchronization error prediction accuracy under different operating conditions, solving the problem of the parameter optimization process being disconnected from the operating state in existing technologies. Furthermore, accurate synchronization error prediction values ​​provide reliable input conditions for sampling time series closed-loop correction, thereby ensuring the effectiveness of the correction and transferring the improved prediction accuracy to the closed-loop correction stage. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an embodiment of the sampling timing error correction method for the merging unit in this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the sampling timing error correction method for the merging unit in this application. Figure 3 A simplified flowchart is provided for an embodiment of the sampling timing error correction method for the merging unit in this application; Figure 4 This is a schematic diagram of the module structure of the sampling timing error correction device for the merging unit in an embodiment of this application; Figure 5 This is a schematic diagram of the device structure of the hardware operating environment involved in the sampling timing error correction method of the merging unit in the embodiments of this application.

[0021] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0022] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0023] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0024] The main solution of this application embodiment is as follows: First, obtain the sampling timing sequence of the merging unit in the intelligent substation; second, collect sampling window data using the sampling timing sequence; third, determine the multi-dimensional error characteristic timing sequence based on the sampling window data, and determine the operating status information of the merging unit based on the multi-dimensional error characteristic timing sequence; fourth, determine the initial parameter combination based on a preset search strategy, and optimize the initial parameter combination using the preset search strategy and operating status information to obtain the target parameter combination; fifth, construct a target prediction model based on the target parameter combination, and determine the synchronization error prediction value based on the multi-dimensional error characteristic timing sequence and the target prediction model; sixth, perform closed-loop correction on the sampling timing sequence based on the synchronization error prediction value to obtain the corrected sampling timing sequence.

[0025] Because the parameter optimization process of the existing unit error prediction model is disconnected from the actual operating state of the equipment, the error prediction accuracy is insufficient under complex working conditions.

[0026] This application provides a solution that acquires sampling window data by obtaining the sampling timing of the merging unit, and determines the multidimensional error feature timing based on this sampling window data. This allows for the extraction of the merging unit's operating status information, enabling quantitative identification of the current operating condition and signal non-stationarity. Furthermore, this operating status information is used as the basis for adjusting a preset search strategy. The search direction is adjusted using operating condition information, and the search step size and search mode are adjusted using non-stationarity information. This links the parameter optimization process with the actual operating status of the equipment, obtaining a target parameter combination that matches the current operating condition. Because the parameter configuration of the target prediction model can adapt to the actual operating status of the merging unit, rather than using a fixed configuration detached from the operating condition, the model can maintain high synchronization error prediction accuracy even under complex operating conditions. This solves the problem of insufficient prediction accuracy caused by the disconnect between the parameter optimization process and the operating status in existing technologies.

[0027] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or processor capable of performing the above functions. The following description uses a processor as an example to illustrate this embodiment and the subsequent embodiments. All actions involving the acquisition of signals, information, or data in this application are performed in accordance with the relevant data protection regulations of the locality and with authorization from the owner of the corresponding device.

[0028] Based on this, embodiments of this application provide a method for correcting sampling timing errors in merging units, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the sampling timing error correction method for the merging unit in this application.

[0029] In this embodiment, the sampling timing error correction method of the merging unit includes steps S10 to S50: Step S10: Obtain the sampling timing of the merging unit in the intelligent substation, and collect sampling window data using the sampling timing.

[0030] It should be noted that the merging unit is a process-level device in a smart substation used to convert the analog signals output from instrument transformers into digital sampled values ​​and output them according to the IEC 61850 protocol. Sampling timing refers to the time reference signal used within the merging unit to control the analog-to-digital converter to perform equally spaced sampling; its frequency stability and phase consistency determine the accuracy of the time stamp of the sampled data. Sampling window data refers to a set of continuous sampling points extracted from a continuous sampling sequence according to a preset time length, serving as the basic unit for subsequent feature extraction.

[0031] In one feasible implementation, step S10 may include: acquiring the sampling timing sequence, preset sampling window length, and window overlap of the merging unit in the smart substation; sampling the voltage and current signals output by the merging unit according to the sampling timing sequence to obtain a continuous sampling sequence; and dividing the continuous sampling sequence into multiple partially overlapping sampling window data based on the sampling window length and the window overlap.

[0032] It should be noted that the sampling window length refers to the time span or number of sampling points contained in a single sampling window. Window overlap refers to the degree of overlap between two adjacent sampling windows on the time axis, used to control the step size of window sliding. A continuous sampling sequence refers to the sequence of original digital signals obtained by sampling voltage and current signals at equal intervals according to the sampling timing sequence.

[0033] Specifically, the nominal frequency and actual time stamp information of the current sampling sequence are obtained through the internal data interface or sampling value message interface of the merging unit; using the sampling sequence as the sampling trigger signal, the voltage signal and current signal output by the merging unit are sampled by analog-to-digital conversion at a sampling frequency of 10kHz to obtain a continuous digital sampling sequence; sampling windows are sequentially extracted from the continuous sampling sequence according to a sampling window length of 20ms and a sliding step size of 10ms, so that adjacent sampling windows form a 50% overlap rate, resulting in multiple partially overlapping sampling window data, each sampling window containing 200 sampling points.

[0034] The voltage sampling sequence, current sampling sequence, and synchronization time stamp sequence in the first sampling window are represented as follows:

[0035]

[0036]

[0037] In the formula, This represents the number of sampling points in a single sampling window.

[0038] It is understandable that since the voltage and current signals output by the merging unit are continuous analog quantities, if the original samples collected in the sampling time sequence are directly processed, there is a lack of sufficient temporal correlation information between adjacent samples, making it impossible to construct structured samples for time series prediction. However, by dividing the continuous sampling sequence into multiple partially overlapping sampling windows according to the preset window length and window overlap, the temporal continuity between adjacent windows is preserved, and the overlapping mechanism reduces the truncation effect of the window boundary on the transient signal feature extraction results, thereby improving the stability and accuracy of subsequent multidimensional error feature time series extraction.

[0039] Step S20: Determine the multidimensional error feature timing based on the sampling window data, and determine the operating status information of the merging unit according to the multidimensional error feature timing.

[0040] It should be noted that multidimensional error feature time series refers to a sequence of data formed by arranging features extracted from multiple consecutive sampling windows, such as amplitude error, phase error, synchronization error, harmonic distortion rate, signal-to-noise ratio, clock jitter index, transient frequency band energy, and temperature change, in chronological order. Operating status information refers to a set of data used to characterize the current operating condition type and signal stability of the merging unit.

