A power distribution network harmonic level estimation method and system

Abstract

The application belongs to the technical field of power distribution networks, and provides a power distribution network harmonic level estimation method and system, which comprises: obtaining power grid real-time operation data; constructing a power distribution network harmonic feature data set based on the obtained real-time operation data; obtaining a harmonic level estimation result at the current time according to the constructed harmonic feature data set and a harmonic level estimation model, to complete the estimation of the harmonic level of the power distribution network; wherein the harmonic level estimation model adopts a long short-term memory network, the long short-term memory network combines short-term memory and long-term memory through gate control, extracts the time sequence correlation information between the constructed harmonic feature data sets, fits the nonlinear relationship between the harmonic feature data based on the extracted time sequence correlation information, and completes the estimation of the harmonic level according to the fitting result.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network technology, and specifically relates to a method and system for estimating harmonic levels in power distribution networks. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] With the widespread integration of power electronic equipment, various problems caused by harmonic pollution, such as additional energy losses, overheating, equipment failures, and accelerated aging, are becoming increasingly prominent, making the demand for harmonic mitigation by power grids and electricity users more urgent. Distribution networks have a large number of nodes and complex network topologies. Due to economic and technical constraints, power quality monitors (PQMs) cannot provide comprehensive coverage of the power system, making it difficult to achieve observable harmonic status across the entire network. Therefore, there is an urgent need for a harmonic level estimation method that is less dependent on PQM devices, replacing "device monitoring" with "computational estimation" to achieve observable harmonic status.

[0004] According to the inventors, current assessments of harmonic levels in distribution networks mostly employ harmonic power flow calculations or harmonic state estimations. However, the accuracy of harmonic power flow calculations is affected by the harmonic admittance matrix and the emission level of harmonic sources. Establishing the node admittance matrix requires accurate harmonic modeling of each component in the system. With the integration of multiple power electronic devices in new power systems, accurate modeling of massive heterogeneous devices becomes increasingly difficult. The integration of distributed renewable energy sources brings more uncertainty to the system power flow, posing a significant challenge to the accuracy of harmonic power flow calculations. Traditional harmonic state estimation is conducted under the premise of knowing the accurate grid topology, estimating the system harmonic impedance through harmonic monitoring data of some known nodes, and then estimating the harmonic state of the entire network. Due to the low coverage of PQM and the dynamic time-varying network structure, traditional methods are difficult to meet the observability requirements of the required data, making them difficult to apply in practical systems. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a method and system for estimating the harmonic level of a distribution network. Based on real-time power grid operation data and a harmonic level estimation model, the method effectively and accurately estimates the current harmonic level of the distribution network.

[0006] According to some embodiments, the first aspect of the present invention provides a method for estimating the harmonic level of a distribution network, employing the following technical solution: A method for estimating harmonic levels in a distribution network includes: Obtain real-time power grid operation data; Based on the acquired real-time operational data, a harmonic characteristic dataset for the power distribution network is constructed. Based on the constructed harmonic characteristic dataset and harmonic level estimation model, the harmonic level estimation result at the current moment is obtained, and the harmonic level of the distribution network is estimated. The harmonic level estimation model employs a long short-term memory network, which combines short-term and long-term memory through gating to extract temporal correlation information between the constructed harmonic feature datasets. Based on the extracted temporal correlation information, the nonlinear relationship between the harmonic feature data is fitted, and the harmonic level is estimated based on the fitting result.

[0007] As a further technical limitation, in the process of harmonic level estimation, the harmonic influence feature data matrix is ​​input into the harmonic level estimation model based on long short-term memory network to obtain the harmonic emission feature data matrix and the harmonic distortion limit exceedance data matrix used to determine the harmonic level of the harmonic source. The evaluation index of harmonic level estimation is calculated based on the obtained harmonic emission feature data matrix and the harmonic distortion limit exceedance data matrix, and the performance of harmonic level estimation is evaluated based on the obtained evaluation index.

[0008] As a further technical limitation, the constructed distribution network harmonic characteristic dataset includes harmonic influence characteristic data, harmonic emission characteristic data, and harmonic distortion exceeding limit characteristic data; wherein, the harmonic influence characteristic data includes at least the effective value of current, the effective value of voltage, active power, and reactive power; the harmonic emission characteristic data includes at least the total harmonic current and the total harmonic voltage; and the harmonic distortion exceeding limit characteristic data includes at least the total harmonic distortion rate of voltage, the principal component content of harmonic voltage, and the principal component of harmonic current.

