A non-intrusive system harmonic impedance estimation method and system

Abstract

The application discloses a non-intervention system harmonic impedance estimation method and system, the method comprises the following steps: collecting harmonic voltage and current vector data; constructing a characteristic vector reflecting amplitude, phase and time variation characteristics; dividing an overlapping sliding time window according to a preset window length and step; constructing an isolation forest model in each window to calculate an abnormal score; fusing multi-window scores to obtain a comprehensive abnormal score and introducing a time consistency constraint; screening or weighted processing according to the score to obtain an effective data set; and estimating system harmonic impedance through a regression method. The application can effectively identify and weaken the adverse effects of background harmonic voltage fluctuations by performing abnormal detection in a local time range, can adaptively identify abnormalities without manual annotation, can improve estimation stability by replacing simple rejection with continuous weighting, and can improve the precision and robustness of harmonic impedance estimation.

Description

Technical Field

[0001] This invention relates to the field of power quality technology, and in particular to a non-interventional method and system for estimating system harmonic impedance. Background Technology

[0002] With the transformation of the power system and rapid economic development, a large number of power electronic devices and industrial nonlinear loads are now connected to the power grid, bringing serious harmonic problems. Harmonics can cause voltage waveform distortion or resonance in the power grid, severely affecting power quality and threatening the stable and safe operation of the power system. Reasonably assessing the harmonic emission levels at the point of common coupling (PCC) on both the system side and the user side is a crucial step in effectively controlling harmonics. This requires accurately estimating the harmonic impedance on the system side; therefore, accurately calculating the harmonic impedance on the system side is key to harmonic source tracing and mitigation.

[0003] Current harmonic impedance estimation methods are mainly divided into two categories: "interventional" and "non-interventional." "Interventional" methods inject artificially generated disturbances into the system or alter the system network topology, injecting harmonic or interharmonic currents, switching branches on and off, or starting and stopping loads to estimate the system's harmonic impedance. However, the injected disturbances can adversely affect system operation. "Non-interventional" methods calculate harmonic impedance without interfering with normal system operation, utilizing natural disturbances from the load or the system itself, as well as measurable parameters. These methods do not affect normal system operation and are therefore widely used.

[0004] Currently, linear regression is widely used in "non-interventional" methods, but it typically requires the following assumptions: within a certain time scale, the system background harmonic voltage fluctuations are small, and the harmonic impedance remains approximately constant. However, in actual power grid operation, due to the influence of multiple harmonic sources, load fluctuations, and random disturbances, the background harmonic voltage often exhibits non-stationary characteristics, resulting in a large number of outlier samples mixed into the measurement data, which significantly reduces the accuracy and stability of harmonic impedance estimation.

[0005] In existing technologies, thresholding or simple statistical methods are commonly used to remove abnormal data. However, these methods rely on human experience, are difficult to adapt to different working conditions, and cannot effectively identify abnormal data caused by background harmonic voltage fluctuations. Summary of the Invention

[0006] To address the aforementioned issues, this invention provides a non-interventional system harmonic impedance estimation method and system that eliminates the need for manual labeling, adaptively identifies background harmonic voltage fluctuations, and improves the accuracy of harmonic impedance estimation.

[0007] This invention provides a non-interventional method for estimating system harmonic impedance, the specific technical solution of which is as follows: S1: Collect harmonic voltage and harmonic current vector data of the node to be analyzed in the power system at multiple discrete moments; S2: Based on the harmonic voltage and current vector data, construct a harmonic data feature vector that reflects the characteristics of harmonic amplitude, phase and time variation; S3: Divide the harmonic data into multiple overlapping sliding time windows according to the preset time window length and sliding step size; S4: Within each sliding time window, based on the harmonic data feature vector within the corresponding time window, construct an isolated forest model and calculate the anomaly score for each harmonic data sample. S5: For harmonic data samples at the same moment, fuse their corresponding anomaly scores in multiple sliding time windows to obtain a comprehensive anomaly score; S6: Based on the comprehensive anomaly score, the harmonic data is filtered or weighted to obtain an effective dataset for non-interventional system harmonic impedance estimation. S7: Based on the filtered harmonic data, the system harmonic impedance is estimated using the regression method.

[0008] Further, step S1 is as follows: Collection of the first h Subharmonics at discrete times k The voltage vector and the current vector; The voltage vector is represented as follows:

[0009] The current vector is represented as follows:

[0010] in, and These represent the amplitudes of the harmonic voltage and current, respectively. and These are the corresponding phase angles.

[0011] Further, in step S2, the harmonic data feature vector is represented as follows:

[0012] The change between adjacent time points is defined as follows:

[0013]

[0014]

[0015] .

[0016] The feature vector reflects both the harmonic amplitude level and short-term fluctuations, and is used to distinguish between background harmonic voltage quasi-steady-state data and abnormal disturbance data.

[0017] Furthermore, the anomaly scores for each harmonic data sample are calculated as follows:

[0018] in, Indicates sample In the sliding time window Abnormal ratings within, Indicates sample Average path length in the isolated forest model This is the normalization factor for the average path length.

[0019] Furthermore, the comprehensive anomaly score is calculated as follows:

[0020]

[0021] in, This indicates the overall anomaly score.