[0041] In one feasible implementation, step S20 may include: determining fundamental characteristics and broadband characteristics based on the sampling window data; generating a multidimensional error feature time series based on the fundamental characteristics and the broadband characteristics; determining the current operating condition and signal non-stationarity of the merging unit based on the multidimensional error feature time series; and using the current operating condition and the signal non-stationarity as the operating status information of the merging unit.

[0042] It should be noted that the fundamental frequency characteristic refers to the amplitude and phase information of the fundamental frequency component extracted from the sampling window. The broadband characteristic refers to the harmonic components, transient components, and broadband noise characteristics extracted from the sampling window. The current operating condition refers to the operating state category of the merging unit at the current moment, including normal steady state, harmonic interference, transient impact, temperature drift, and clock anomaly. Signal non-stationarity is a quantitative indicator that reflects the degree of change in the statistical characteristics of the signal between adjacent sampling windows.

[0043] Specifically, firstly, the data from each sampling window is input into the fundamental frequency processing branch and the broadband processing branch, respectively. The fundamental frequency processing branch uses an adaptive finite impulse response filter to extract the fundamental frequency component centered on the grid fundamental frequency, performs time alignment of different sampling channels according to the synchronization reference time scale, and calculates the fundamental frequency amplitude and phase using discrete Fourier transform. The broadband processing branch uses wavelet thresholding to suppress random noise and employs fast Fourier transform to extract harmonic amplitude, harmonic distortion rate, and transient frequency band energy.

[0044] Calculate the amplitude error of the data in the k-th sampling window based on the fundamental amplitude obtained from the fundamental amplitude processing branch and the reference amplitude. :

[0045] In the formula, To measure the fundamental amplitude of the combined unit's measurement channel, This is the corresponding reference amplitude.

[0046] Phase error of the data in the kth sampling window for:

[0047] In the formula, To measure the fundamental phase of the merging unit's measurement channel, For synchronization reference phase.

[0048] Synchronization error Calculated based on the deviation between the sampling timescale of the merging unit and the synchronous reference timescale:

[0049] In the formula, For the sampling time scale of the merging unit, For reference time scale, the harmonic distortion rate of the broadband processing branch output is expressed as:

[0050] In the formula, The fundamental amplitude in the data of the k-th sampling window. The amplitude of the nth harmonic. The highest harmonic order used in the calculation.

[0051] Clock jitter is expressed as:

[0052] In the formula, This is the rated sampling period.

[0053] By fusing amplitude error, phase error, synchronization error, harmonic distortion rate, signal-to-noise ratio, clock jitter index, transient frequency band energy, and temperature change, a multi-dimensional error feature vector corresponding to the data of the k-th sampling window is obtained:

[0054] In the formula, For signal-to-noise ratio, Transient frequency band energy, This represents the temperature change of the device. The feature vectors corresponding to data from multiple consecutive sampling windows are arranged in chronological order to form a multidimensional error feature time series.

[0055] The current operating status of the merging unit is identified based on the multidimensional error characteristic timing. This embodiment classifies operating conditions into five categories: normal steady state, harmonic interference, transient impact, temperature drift, and clock anomaly. When the harmonic distortion rate, transient frequency band energy, device temperature change rate, and clock jitter index are all within the normal range, it is determined to be a normal steady state operating condition; when the harmonic distortion rate continuously exceeds a preset threshold, it is determined to be a harmonic interference operating condition; when the transient frequency band energy or the peak factor of the sampled signal increases significantly in a short period of time, it is determined to be a transient impact operating condition; when the device temperature continuously changes and the amplitude error or phase error shows a slow drift trend, it is determined to be a temperature drift operating condition; when the synchronization error or clock jitter index continuously increases, it is determined to be a clock anomaly operating condition.

[0056] To reflect the degree of change in the statistical characteristics of the signal between adjacent sampling windows, a signal non-stationarity index is constructed:

[0057] In the formula, Let be the signal non-stationarity of the data in the k-th sampling window. and These represent the j-th type of working condition characteristics in adjacent sampling window data. For the corresponding feature reference value, These are the weighting coefficients. A positive number is set to prevent the denominator from being zero. When the merging unit is in a normal steady state, the changes in various characteristics are small, and the non-stationarity remains at a low level; when harmonic interference, transient impact, or clock anomalies occur, the changes in characteristics between adjacent windows increase, and the corresponding non-stationarity increases accordingly.

[0058] Understandably, since the sampled signal output by the merging unit contains both the fundamental component used for measurement and protection, and the broadband component reflecting harmonic pollution and transient interference, parallel processing by the fundamental and broadband processing branches can accurately extract the fundamental amplitude and phase while retaining harmonic and transient information, avoiding the loss of effective features by a single filtering method. Because the variation patterns of various error characteristics differ significantly under different operating conditions, fusing and characterizing various features through multi-dimensional error feature time series can provide a complete feature basis for operating condition identification. Constructing signal non-stationarity to quantitatively reflect the degree of change in signal statistical characteristics between adjacent windows can provide a quantitative basis for the adaptive adjustment of subsequent search strategies, enabling the optimization process to dynamically adapt to the actual signal fluctuation level.

[0059] Step S30: Determine the initial parameter combination based on the preset search strategy, and optimize the initial parameter combination by using the preset search strategy and the running status information to obtain the target parameter combination.

[0060] It should be noted that the preset search strategy is implemented based on a metaheuristic algorithm of population iteration. In a preferred embodiment of this application, the Gannet optimization algorithm is used as the specific implementation method. The initial parameter combination refers to the initial set of values ​​for the prediction model parameters to be optimized, including the filter order, wavelet threshold parameter, number of input history windows, number of LSTM (Long Short-Term Memory) hidden layer nodes, learning rate, batch size, dropout rate, number of prediction windows, and amplitude error warning threshold, phase error warning threshold, and synchronization error warning threshold. The target parameter combination refers to the optimal set of parameter values ​​that meet the preset performance conditions after search optimization.