[0009] As a further technical limitation, the real-time power grid operation data obtained includes at least distribution network voltage, distribution network current, distribution network power, and continuous data of harmonic source grid connection points, including harmonic source timing operation and harmonic disturbance data.

[0010] As a further technical limitation, the network structure of the harmonic level estimation model includes an input layer, an output layer, and at least one long short-term memory layer; wherein, the input layer is used to receive the harmonic correlation feature data matrix at time t. , for Harmonic influence characteristic data matrix at time point for Harmonic emission characteristic data matrix at time t, for The harmonic distortion over-limit feature data matrix at time t; each of the long short-term memory layers includes a forget gate, an input gate, an output gate, and a cell state; the output layer is used to output the harmonic emission feature data matrix estimated at time t. Harmonic distortion exceeding limit characteristic data matrix .

[0011] As a further technical limitation, in the process of fitting the nonlinear relationship between harmonic characteristic data, the root mean square error loss function and the mean absolute error loss function are used for repeated iterative training to improve the fitting accuracy.

[0012] According to some embodiments, the second aspect of the present invention provides a distribution network harmonic level estimation system, which adopts the following technical solution: A system for estimating the harmonic level of a power distribution network, comprising: The acquisition module is configured to acquire real-time operating data of the power grid; The building module is configured to construct a distribution network harmonic characteristic dataset based on the acquired real-time operational data; The estimation module is configured to obtain the harmonic level estimation result at the current moment based on the constructed harmonic feature dataset and harmonic level estimation model, thereby completing the estimation of the harmonic level of the distribution network. The harmonic level estimation model employs a long short-term memory network, which combines short-term and long-term memory through gating to extract temporal correlation information between the constructed harmonic feature datasets. Based on the extracted temporal correlation information, the nonlinear relationship between the harmonic feature data is fitted, and the harmonic level is estimated based on the fitting result.

[0013] According to some embodiments, a third aspect of the present invention provides a computer-readable storage medium, employing the following technical solution: A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in a method for estimating the harmonic level of a distribution network as described in the first aspect of the present invention.

[0014] According to some embodiments, the fourth aspect of the present invention provides an electronic device, which adopts the following technical solution: An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps in a method for estimating the harmonic level of a power distribution network as described in the first aspect of the present invention.

[0015] According to some embodiments, the fifth aspect of the present invention provides a computer program product, which adopts the following technical solution: A computer program product includes software code, wherein the program in the software code performs the steps of a method for estimating the harmonic level of a distribution network as described in the first aspect of the present invention.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention fully utilizes the widely deployed smart meters and electricity information collection systems to obtain real-time power grid operation data, eliminating the need for additional fixed power quality monitoring devices. It can achieve harmonic level estimation of the distribution network by combining with a harmonic level estimation model. It does not require prior information such as the distribution network topology, line parameters, and harmonic impedance, thus overcoming the dependence of traditional harmonic power flow calculation and harmonic state estimation methods on accurate modeling of system parameters. It is suitable for distribution networks with dynamically changing topologies. Attached Figure Description

[0017] The accompanying drawings, which form part of this embodiment, are used to provide a further understanding of this embodiment. The illustrative embodiments and their descriptions are used to explain this embodiment and do not constitute an improper limitation of this embodiment.

[0018] Figure 1 This is a flowchart of a method for estimating the harmonic level of a distribution network according to Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the LSTM neuron structure in Embodiment 1 of the present invention; Figure 3 This is an architecture diagram of the LSTM-based harmonic level estimation model in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of the harmonic level estimation process based on LSTM in Embodiment 1 of the present invention; Figure 5 This is a schematic diagram of the amplitude of each harmonic current and voltage in Embodiment 1 of the present invention; Figure 6 This is a schematic diagram of the total harmonic current estimation results in Embodiment 1 of the present invention; Figure 7 This is a schematic diagram of the total harmonic voltage estimation results in Embodiment 1 of the present invention; Figure 8 This is a schematic diagram of the voltage total harmonic distortion rate estimation results in Embodiment 1 of the present invention; Figure 9 This is a schematic diagram of the estimation results of the 5th harmonic voltage content in Embodiment 1 of the present invention; Figure 10 This is a schematic diagram of the 5th harmonic current estimation results in Embodiment 1 of the present invention; Figure 11 This is a comparative diagram of the total harmonic current estimation results in Embodiment 1 of the present invention; Figure 12 This is a comparative schematic diagram of the total harmonic voltage estimation results in Embodiment 1 of the present invention; Figure 13 This is a comparative diagram of the voltage total harmonic distortion rate estimation results in Embodiment 1 of the present invention; Figure 14This is a comparative diagram of the estimation results of the fifth harmonic voltage content in Embodiment 1 of the present invention; Figure 15 This is a comparative diagram of the estimation results of the 5th harmonic current in Embodiment 1 of the present invention; Figure 16 This is a structural block diagram of a power distribution network harmonic level estimation system according to Embodiment 2 of the present invention. Detailed Implementation

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0020] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0022] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0023] In this invention, terms such as "fixed connection," "connected," and "linked" should be interpreted broadly, indicating a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can determine the specific meaning of these terms in this invention based on the specific circumstances, and they should not be construed as limitations on the invention.

[0024] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0025] Example 1 Embodiment 1 of this invention introduces a method for estimating the harmonic level of a power distribution network.

[0026] like Figure 1 The method for estimating the harmonic level of a distribution network, as shown, includes: Obtain real-time power grid operation data; Based on the acquired real-time operational data, a harmonic characteristic dataset for the power distribution network is constructed. Based on the constructed harmonic characteristic dataset and harmonic level estimation model, the harmonic level estimation result at the current moment is obtained, and the harmonic level of the distribution network is estimated. The harmonic level estimation model employs a long short-term memory network, which combines short-term and long-term memory through gating to extract temporal correlation information between the constructed harmonic feature datasets. Based on the extracted temporal correlation information, the nonlinear relationship between the harmonic feature data is fitted, and the harmonic level is estimated based on the fitting result.

[0027] This embodiment estimates the harmonic level of the distribution network based on user operational data. It utilizes a long short-term memory (LSTM) network to evaluate the harmonic emission level corresponding to the basic operating state of harmonic sources. Specifically: First, based on harmonic data and operational data monitored by portable power quality monitors (PPQMs) and smart meters, a harmonic relevant characteristic data (HRCD) set is constructed. Second, based on the LSTM-based multivariate time-series data coupling relationship characterization method, a nonlinear relationship is fitted to the HRCD to construct an LSTM-based harmonic level estimation model. Finally, based on the real-time monitoring and uploaded operational data from smart meters, the harmonic level of the current user's grid connection point is estimated. In areas where smart meters have been deployed, harmonic emission level sensing of harmonic sources can be achieved with zero-cost additional hardware deployment. This embodiment effectively estimates the user's current harmonic level based on real-time grid operational data, exhibiting high accuracy and stability.

[0028] This embodiment uses a portable power quality monitoring device to continuously sample data from the grid connection points of harmonic sources in the distribution network, obtaining the time-series operation and harmonic disturbance data of the harmonic sources. Smart meters and electricity consumption information collection systems also store a large amount of multi-dimensional conventional electricity consumption data, such as voltage, current, and power, which contains the variation patterns of the system's harmonic distortion level and is closely related to the harmonic state of the harmonic sources. Therefore, the aforementioned multi-dimensional operational data and harmonic disturbance data are collectively referred to as HRCD and used as the input data matrix to train the neural network. First, the multi-dimensional data is processed to construct the HRCD dataset.

[0029] In this embodiment, the harmonic data that the PPQM can collect includes the amplitude of each harmonic current and voltage, harmonic distortion rate, etc. A typical PPQM can measure the 2nd to 50th harmonic current and voltage. To reduce the amount of input and output data for the harmonic level estimation model and facilitate training and calculation, the measured harmonic electrical quantities are first converted into total harmonic electrical quantities. The total harmonic current and total harmonic voltage can be respectively... and ;in, , These are the total harmonic current and total harmonic voltage at the monitoring point, respectively. , The monitoring points are respectively the first The second harmonic current and the first Subharmonic voltage; The maximum harmonic order under consideration.

[0030] The total harmonic current and voltage can be used to understand the harmonic emission level of a harmonic source, hence the term harmonic emission characteristic data. However, the emission level of harmonic current or voltage alone cannot quickly determine whether the harmonic distortion at a node exceeds the limit; the harmonic distortion rate at the monitoring point also needs to be known.