[0061] Specifically, the filter order, wavelet threshold parameters, number of input history windows, number of hidden layer nodes in the Long Short-Term Memory network, learning rate, batch size, dropout rate, number of prediction windows, and various error warning thresholds are encoded into individual position vectors, with each individual corresponding to a set of candidate parameter combinations. An initial population is generated using a Tent chaotic mapping, mapping the chaotic sequence to the search interval of each parameter, ensuring a uniform distribution of individuals within the parameter space. In each iteration, based on the current operating condition in the operating status information, elite individuals are selected from a pre-established elite subpopulation corresponding to each operating condition. The position information of these elite individuals guides the current individual's position update direction. Based on the signal non-stationarity in the operating status information, the position update step size and the ratio of global search to local exploration are adaptively adjusted, dynamically adjusting the search range according to signal fluctuations. A multi-objective function fusing prediction bias, false negative rate, false positive rate, computational latency, and resource consumption is used to evaluate the fitness of each individual. The search strategy is switched by controlling the Piecewise chaotic mapping, and the individual position is updated iteratively until the preset number of iterations is reached. The parameter combination corresponding to the individual with the best fitness is taken as the target parameter combination.

[0062] Understandably, since the operating status information reflects the actual operating characteristics of the merging unit at the current moment, using it as the basis for adjusting the preset search strategy links the parameter combination optimization process to the actual operating status of the merging unit, rather than adopting a fixed search method detached from the operating conditions. This allows the obtained target parameter combination to adapt to the current operating conditions, improving the matching degree between the parameter combination and the current operating conditions. Because the target parameter combination matches the current operating conditions, the target prediction model constructed based on this can maintain high prediction accuracy under the current operating conditions, providing accurate input conditions for subsequent sampling time-series closed-loop correction, thereby ensuring the effectiveness of the closed-loop correction.

[0063] Step S40: Construct a target prediction model based on the target parameter combination, and determine the synchronization error prediction value based on the multidimensional error feature time series and the target prediction model.

[0064] It should be noted that the target prediction model refers to a neural network prediction model configured with a combination of target parameters, used to predict the synchronization error value at future times based on historical multidimensional error characteristics. The predicted synchronization error value refers to the estimated synchronization error value of the future sampling window output by the model.

[0065] In one feasible implementation, step S40 may include: inputting the multidimensional error features into the target prediction model for feature extraction to obtain time-dependent features; and determining the synchronization error prediction value based on the time-dependent features.

[0066] It should be noted that time-dependent features refer to the feature representations extracted by the target prediction model from the time series of multi-dimensional error features, which reflect the correlation between various error features at different time steps.

[0067] Specifically, a multi-task LSTM error prediction model is constructed based on the target parameter combination obtained from the gantry optimization algorithm. The multi-task LSTM error prediction model includes a shared temporal feature extraction layer and branches for amplitude error prediction, phase error prediction, and synchronization error prediction. The shared LSTM layer receives multi-dimensional error features corresponding to L consecutive sampling data windows and extracts the temporal dependencies between these error features. The three prediction branches output the amplitude error prediction sequence, phase error prediction sequence, and synchronization error prediction sequence for the next K sampling data windows, respectively.

[0068] The total loss of the model is the weighted sum of the prediction losses of the three types of errors:

[0069] In the formula, , and These are the prediction losses for amplitude error, phase error, and synchronization error, respectively. , , These are the corresponding weighting coefficients.

[0070] Furthermore, the error prediction values ​​of multiple future sampling data windows are compared with their corresponding dynamic warning thresholds. When any error prediction value reaches the first warning threshold, a prompt-level warning is output; when any error prediction value reaches the second warning threshold, or when multiple consecutive prediction windows reach the first warning threshold, an alarm-level warning is output; when any error prediction value reaches the third warning threshold, a severe warning is output.

[0071] Understandably, since the multidimensional error feature time series contains various features related to synchronization error, such as amplitude error, phase error, harmonic distortion rate, and clock jitter index, extracting time-dependent features through the target prediction model can capture the evolution of each error feature at different times and their mutual coupling relationship, thereby improving the accuracy of the synchronization error prediction value. The accurate synchronization error prediction value provides reliable input conditions for subsequent sampling time series closed-loop correction, enabling the correction amount to be generated in advance according to the predicted offset direction and degree and applied to the sampling time series, thereby achieving the active maintenance of the sampling time series.

[0072] Step S50: Perform closed-loop correction on the sampling timing based on the predicted synchronization error value to obtain the corrected sampling timing.

[0073] It should be noted that the corrected sampling timing refers to the sampling timing after frequency and / or phase adjustment, in which synchronization error is suppressed.

[0074] In one feasible implementation, step S50 may include: determining a sampling timing correction amount based on the predicted synchronization error value; adjusting the frequency and / or phase of the sampling timing based on the sampling timing correction amount to obtain the corrected sampling timing.

[0075] It should be noted that sampling timing correction refers to control quantities used to adjust the sampling timing frequency or phase. The sampling timing frequency refers to the oscillation frequency of the sampling clock, which determines the length of the sampling interval; the sampling timing phase refers to the offset of the sampling moment relative to the reference time scale, which determines the absolute position of the sampling moment.

[0076] Specifically, obtain the synchronization error prediction value of the next sampling window output by the target prediction model, and calculate the sampling clock correction amount based on the synchronization error prediction result of the next sampling data window:

[0077] In the formula, This is the sampling clock frequency correction amount. and These are the proportional adjustment coefficient and the integral adjustment coefficient, respectively. This is the predicted synchronization error value for the next sampling window.

[0078] After the correction value is limited, it is sent to the clock control module of the merging unit to adjust the frequency or phase of the sampling clock. After the adjustment is completed, the corrected sampling timing is used as the new sampling timing reference for data acquisition in subsequent sampling cycles.

[0079] Understandably, since the predicted synchronization error reflects the direction and extent of potential deviations in the sampling timing at future moments, using it as the input for closed-loop correction allows the correction action to be applied to the sampling timing in advance, rather than waiting for the error to accumulate to exceed the limit before processing, thus achieving proactive maintenance of the sampling timing. Limiting processing avoids timing jitter caused by excessive correction, ensuring the smoothness of the correction process. By sending the correction amount to the clock control module to perform physical-level frequency or phase adjustment, the correction result can be directly applied to the sampling process of the merging unit, forming a closed-loop structure of "prediction-correction-re-prediction," improving the time-stamp accuracy of the output data of the merging unit.