[0031] Since harmonic voltage is almost always relative to the fundamental voltage, and its fluctuation range is generally small; while current fluctuation is more random, even a small-amplitude harmonic current can cause a large total harmonic distortion (THD). Therefore, this embodiment selects voltage parameters as the criterion for judging harmonic distortion exceeding the limit, and adopts the national standard for the limit of voltage total harmonic distortion for the judgment of exceeding the limit; voltage total harmonic distortion for Where U1 is the effective value of the fundamental voltage at the monitoring point.

[0032] The total harmonic distortion rate of voltage reflects the degree of harmonic distortion at a node. By comparing its value with the limit specified in the standard, it is possible to quickly determine whether the harmonic distortion at the monitoring point is qualified. Therefore, it is called the harmonic distortion over-limit characteristic data.

[0033] National standards also specify limits for the content of single-harmonic voltage and the amplitude of single-harmonic current, therefore, single-harmonic voltage and current also need to be considered. Considering computational resources and application requirements, in actual operation, it is preferable to conduct a rough harmonic level assessment based on low-cost hardware deployment and low computational overhead. Therefore, for single-harmonic voltage and current, this embodiment only analyzes their principal components. Based on the monitored harmonic voltage and current spectra, the principal components of the harmonic voltage at the monitored point can be obtained. Harmonic current principal components Harmonic voltage principal component content for .

[0034] Single harmonic current and voltage are also among the power quality assessment indicators; therefore, the principal components of harmonic voltage are considered. Harmonic current principal components Also known as harmonic distortion over-limit characteristic data.

[0035] The operating state and conditions of a harmonic source directly affect its harmonic emission level. Under different operating conditions, the basic electrical quantities of a harmonic source, such as voltage, current, and power, will vary. Therefore, by determining the basic electrical quantities under typical (or specific) operating conditions, the corresponding harmonic emission level under different basic electrical quantity conditions can be determined. The nonlinear relationship between these basic electrical quantities and the harmonic level of the harmonic source can be fitted using a neural network. Easily available multivariate operating data such as voltage, current, and power are referred to as harmonic influence characteristic data, and the harmonic level is assessed based on this data after the neural network is trained.

[0036] According to their relationship with the harmonic level of the monitoring point, HRCD can be divided into three categories: harmonic influence characteristic data, harmonic emission characteristic data, and harmonic distortion exceeding the limit characteristic data. They together constitute the HRCD set shown in Table 1.

[0037] Table 1 HRCD Set

[0038] Harmonic distortion (HDC) is a waveform distortion problem in the steady-state domain of power quality, often exhibiting periodic and temporal characteristics. Its time-varying properties are correlated with the temporal variations of voltage, current, and power, thus revealing crucial temporal features hidden within the HRCD. Recurrent neural networks (RNNs) are highly effective at processing time-series data, capable of extracting temporal and semantic information and retaining some memory of processed data. However, as time progresses and the number of network layers increases, traditional RNNs are prone to problems such as vanishing or exploding gradients, limiting their memory to short-term. LSTM networks, through gating, combine short-term and long-term memory, enabling them to retain data for longer periods and mitigating the vanishing gradient problem to some extent. Therefore, this embodiment selects an LSTM network to extract the temporal correlation characteristics between multivariate HRCDs.

[0039] LSTM networks consist of several such Figure 2 The LSTM neurons shown are composed of one hidden state. And 3 doors (input doors) Output gate And the Gate of Oblivion ).

[0040] Forgotten Gate Used to control the cell state at the previous moment. How much needs to be forgotten, i.e. ;in, It is the sigmoid activation function; The weight of the forgetting gate; This is the hidden state from the previous moment; For the current input; This is an offset for the forget gate.

[0041] The input gate consists of two parts. The first part uses the sigmoid activation function, and the output is... The second part uses the tanh hyperbolic tangent activation function, and the output is... The results of the two are multiplied together to update the cell state, i.e. ; ; in, The weights of the input gates; For the input gate bias; Weights for cell states; This is a bias for the cell state.

[0042] The forget gate and the input gate work together to influence the current cell state. ,Right now ;in, The Hadamard product represents the multiplication of elements at corresponding positions in a matrix.

[0043] Output gate Determine the next hidden state and use it for prediction; and current input The modified cell state is passed to the sigmoid function, then passed to the tanh function, and finally the current hidden state is output. and the current cell state And the hidden state is moved to the next time series, i.e. ; ; in, The weights of the output gates; This is the bias of the output gate.