[0080] This embodiment provides a method for correcting sampling timing errors in a merging unit. By acquiring the sampling timing of the merging unit and collecting sampling window data, multi-dimensional error characteristic timing is determined based on this sampling window data, and operating status information is identified. This incorporates the current operating condition of the merging unit and the degree of signal non-stationarity into parameter optimization. Furthermore, the operating status information is used as the basis for adjusting a preset search strategy, linking the parameter combination optimization process to the actual operating state of the merging unit, rather than using a fixed search method detached from the operating conditions. This yields a target parameter combination that matches the current operating state. Since the target parameter combination can adapt to the current operating conditions, the target prediction model constructed accordingly can maintain high synchronization error prediction accuracy under the current operating conditions. Based on this, closed-loop correction is performed on the sampling timing according to the predicted synchronization error value, allowing the correction amount to be applied to the sampling timing in advance according to the predicted offset direction and degree, forming a "prediction-correction-re-prediction" closed-loop structure. This achieves adaptive matching between the prediction model parameters and the actual operating state of the merging unit, improving the synchronization error prediction accuracy and the effectiveness of sampling timing correction under complex operating conditions.

[0081] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in Embodiment 1 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Step S30 may include steps S31 to S36: Step S31: Generate an initial parameter combination according to a preset search strategy.

[0082] Specifically, the types of parameters to be optimized and their value ranges are determined. Parameter types include filter order, wavelet threshold parameter, number of input history windows, number of hidden layer nodes in the Long Short-Term Memory network, learning rate, batch size, dropout rate, number of prediction windows, amplitude error warning threshold, phase error warning threshold, and synchronization error warning threshold. The signal processing parameters, temporal window parameters, Long Short-Term Memory network model parameters, and error warning parameters are used to construct the gannet individual encoding vector.

[0083] In the formula, Let the filter order be . For wavelet threshold parameters, To input the number of history windows, and This represents the number of hidden nodes in the LSTM layer. For learning rate, For batch size, For the discard rate, To predict the number of windows, , and These are the warning thresholds for amplitude error, phase error, and synchronization error, respectively.

[0084] The initial gannet population was generated using Tent chaotic mapping:

[0085] in, It is a chaotic variable.

[0086] Map the chaotic sequence to the search range of each parameter to be optimized:

[0087] In the formula, and The first The lower and upper limits of each parameter.

[0088] Combine all generated individuals to obtain the initial parameter combination, with each individual corresponding to a set of candidate parameter combinations.

[0089] Understandably, since the types of parameters to be optimized are diverse and their value ranges vary, directly generating initial parameter combinations randomly can easily lead to uneven distribution of individuals in the parameter space. Generating an initial population through Tent chaotic mapping can improve the problem of uneven distribution of individuals in the population caused by random initialization, increase the diversity of the initial population, make the candidate parameter combinations cover the parameter space more evenly, provide a wider initial starting point for subsequent iterative searches, and thus increase the probability of finding the globally optimal parameter combination.

[0090] Step S32: Adjust the preset search strategy according to the running status information to obtain an updated search strategy.

[0091] It should be noted that updating the search strategy refers to adjusting the search direction, search step size, and search mode of the preset search strategy based on the running status information to obtain a search strategy that matches the current running status.

[0092] In one feasible implementation, step S32 may include: obtaining the current operating condition and signal non-stationarity from the operating status information, and obtaining the search direction, search step size, and search mode of the preset search strategy; adjusting the search direction according to the current operating condition; adjusting the search step size and the search mode according to the signal non-stationarity; and obtaining an updated search strategy according to the adjusted search direction, search step size, and search mode.

[0093] It should be noted that the search direction refers to the direction in which the individual moves within the parameter space, guiding it towards a more optimal region. The search step size refers to the distance the individual moves in each step within the parameter space, controlling the magnitude of parameter adjustments. The search pattern refers to the way the individual moves within the parameter space, determining its tendency towards global exploration versus local development.

[0094] Specifically, after the system identifies the current operating condition, it selects elite individuals from the corresponding elite subpopulation and obtains their location information. Based on the difference between the location of the elite individual and the current individual's location, it determines the current individual's position update direction. Simultaneously, it obtains the current signal non-stationarity and calculates the adaptive Lévy flight step size factor based on this non-stationarity.

[0095] In the formula, Let Lévy's flight step size factor be the value corresponding to the t-th iteration. and These are the minimum and maximum values ​​of the step size factor, respectively, where t is the current iteration number and T is the maximum iteration number. and For adjustment coefficients, It is an index of signal non-stationarity.

[0096] In this embodiment, the following is taken , , , When the non-stationarity of the signal increases, the Levy flight step size increases accordingly, allowing the gannet individuals to expand their search range; when the signal returns to stability, the Levy flight step size gradually decreases, enabling the algorithm to perform a local fine search within the current optimal parameter region.

[0097] The adaptive Lévy flight step size factor is used as the search step size for the current iteration. The search mode for the current iteration is determined based on the signal non-stationarity: when the signal non-stationarity increases, the search mode is determined to be a global search mode; when the signal non-stationarity decreases, the search mode is determined to be a local exploration mode. The adjusted search direction, search step size, and search mode are combined to obtain the updated search strategy.

[0098] Understandably, since the current operating conditions reflect the specific operating state of the merging unit, the distribution area of ​​the optimal parameter combination differs under different operating conditions. By selecting elite individuals from the elite subpopulation of the corresponding operating conditions to guide the search direction, we can utilize the excellent parameter information accumulated under similar operating conditions, reduce dependence on a single optimal individual, and lower the risk of premature convergence. Since the non-stationarity of the signal reflects the degree of change of the current signal, it can be used as the basis for adjusting the search step size and search mode, so that the search range is dynamically adjusted with the degree of signal fluctuation. When the signal changes drastically, the search range is expanded to track the changes in operating conditions, and when the signal is stable, the search range is narrowed to improve the optimization accuracy. Thus, the search strategy can adapt to the changes in the operating state of the merging unit and improve the efficiency and accuracy of parameter optimization.

[0099] Step S33: Based on the updated search strategy, perform search optimization on the initial parameter combination to obtain the process parameter combination.

[0100] It should be noted that the process parameter combination refers to the candidate parameter combination corresponding to the individual gannet after position update in the current iteration round, which serves as an intermediate transition result between the initial parameter combination and the final target parameter combination.

[0101] Specifically, during the search phase, a Piecewise chaotic mapping is used to generate position update strategies and select parameters, ensuring a more balanced selection probability for different search strategies. For five operating conditions—normal steady state, harmonic interference, transient impact, temperature drift, and clock anomalies—corresponding elite subpopulations are established. After each iteration, the top 10% of gannets in fitness for each operating condition are saved to the corresponding elite subpopulation. When the system identifies the current operating condition, elite individuals are selected from the corresponding elite subpopulations to guide the current gannet individual in position updates, thereby reducing the population's dependence on a single globally optimal individual. Based on signal non-stationarity, a search development ratio and Lévy flight step size adjustment process are introduced. When signal non-stationarity increases, the global search probability and Lévy flight step size are increased, enabling the gannet population to expand its search range and adapt to rapid changes in operating conditions. When signal non-stationarity decreases, the local development probability is increased and the position update step size is decreased, allowing the algorithm to perform fine-grained searches within the optimal parameter region. The parameter combination corresponding to the updated individual position is used as the process parameter combination.

[0102] Understandably, since the Piecewise chaotic mapping can achieve a relatively balanced selection probability among different search strategies, its control strategy switching can improve the diversity and adaptability of the search; by guiding position updates through the elite subpopulation of operating conditions, it can utilize the excellent parameter information accumulated under similar operating conditions, reduce dependence on a single optimal individual, and reduce the risk of premature convergence; by adaptively adjusting the search step size and search development ratio through signal non-stationarity, the search strategy can be dynamically adjusted according to the changes in the operating state of the merged units, thereby improving the accuracy of local optimization while maintaining global search capability, and obtaining a higher quality combination of process parameters.

[0103] Step S34: Perform a performance evaluation on the combination of process parameters to obtain the performance evaluation results.

[0104] It should be noted that the performance evaluation result refers to the fitness value calculated based on the multi-objective fitness function for the combination of process parameters, which is used to measure the overall performance of the parameter combination in terms of prediction accuracy, alarm performance and operating efficiency.

[0105] Specifically, the prediction performance evaluation value is determined based on the amplitude error prediction deviation, phase error prediction deviation, synchronization error prediction deviation, severe error false alarm rate, and false alarm rate; the online operation constraint evaluation value is determined based on at least one of the single-window computation delay, computational resource consumption, and storage resource consumption of the Long Short-Term Memory Network error prediction model; the prediction performance evaluation value and the online operation constraint evaluation value are weighted to obtain the fitness value of the individual gannet.

[0106] In the formula, , and These are the prediction deviations for amplitude error, phase error, and synchronization error, respectively. The rate of false negatives is a serious error. This represents the false alarm rate under normal conditions. Calculate time for a single window. To the maximum allowable computation time, This refers to the model's resource consumption. The maximum allowed resource usage, To constrain penalty items.

[0107] When the model computation time exceeds the preset real-time requirements, or the model resource consumption exceeds the allowable range of the edge processor, the constraint penalty term increases, preventing the corresponding gannet individual from becoming the final optimal solution. The calculated fitness value is used as the performance evaluation result.

[0108] Understandably, since prediction accuracy is not the only requirement in the online monitoring scenario of the merged unit, the model also needs to meet the real-time calculation requirements under the condition of continuous input of sampled data. The performance of each parameter combination in terms of prediction performance, alarm accuracy and operation efficiency is comprehensively evaluated through a multi-objective fitness function, and a penalty term is used to constrain the calculation latency or resource consumption when the limit is exceeded, so that the final selected parameter combination can meet the dual requirements of prediction accuracy and real-time operation.

[0109] Step S35: Based on the performance evaluation results, determine candidate parameter combinations from the process parameter combinations and the initial parameter combinations.

[0110] It should be noted that the candidate parameter combination refers to the parameter combination with better fitness selected from the process parameter combination and the initial parameter combination based on the performance evaluation results in the current iteration, which will be used as the initial parameter combination for the next iteration.

[0111] Specifically, the performance evaluation results of the process parameter combination and the initial parameter combination are obtained, and their fitness values ​​are compared. If the fitness value of the process parameter combination is better than that of the initial parameter combination, the process parameter combination is selected as a candidate parameter combination. If the fitness value of the initial parameter combination is better than or equal to that of the process parameter combination, the initial parameter combination is selected as a candidate parameter combination. Simultaneously, the globally optimal parameter combination is updated based on the performance evaluation results. If the fitness value of the current candidate parameter combination is better than that of the globally optimal parameter combination, the current candidate parameter combination is selected as the new globally optimal parameter combination.

[0112] Understandably, by comparing the fitness values ​​of the process parameter combination with those of the initial parameter combination, and selecting the parameter combination with better fitness as the basis for the next iteration, the quality of individuals in the population can be gradually improved during the iteration process. This avoids the decline in search efficiency caused by inferior individuals entering the next iteration, thereby accelerating the convergence process and improving the quality of the final parameter combination. At the same time, by synchronously updating the globally optimal parameter combination, it is ensured that the discovered optimal solution is always preserved during the iteration process, preventing the loss of the found high-quality parameter combination due to individual degeneration in subsequent search processes.

[0113] Step S36: Use the candidate parameter combination as the initial parameter combination, return to execute the step of adjusting the preset search strategy according to the running status information until the termination condition is met, and use the candidate parameter combination as the target parameter combination.

[0114] Specifically, the determined candidate parameter combination is used as the initial parameter combination for the next iteration, and the iteration count is incremented by 1. The process returns to adjust the preset search strategy based on the running status information to obtain an updated search strategy. The current running status information is then re-acquired, and the search direction, search step size, and search mode of the preset search strategy are readjusted based on the current running conditions and signal non-stationarity in the current running status information to obtain an updated search strategy. Based on the updated search strategy, the new initial parameter combination is searched and optimized to obtain a new process parameter combination. The performance of the new process parameter combination is evaluated to obtain a new performance evaluation result. Based on the new performance evaluation result, a new candidate parameter combination is determined from the new process parameter combination and the current initial parameter combination. If the fitness value of the new candidate parameter combination is better than the fitness value of the globally optimal parameter combination, the globally optimal parameter combination is updated. This iterative process is repeated until the preset maximum number of iterations is reached. After reaching the maximum number of iterations, the globally optimal parameter combination is output as the target parameter combination.