[0044] This embodiment uses, as follows: Figure 3 The model shown is a harmonic level estimation model based on LSTM; where, for t The HRCD matrix input at each time step; for t The harmonic level matrix obtained from the time-matrix model estimation; Y This is the true value label matrix, which is the historical harmonic level matrix of actual measurements.

[0045] In this embodiment, the HRCD set is used as the initial feature matrix x, and harmonic level data is selected as the true value label matrix Y for model training: ; ; ; ; in, , , , respectively, are the harmonic influence, emission, and distortion exceeding the limit characteristic data matrix at time t; y is the harmonic level estimation matrix for the entire monitoring period; , Let t be the emission characteristic matrix estimated by the model at time t; , These are the actual harmonic distortion over-limit characteristic matrices at time t.

[0046] In this embodiment, to minimize the error between the estimated result and the real data and to prevent overfitting, a combined loss function is used during training. Adjustment, that is ;in, The root mean square error loss function is used to reduce the error between the estimated result and the true value. ; The loss function, which incorporates L2 regularization, can simulate random noise and prevent overfitting. ; This is a hyperparameter used to control the strength of regularization; for t The matrix of true values ​​at each time step; for t The harmonic level matrix obtained from the time-matrix model estimation; This represents the total sample time length. for t The weighted sparse vector at time step.

[0047] The initial feature matrix x is processed by an LSTM network to extract temporal correlation information, fitting the nonlinear relationship between multivariate HRCDs, and iteratively trained using a loss function to improve the fitting accuracy. This embodiment uses the following commonly used evaluation metrics to assess the performance of the proposed method: 1) Root mean square error (RMSE). ; 2) Mean absolute error (MAE), i.e. ; 3) Coefficient of determination, i.e. ; in, , The root mean square error loss function and the mean absolute error loss function should be as small as possible; The coefficient of determination reflects the goodness of fit of the model; the closer its value is to 1, the better the fit. The total number of samples; , The first The true and estimated values ​​of each sample; This represents the average value of the sample.

[0048] This embodiment uses Power Quality Monitoring (PPQM) to monitor harmonic sources for a period of time, aiming to understand the harmonic emission levels under typical operating conditions. The required monitoring time varies depending on the type of harmonic source. For traditional large-scale harmonic sources such as factories, harmonic characteristics do not change significantly over time, and a shorter monitoring period is sufficient to obtain representative data. In this case, a monitoring period of one week to one month can reflect its harmonic characteristics throughout the year. If the factory's production has cyclical variations, at least one complete production cycle needs to be monitored to observe the harmonic performance under different production loads. The longer the production cycle, the longer the monitoring time should be. For distributed photovoltaic, wind power, and other new energy sources, their operating conditions have strong regularity and are closely related to the natural environment. Therefore, typical weather conditions, such as sunny days and rainy days, can be selected to conduct power quality monitoring during key periods or throughout the day to understand the correlation between harmonics and operating conditions under typical weather conditions.

[0049] This embodiment combines the voltage, current, power, and other operating data monitored and uploaded by the smart meter and user information collection system with the harmonic data monitored by PPQM to form harmonic-related feature data, and forms a time-series HRCD set, which serves as the input data for the neural network.

[0050] An LSTM-based harmonic level estimation model was trained to construct a mapping relationship between harmonic levels and operational data under typical operating conditions. Various learning parameters of the neural network were appropriately set, and all data in the HRCD dataset were divided into training and test sets. After training on the training set, the model's estimation accuracy was validated using data from the test set. Training ended when the model's estimation accuracy reached a certain requirement.

[0051] After the neural network is trained, the harmonic levels under corresponding operating conditions can be estimated using the real-time operating data uploaded by the smart meter. The harmonic level estimation process based on LSTM is as follows: Figure 4 As shown.

[0052] Smart meters monitor and record users' voltage, current, and power data in real time, which is the harmonic influence feature data matrix A required for the harmonic level estimation model based on LSTM. This data is then uploaded through the electricity information collection system. Matrix A is input into the harmonic level estimation model, and finally, the model outputs the harmonic emission feature data matrix E and the harmonic distortion exceeding the limit feature data matrix O. This allows for timely understanding of the harmonic level of the harmonic source and provides data support for harmonic mitigation.

[0053] The characteristics of the measuring device are shown in Table 2. Since harmonic distortion is a steady-state power quality problem in the power system, the harmonic level generally remains relatively stable within 15 minutes. Therefore, selecting one set of data from the smart meter to be uploaded at 15-minute intervals can meet the requirements for real-time harmonic assessment.

[0054] Table 2 Characteristics of the measuring device

[0055] Taking industrial users as an example, the accuracy and limitations of the harmonic level estimation method proposed in this embodiment are verified.

[0056] This embodiment selects a factory with relatively fixed load and production status, and whose harmonic characteristics are relatively stable and do not change significantly over time for analysis. For this type of traditional large harmonic source, a monitoring period of one week to one month is sufficient to reflect its harmonic characteristics throughout the year. In order to understand the correlation between the factory's harmonic levels and operating data under various operating conditions, PPQM was used to sample data continuously for 30 days on the low-voltage side of the plant's auxiliary transformer, and a case study analysis was carried out.

[0057] This embodiment performs spectral analysis on the monitored harmonic data to obtain the following results: Figure 5 The amplitudes of each harmonic current and voltage are shown; it can be seen that the amplitudes of the 5th harmonic current and voltage are the largest, which are the main components of the harmonic current and voltage, the main contributing factors to harmonic distortion, and the most likely to exceed the standard. Therefore, they need to be given special attention.

[0058] The PPQM sampling interval was 1 minute. Power quality monitoring was conducted on the factory for 30 consecutive days, resulting in 9 sets of 1440×30 data points, which constituted the HRCD set, i.e., the input matrix of the LSTM harmonic level estimation model. x The dimension is 43200×9, that is ; in, , , , , , , , and All t The monitoring values ​​of each type of HRCD data at each time point are respectively t The current, voltage, active power, reactive power, total harmonic current, total harmonic voltage, total harmonic distortion rate of voltage, principal component content of harmonic voltage, and principal component of harmonic current at any given time.

[0059] In this embodiment, 80% of the data is selected as the training dataset, and the remaining 20% ​​is selected as the test dataset.

[0060] In this embodiment, the input to the harmonic level estimation model includes two parts: ① initial feature matrix x , containing matrix A , E , O Each column stores the HRCD values ​​at different times; ② Real value label matrix Y It describes the harmonic emission and over-limit characteristics, i.e., the total harmonic current. Total harmonic voltage Total harmonic distortion of voltage And the principal components of harmonic voltage and current , The changes over time are shown in the diagram, with each row representing a type of HRCD and each column representing the harmonic level at different times.

[0061] In model training, the learning rate of the neural network was set to 0.001, the minimum training error was set to 0.00001, the number of training iterations was set to 200, and the number of hidden layers was set to 5. The number of neurons in the hidden layers was... ;in, This represents the number of neurons in the hidden layer. This represents the total number of samples in the training set. and These represent the number of neurons in the input layer and the output layer, respectively. The variable can be any value chosen, generally within the range of 2 to 10; in this embodiment, 5 is selected. Therefore, the number of neurons in the hidden layer is set to 36.

[0062] The LSTM model was trained using the training dataset, and the harmonic levels at the measurement points were simulated and estimated using the test dataset. Partial estimation results from the test set are shown below. Figure 6 , Figure 7 , Figure 8 , Figure 9 and Figure 10 As shown in the figure; where the blue curve represents the actual values ​​of the sample, and the red curve represents the estimated values.

[0063] Therefore, the method proposed in this embodiment can accurately estimate the harmonic level at the monitoring point. The estimated values ​​of total harmonic current, total harmonic voltage, total harmonic distortion of voltage, and principal components of harmonic voltage and current are basically consistent with the actual values, with small errors. The performance of the harmonic level estimation model, i.e., the estimation accuracy of the test dataset, is evaluated based on the previous steps, and the estimation error of the test dataset is shown in Table 3.

[0064] Table 3 Estimation error of the test dataset

[0065] Table 3 shows that the harmonic level estimation model estimated by LSTM... , , , and mean square error L RMSE and mean absolute error L MAE All coefficients were below 0.6, demonstrating that the deviation between the model's estimates and actual values ​​was small; and the coefficients of determination for the five harmonic level indicators were... All values ​​are above 0.91 and close to 1, demonstrating that the model's estimates fit the actual values ​​well. This proves that the LSTM-based harmonic level estimation method proposed in this embodiment has good accuracy and feasibility.