[0115] Understandably, through multiple iterations, individuals in the population dynamically adjust their search strategies based on operational status information in each iteration, gradually moving closer to a better region; candidate parameter combinations are continuously replaced by better solutions in each round of iteration, so that the quality of parameter combinations is continuously improved during the iteration process; after reaching the preset number of iterations, the output target parameter combination is the optimal fitness parameter combination found during the search process, and using it to build a prediction model can enable the model to maintain high prediction accuracy and real-time performance under the current operating conditions.

[0116] This embodiment provides a method for correcting sampling timing errors in merging units. By employing Tent chaotic mapping to generate an initial population, the candidate parameter combinations are more evenly distributed across the parameter space, providing a broader initial starting point for subsequent iterative searches. Piecewise chaotic mapping controls the switching of search strategies, ensuring a more balanced selection probability for different strategies and improving search diversity and adaptability. By constructing elite subpopulations divided according to operating conditions, the top 10% of individuals in fitness for each condition are saved after each iteration. Upon identifying the current operating condition, elite individuals are selected from the corresponding elite subpopulation to guide the search direction, enabling the algorithm to utilize the superior parameter information accumulated under similar conditions, reducing... By reducing reliance on a single globally optimal individual, the risk of premature convergence is lowered. The search range is dynamically adjusted based on the signal non-stationarity by adaptively adjusting the search step size and search development ratio, allowing for dynamic adjustment of the search range according to signal fluctuations. The search range expands to track changes in operating conditions when the signal changes drastically, and shrinks to improve optimization accuracy when the signal is stable. A multi-objective fitness function comprehensively evaluates the performance of each parameter combination in terms of prediction performance, alarm accuracy, and operational efficiency. Penalty terms are used to constrain computational delays or resource usage when limits are exceeded, ensuring that the final selected target parameter combinations simultaneously meet the dual requirements of prediction accuracy and real-time operation. This further improves the accuracy of merging unit synchronization error prediction and the effectiveness of sampling timing closed-loop correction under complex operating conditions.

[0117] For example, to help understand the implementation flow of the merging unit sampling timing error correction method obtained in this embodiment combined with the above embodiment one, please refer to... Figure 3 , Figure 3 A simplified flowchart of a method for correcting sampling timing errors in merging units is provided, specifically: First, during the data acquisition phase, voltage, current, and synchronization time stamp signals output by the merging unit in the smart substation are acquired. The merging unit uses its own sampling timing as a trigger reference to perform analog-to-digital conversion on these signals, forming a continuous raw sampling sequence. During this process, based on the preset sampling window length and window overlap, the continuous sampling sequence is divided into multiple partially overlapping sampling window data. This partially overlapping division preserves the temporal continuity between adjacent windows, providing a structured data foundation for subsequent feature extraction.

[0118] The feature extraction stage then begins, where data from each sample window is fed into the fundamental frequency processing branch and the broadband processing branch for parallel processing. The fundamental frequency processing branch extracts the amplitude and phase information of the fundamental frequency component in the sampled signal, while the broadband processing branch extracts the harmonic components, transient components, and broadband noise features. After parallel processing by both branches, the fundamental frequency amplitude and phase are compared with the corresponding reference values ​​to determine the amplitude error and phase error. Simultaneously, the synchronization error is determined based on the deviation between the sampling time scale and the synchronization reference time scale. These three errors are then fused with features such as harmonic distortion rate, signal-to-noise ratio, clock jitter index, transient frequency band energy, and device temperature change output from the broadband processing branch to form a multidimensional error feature vector corresponding to the current sampling window. The feature vectors of multiple consecutive windows are then arranged in chronological order to form a multidimensional error feature time sequence. Based on this, the current operating condition of the merging unit is identified according to various features in the multidimensional error characteristic time series. Normal steady state occurs when all features are within the normal range; harmonic distortion rate continuously exceeds the threshold, indicating harmonic interference; transient energy or peak factor increases significantly in a short period, indicating transient impact; temperature drift occurs when temperature changes continuously and error shows a slow drift trend; and clock anomaly occurs when synchronization error or clock jitter continuously increases. Simultaneously, signal non-stationarity is calculated based on the degree of change of each feature between adjacent windows to quantify the severity of signal fluctuations. The identified operating condition and the calculated non-stationarity are combined as the operating status information of the merging unit.

[0119] After obtaining the operational status information, the parameter optimization phase begins. The filter order, wavelet threshold parameters, number of input historical windows, number of hidden layer nodes in the Long Short-Term Memory network, learning rate, batch size, dropout rate, number of prediction windows, and error warning thresholds at each level constitute the parameter combination to be optimized. A chaotic mapping is used to generate an initial population, ensuring a uniform distribution of candidate parameter combinations in the parameter space, with each individual corresponding to a set of candidate parameter combinations. Subsequently, the preset search strategy is adjusted based on the current operational status information. Elite individuals are selected from the elite subpopulation corresponding to the current operational condition to guide the search direction, making the search converge towards the historically excellent parameter region under this condition. Simultaneously, the search step size and search mode are adaptively adjusted according to the magnitude of signal non-stationarity; when non-stationarity increases, the search step size is increased, tending towards global exploration; when non-stationarity decreases, the search step size is decreased, tending towards local fine-tuning. Based on the adjusted strategy, the parameter combination is searched for optimization and its performance is evaluated. The evaluation process comprehensively considers multiple objectives, including amplitude error prediction deviation, phase error prediction deviation, synchronization error prediction deviation, severe error false alarm rate, normal state false alarm rate, model computation latency, and processor resource consumption, to obtain a comprehensive fitness value. Based on the fitness value, the optimal parameter combination is selected and iteratively updated until the preset maximum number of iterations is reached. The parameter combination with the best fitness is then output as the target parameter combination.

[0120] After parameter optimization, the error prediction stage begins. A target prediction model based on a Long Short-Term Memory (LSTM) network is constructed using the structural and training parameters from the target parameter combination. The previously determined multi-dimensional error features are then input into this model. The model first models the temporal dependencies between sampling windows through a shared temporal feature extraction layer, obtaining temporal dependency features. These features are then input into the amplitude error prediction branch, phase error prediction branch, and synchronization error prediction branch, respectively. Each branch outputs the corresponding error prediction sequences for multiple future sampling windows, providing not only an error estimate for the next time step but also reflecting the error's trend over a future period.