[0066] To verify the superiority of the method in this embodiment, three commonly used prediction algorithms—traditional RNN, random forest (RF), and back propagation neural network (BPNN)—were used to estimate the user's harmonic levels. The results were compared with the method described in this embodiment. The comparisons are as follows: Figure 11 , Figure 12 , Figure 13 , Figure 14 and Figure 15As shown, the estimated values ​​obtained by the method in this embodiment match the actual data much better than other models. It not only maintains more accurate tracking of the overall trend but also exhibits more stable fitting results when data fluctuations are significant, with minimal curve fluctuation amplitude and error. In contrast, other models show obvious biases during the fitting process, especially in areas of drastic data change, where errors are large and they fail to capture subtle data fluctuations effectively. Therefore, the method in this embodiment can accurately capture the changing trends of harmonic levels under a wider range of load variation conditions, demonstrating higher fitting accuracy and better time-series data modeling capabilities.

[0067] The performance evaluation metrics of RNN, RF, and BPNN in harmonic level estimation are compared, and the estimation error comparison is shown in Table 4. It can be seen that the three methods... L RMSE and L MAE All are greater than 0.6 (except for the index used by BPNN to estimate harmonic currents), and The values ​​are generally low, differing significantly from the value of 1, indicating a low degree of fit. Comparing Tables 3 and 4, we can obtain the results of the method in this embodiment. L RMSE and L MAE The accuracy is significantly lower than the other three methods, indicating that the method in this embodiment has the smallest error and the highest accuracy in harmonic estimation. In this embodiment... The values ​​are also significantly higher than other methods, indicating that it has the strongest fit to various harmonic disturbance indices and can more accurately reflect the changing trends of actual data. Although other models perform better on some indices, they generally have larger errors and weaker fits compared to the method in this embodiment. Therefore, Table 4 confirms the superiority of the proposed method in harmonic level estimation.

[0068] Table 4. Comparison of Estimation Errors

[0069] This embodiment applies LSTM to harmonic level estimation, using a data-driven approach to uncover the temporal variation patterns of harmonic emission levels hidden in multivariate monitoring data. It eliminates the need for system parameters such as harmonic impedance and network topology; harmonic levels can be estimated solely from conventional operational data monitored by smart meters. This provides a new approach to harmonic sensing in areas without online monitoring devices, freeing them from the constraints of configuring power quality monitoring equipment. Simulation analysis using actual monitoring data verifies the effectiveness and accuracy of the proposed method. The proposed model exhibits small estimation errors and high fitting accuracy, providing accurate guidance for harmonic mitigation strategies. Furthermore, the proposed harmonic level estimation method is low-cost and possesses strong application prospects and widespread value.

[0070] Example 2 Embodiment 2 of the present invention introduces a system for estimating the harmonic level of a power distribution network.

[0071] like Figure 16 The system shown includes a distribution network harmonic level estimation system. The acquisition module is configured to acquire real-time operating data of the power grid; The building module is configured to construct a distribution network harmonic characteristic dataset based on the acquired real-time operational data; The estimation module is configured to obtain the harmonic level estimation result at the current moment based on the constructed harmonic feature dataset and harmonic level estimation model, thereby completing the estimation of the harmonic level of the distribution network. The harmonic level estimation model employs a long short-term memory network, which combines short-term and long-term memory through gating to extract temporal correlation information between the constructed harmonic feature datasets. Based on the extracted temporal correlation information, the nonlinear relationship between the harmonic feature data is fitted, and the harmonic level is estimated based on the fitting result.

[0072] The detailed steps are the same as those provided in Example 1 for estimating the harmonic level of a distribution network, and will not be repeated here.

[0073] Example 3 Embodiment 3 of the present invention provides a computer-readable storage medium.

[0074] A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in a method for estimating the harmonic level of a distribution network as described in Embodiment 1 of the present invention.

[0075] The detailed steps are the same as those provided in Example 1 for estimating the harmonic level of a distribution network, and will not be repeated here.

[0076] Example 4 Embodiment 4 of the present invention provides an electronic device.

[0077] An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor. When the processor executes the program, it implements the steps in a method for estimating the harmonic level of a power distribution network as described in Embodiment 1 of the present invention.

[0078] The detailed steps are the same as those provided in Example 1 for estimating the harmonic level of a distribution network, and will not be repeated here.

[0079] Example 5 Embodiment 5 of the present invention provides a computer program product.

[0080] A computer program product includes software code, wherein the program in the software code performs the steps of a method for estimating the harmonic level of a distribution network as described in Embodiment 1 of the present invention.

[0081] The detailed steps are the same as those provided in Example 1 for estimating the harmonic level of a distribution network, and will not be repeated here.

[0082] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0083] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0084] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0085] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The steps of the function specified in one or more boxes.

[0086] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0087] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0088] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A method for estimating harmonic levels in a distribution network, characterized in that, include: Obtain real-time power grid operation data; Based on the acquired real-time operational data, a harmonic characteristic dataset for the power distribution network is constructed. Based on the constructed harmonic characteristic dataset and harmonic level estimation model, the harmonic level estimation result at the current moment is obtained, and the harmonic level of the distribution network is estimated. The harmonic level estimation model employs a long short-term memory network, which combines short-term and long-term memory through gating to extract temporal correlation information between the constructed harmonic feature datasets. Based on the extracted temporal correlation information, the nonlinear relationship between the harmonic feature data is fitted, and the harmonic level is estimated based on the fitting result.

2. The method for estimating the harmonic level of a distribution network as described in claim 1, characterized in that, In the process of harmonic level estimation, the harmonic influence feature data matrix is ​​input into the harmonic level estimation model based on long short-term memory network to obtain the harmonic emission feature data matrix and the harmonic distortion limit exceedance data matrix used to determine the harmonic level of the harmonic source. The evaluation index of harmonic level estimation is calculated based on the obtained harmonic emission feature data matrix and the harmonic distortion limit exceedance data matrix. The performance of harmonic level estimation is evaluated based on the obtained evaluation index.

3. The method for estimating the harmonic level of a distribution network as described in claim 1, characterized in that, The constructed distribution network harmonic characteristic dataset includes harmonic impact characteristic data, harmonic emission characteristic data, and harmonic distortion exceeding limit characteristic data. The harmonic impact characteristic data includes at least the RMS value of current, RMS value of voltage, active power, and reactive power. The harmonic emission characteristic data includes at least the total harmonic current and total harmonic voltage. The harmonic distortion exceeding limit characteristic data includes at least the total harmonic distortion rate of voltage, the principal component content of harmonic voltage, and the principal component of harmonic current.

4. The method for estimating the harmonic level of a distribution network as described in claim 1, characterized in that, The acquired real-time power grid operation data includes at least distribution network voltage, distribution network current, distribution network power, and continuous data of harmonic source grid connection points, including harmonic source timing operation and harmonic disturbance data.

5. The method for estimating the harmonic level of a distribution network as described in claim 1, characterized in that... The network structure of the harmonic level estimation model includes an input layer, an output layer, and at least one long short-term memory layer; wherein, the input layer is used to receive the harmonic correlation feature data matrix at time t. , for Harmonic influence characteristic data matrix at time point for Harmonic emission characteristic data matrix at time t, for The harmonic distortion over-limit feature data matrix at time t; each of the long short-term memory layers includes a forget gate, an input gate, an output gate, and a cell state; the output layer is used to output the harmonic emission feature data matrix estimated at time t. Harmonic distortion exceeding limit characteristic data matrix .

6. The method for estimating the harmonic level of a distribution network as described in claim 1, characterized in that, In the process of fitting the nonlinear relationship between harmonic characteristic data, the root mean square error loss function and the mean absolute error loss function are used for iterative training to improve the fitting accuracy.

7. A system for estimating the harmonic level of a power distribution network, characterized in that, include: The acquisition module is configured to acquire real-time operating data of the power grid; The building module is configured to construct a distribution network harmonic characteristic dataset based on the acquired real-time operational data; The estimation module is configured to obtain the harmonic level estimation result at the current moment based on the constructed harmonic feature dataset and harmonic level estimation model, thereby completing the estimation of the harmonic level of the distribution network. The harmonic level estimation model employs a long short-term memory network, which combines short-term and long-term memory through gating to extract temporal correlation information between the constructed harmonic feature datasets. Based on the extracted temporal correlation information, the nonlinear relationship between the harmonic feature data is fitted, and the harmonic level is estimated based on the fitting result.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of a method for estimating the harmonic level of a distribution network as described in any one of claims 1-6.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the steps of a method for estimating the harmonic level of a distribution network as described in any one of claims 1-6.

10. A computer program product, comprising software code, characterized in that, The program in the software code performs the steps of a method for estimating the harmonic level of a distribution network as described in any one of claims 1-6.

Images