[0121] Finally, the closed-loop correction stage begins. On one hand, the predicted values ​​of various errors are compared with the corresponding warning thresholds. When the predicted value reaches the threshold of different levels, the corresponding level of warning information is triggered. On the other hand, the predicted value of the synchronization error for the next sampling window is extracted from the output of the synchronization error prediction branch. The frequency correction amount of the sampling timing is determined through proportional-integral adjustment, and after amplitude limiting, it is converted into a clock control word and sent to the clock control module of the merging unit. The clock control module adjusts the frequency or phase of the sampling timing to make the adjusted sampling timing approach the ideal timing. The corrected sampling timing serves as a new sampling reference for data acquisition in subsequent sampling cycles. Thus, a complete closed-loop feedback structure is formed from data acquisition, feature extraction, parameter optimization, error prediction to closed-loop correction, ensuring that the sampling timing remains accurate under the action of correction.

[0122] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the sampling timing error correction method of the merging unit in this application. Any simple transformations based on this technical concept are within the protection scope of this application.

[0123] This application also provides a device for correcting sampling timing errors in merging units. Please refer to [reference needed]. Figure 4 The merging unit sampling timing error correction device includes: Data sampling module 10 is used to acquire the sampling timing of the merging unit in the smart substation, and to collect sampling window data based on the sampling timing. The feature extraction module 20 is used to determine the multidimensional error feature time series based on the sampling window data, and to determine the running status information of the merging unit according to the multidimensional error feature time series; The parameter optimization module 30 is used to determine an initial parameter combination based on a preset search strategy, and to perform search optimization on the initial parameter combination using the preset search strategy and the running status information to obtain a target parameter combination. Error prediction module 40 is used to construct a target prediction model based on the target parameter combination, and to determine the synchronous error prediction value based on the multidimensional error feature time series and the target prediction model; The timing correction module 50 is used to perform closed-loop correction on the sampling timing based on the predicted synchronization error value to obtain the corrected sampling timing.

[0124] The merging unit sampling timing error correction device provided in this application, employing the merging unit sampling timing error correction method in the above embodiments, can solve the technical problem of insufficient error prediction accuracy under complex operating conditions due to the disconnect between the parameter optimization process of the merging unit error prediction model and the actual operating state of the equipment. Compared with the prior art, the beneficial effects of the merging unit sampling timing error correction device provided in this application are the same as those of the merging unit sampling timing error correction method provided in the above embodiments, and other technical features in the merging unit sampling timing error correction device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0125] In one embodiment, the parameter optimization module 30 is further configured to generate an initial parameter combination according to a preset search strategy; The preset search strategy is adjusted based on the operating status information to obtain an updated search strategy; Based on the updated search strategy, the initial parameter combination is optimized to obtain the process parameter combination; The performance of the process parameter combination is evaluated to obtain the performance evaluation results; Based on the performance evaluation results, candidate parameter combinations are determined from the process parameter combinations and the initial parameter combinations; The candidate parameter combination is used as the initial parameter combination, and the process returns to the step of adjusting the preset search strategy according to the running status information until the termination condition is met, and the candidate parameter combination is used as the target parameter combination.

[0126] In one embodiment, the parameter optimization module 30 is further configured to obtain the current operating condition and signal non-stationarity in the operating status information, and to obtain the search direction, search step size and search mode of the preset search strategy; Adjust the search direction based on the current operating conditions; Adjust the search step size and the search mode according to the non-stationarity of the signal; An updated search strategy is obtained based on the adjusted search direction, search step size, and search pattern.

[0127] In one embodiment, the feature extraction module 20 is further configured to determine fundamental frequency features and broadband features based on the sampling window data; Based on the fundamental wave characteristics and the broadband characteristics, a multidimensional error feature time series is generated. Based on the multidimensional error characteristic time series, the current operating condition and signal non-stationarity of the merging unit are determined; The current operating conditions and the signal non-stationarity are used as the operating status information of the merging unit.

[0128] In one embodiment, the error prediction module 40 is further configured to input the multidimensional error features into the target prediction model in a time series for feature extraction to obtain time-dependent features; Based on the time-dependent characteristics, the predicted value of the synchronization error is determined.

[0129] In one embodiment, the timing correction module 50 is further configured to determine the sampling timing correction amount based on the synchronization error prediction value; Based on the sampling timing correction amount, the frequency and / or phase of the sampling timing are adjusted to obtain the corrected sampling timing.

[0130] In one embodiment, the data sampling module 10 is further configured to acquire the sampling timing of the merging unit in the smart substation, the preset sampling window length, and the window overlap. The voltage and current signals output by the merging unit are sampled according to the sampling timing to obtain a continuous sampling sequence; Based on the sampling window length and the window overlap, the continuous sampling sequence is divided into multiple partially overlapping sampling window data.

[0131] This application provides a merging unit sampling timing error correction device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the merging unit sampling timing error correction method in the above embodiment 1.

[0132] The following is for reference. Figure 5This document illustrates a structural schematic diagram suitable for implementing a merging unit sampling timing error correction device in the embodiments of this application. The merging unit sampling timing error correction device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The sampling timing error correction device for the merging unit shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0133] like Figure 5 As shown, the merging unit sampling timing error correction device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in ROM (Read Only Memory) 1002 or a program loaded from storage device 1003 into RAM (Random Access Memory) 1004. RAM 1004 also stores various programs and data required for the operation of the merging unit sampling timing error correction device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, LCDs (Liquid Crystal Displays), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the merging unit sampling timing error correction device to communicate wirelessly or wiredly with other devices to exchange data. Although merging unit sampling timing error correction devices with various systems are shown in the figures, it should be understood that it is not required to implement or possess all of the systems shown. More or fewer systems can be implemented alternatively.

[0134] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0135] The merging unit sampling timing error correction device provided in this application, employing the merging unit sampling timing error correction method in the above embodiments, can solve the technical problem of insufficient error prediction accuracy under complex operating conditions due to the disconnect between the parameter optimization process of the merging unit error prediction model and the actual operating state of the equipment. Compared with the prior art, the beneficial effects of the merging unit sampling timing error correction device provided in this application are the same as those of the merging unit sampling timing error correction method provided in the above embodiments, and other technical features in this merging unit sampling timing error correction device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0136] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0137] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0138] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the merging unit sampling timing error correction method in the above embodiments.

[0139] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory or Flash Memory), optical fibers, CD-ROM (CD-Read Only Memory), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0140] The aforementioned computer-readable storage medium may be included in the merging unit sampling timing error correction device; or it may exist independently and not be assembled into the merging unit sampling timing error correction device.

[0141] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the merging unit sampling timing error correction device, the merging unit sampling timing error correction device performs the following actions: acquires the sampling timing of the merging units in the intelligent substation, and obtains sampling window data by collecting the sampling timing data; determines multi-dimensional error characteristic timing based on the sampling window data, and determines the operating status information of the merging units based on the multi-dimensional error characteristic timing; determines an initial parameter combination based on a preset search strategy, and optimizes the initial parameter combination using the preset search strategy and operating status information to obtain a target parameter combination; constructs a target prediction model based on the target parameter combination, and determines the synchronization error prediction value based on the multi-dimensional error characteristic timing and the target prediction model; and performs closed-loop correction on the sampling timing based on the synchronization error prediction value to obtain the corrected sampling timing.

[0142] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including LAN (Local Area Network) or WAN (Wide Area Network)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0143] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0144] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0145] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described merging unit sampling timing error correction method. This solves the technical problem of insufficient error prediction accuracy under complex operating conditions due to the disconnect between the parameter optimization process of the merging unit error prediction model and the actual operating state of the equipment. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the merging unit sampling timing error correction method provided in the above embodiments, and will not be repeated here.

[0146] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the merging unit sampling timing error correction method described above.

[0147] The computer program product provided in this application can solve the technical problem that the parameter optimization process of the merging unit error prediction model is disconnected from the actual operating state of the equipment, resulting in insufficient error prediction accuracy under complex operating conditions. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the merging unit sampling timing error correction method provided in the above embodiments, and will not be repeated here.

[0148] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for correcting sampling timing errors in a merging unit, characterized in that, The method includes: The sampling timing sequence of the merging unit in the intelligent substation is obtained, and sampling window data is obtained by collecting data based on the sampling timing sequence. The multidimensional error feature timing is determined based on the sampling window data, and the operating status information of the merging unit is determined based on the multidimensional error feature timing. An initial parameter combination is determined based on a preset search strategy, and the initial parameter combination is searched and optimized using the preset search strategy and the running status information to obtain a target parameter combination. A target prediction model is constructed based on the target parameter combination, and the synchronization error prediction value is determined based on the multidimensional error feature time series and the target prediction model. The sampling timing is corrected by performing closed-loop correction based on the predicted synchronization error value to obtain the corrected sampling timing.

2. The method as described in claim 1, characterized in that, The process of determining an initial parameter combination based on a preset search strategy, and then optimizing the initial parameter combination using the preset search strategy and the running status information to obtain a target parameter combination, includes: Generate initial parameter combinations based on a preset search strategy; The preset search strategy is adjusted based on the running status information to obtain an updated search strategy; Based on the updated search strategy, the initial parameter combination is optimized to obtain the process parameter combination; The performance of the process parameter combination is evaluated to obtain the performance evaluation results; Based on the performance evaluation results, candidate parameter combinations are determined from the process parameter combinations and the initial parameter combinations; The candidate parameter combination is used as the initial parameter combination, and the process returns to the step of adjusting the preset search strategy according to the running status information until the termination condition is met, and the candidate parameter combination is used as the target parameter combination.

3. The method as described in claim 2, characterized in that, The step of adjusting the preset search strategy based on the running status information to obtain an updated search strategy includes: The current operating condition and signal non-stationarity in the operating status information are obtained, and the search direction, search step size and search mode of the preset search strategy are obtained. Adjust the search direction based on the current operating conditions; Adjust the search step size and the search mode according to the non-stationarity of the signal; An updated search strategy is obtained based on the adjusted search direction, search step size, and search pattern.

4. The method as described in claim 1, characterized in that, The step of determining the multidimensional error feature time series based on the sampling window data, and determining the operating status information of the merging unit based on the multidimensional error feature time series, includes: Based on the sampling window data, the fundamental frequency characteristics and broadband characteristics are determined; Based on the fundamental wave characteristics and the broadband characteristics, a multidimensional error feature time series is generated. Based on the multidimensional error characteristic time series, the current operating condition and signal non-stationarity of the merging unit are determined; The current operating conditions and the signal non-stationarity are used as the operating status information of the merging unit.

5. The method as described in claim 1, characterized in that, Based on the multidimensional error feature time series and the target prediction model, the synchronization error prediction value is determined, including: The multidimensional error features are input into the target prediction model in time series for feature extraction to obtain time-dependent features; Based on the time-dependent characteristics, the predicted value of the synchronization error is determined.

6. The method as described in claim 1, characterized in that, The step of performing closed-loop correction on the sampling timing based on the predicted synchronization error value to obtain the corrected sampling timing includes: Based on the predicted synchronization error value, determine the sampling timing correction amount; Based on the sampling timing correction amount, the frequency and / or phase of the sampling timing are adjusted to obtain the corrected sampling timing.

7. The method as described in claim 1, characterized in that, The step of acquiring the sampling timing sequence of the merging unit in the intelligent substation, and collecting sampling window data using the sampling timing sequence, includes: Obtain the sampling timing, preset sampling window length, and window overlap of the merging unit in the smart substation; The voltage and current signals output by the merging unit are sampled according to the sampling timing to obtain a continuous sampling sequence; Based on the sampling window length and the window overlap, the continuous sampling sequence is divided into multiple partially overlapping sampling window data.

8. A device for correcting sampling timing errors in a merging unit, characterized in that, The device includes: The data sampling module is used to obtain the sampling timing of the merging unit in the smart substation, and to collect sampling window data based on the sampling timing. The feature extraction module is used to determine the multidimensional error feature time series based on the sampling window data, and to determine the running status information of the merging unit based on the multidimensional error feature time series; The parameter optimization module is used to determine the initial parameter combination based on a preset search strategy, and to optimize the initial parameter combination by the preset search strategy and the running status information to obtain the target parameter combination. The error prediction module is used to construct a target prediction model based on the target parameter combination, and to determine the synchronous error prediction value based on the multidimensional error feature time series and the target prediction model. The timing correction module is used to perform closed-loop correction on the sampling timing based on the predicted synchronization error value to obtain the corrected sampling timing.

9. A device for correcting sampling timing errors in a merging unit, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the merging unit sampling timing error correction method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the merging unit sampling timing error correction method as described in any one of claims 1 to 7.

Images