[0022] Furthermore, the filtering process is as follows: Set an abnormal threshold The comprehensive anomaly score is compared with the anomaly threshold. Compare; If the overall anomaly score is not greater than the anomaly threshold If so, the corresponding harmonic data will be retained as valid data.

[0023] Furthermore, the weighting process is as follows: The weighting coefficients are constructed based on the comprehensive anomaly score, and are expressed as follows:

[0024] in, For weight adjustment parameters, This indicates the overall anomaly score; Weighting coefficients at each time point The corresponding harmonic data samples are weighted and labeled to obtain a weighted harmonic dataset.

[0025] By using a continuous weighting method to process outlier data, rather than simply removing it, the integrity of the data is preserved while the impact of outliers is weakened, thus improving the stability of impedance estimation.

[0026] Further, step S7 is as follows: A harmonic impedance regression model was established based on the selected harmonic data.

[0027] in, Indicates the first h Subharmonics at discrete times k Harmonic voltage, Indicates the first h Subharmonics at discrete times k Harmonic currents, for h Subsystem harmonic impedance for h Sub-background harmonic voltage.

[0028] The present invention also provides a non-interventional system harmonic impedance estimation system, including a data acquisition module for acquiring harmonic voltage and harmonic current vector data of the node to be analyzed in the power system at multiple discrete times; The feature construction module constructs a harmonic data feature vector that reflects the harmonic amplitude, phase, and time variation characteristics based on the harmonic voltage and current vector data. The sliding window partitioning module divides the harmonic data into multiple overlapping sliding time windows according to the preset time window length and sliding step size; The anomaly detection module establishes an isolated forest model based on the harmonic data feature vector within each sliding time window and calculates the anomaly score for each harmonic data sample. The scoring fusion module merges the anomaly scores corresponding to harmonic data samples at the same moment in multiple sliding time windows to obtain a comprehensive anomaly score. The data filtering module filters or weights the harmonic data based on the comprehensive anomaly score to obtain an effective dataset for non-interventional system harmonic impedance estimation. The impedance estimation module uses a regression method to estimate the system harmonic impedance based on the filtered harmonic data.

[0029] Furthermore, the data filtering module includes a threshold filtering unit and / or a weighted processing unit; The threshold filtering unit performs binarization filtering on harmonic data according to a preset abnormal threshold, and outputs harmonic data with a comprehensive abnormal score less than or equal to the abnormal threshold as a valid dataset. The weighting processing unit is used to construct exponentially decaying weight coefficients based on the comprehensive anomaly score, and associate the weight coefficients with the harmonic voltage and harmonic current data at the corresponding time to construct a weighted harmonic dataset as a valid dataset. The beneficial effects of this invention are as follows: 1. This invention detects anomalies in harmonic data within a local time range and introduces time consistency constraints, effectively identifying and mitigating the adverse effects of background harmonic voltage fluctuations on harmonic impedance estimation results. This improves the accuracy and robustness of harmonic impedance estimation and solves the problem of decreased accuracy in harmonic impedance estimation caused by non-stationary changes in background harmonic voltage.

[0030] 2. This invention uses an isolated forest model to achieve unsupervised anomaly detection, eliminating the need for manual labeling of anomaly samples or setting empirical thresholds, and can adapt to the background harmonic voltage fluctuation characteristics under different operating conditions.

[0031] 3. This invention introduces time consistency constraints by using a sliding time window overlap design and multi-window anomaly scoring fusion, which avoids misjudgment caused by instantaneous noise and improves the reliability of anomaly detection. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0033] Figure 2 This is a schematic diagram of the Norton equivalent circuit of the power system in Embodiment 1 of the present invention.

[0034] Figure 3 This is a schematic diagram illustrating the principle of harmonic data filtering using a sliding time window in this invention.

[0035] Figure 4 This is a schematic diagram illustrating the isolation forest screening principle of this invention.

[0036] Figure 5 This is a schematic diagram of a 14-node power system in Embodiment 1 of the present invention.

[0037] Figure 6 This is a schematic diagram of the fifth harmonic voltage and current amplitude data in Embodiment 1 of the present invention.

[0038] Figure 7 This is a schematic diagram comparing harmonic data filtering before and after in Embodiment 1 of the present invention. Detailed Implementation

[0039] The technical solutions in the embodiments of the present invention are clearly and completely described in the following description. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0040] In the description of the embodiments of the present invention, it should be noted that the indicated orientation or positional relationship is based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship in which the product of the invention is conventionally placed during use, or the orientation or positional relationship in which those skilled in the art conventionally understand it during use. This is only for the convenience of describing the present invention and simplifying the description, and is not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of the present invention. Furthermore, the terms "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0041] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set" and "connection" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in the present invention based on the specific circumstances.

[0042] Example 1 Embodiment 1 of the present invention discloses a non-interventional system harmonic impedance estimation method, such as... Figure 1 As shown, the details are as follows: S1: Collect harmonic voltage and harmonic current vector data of the node to be analyzed in the power system at multiple discrete moments; Specifically, in this embodiment, the Norton circuit divides the power system into the system side and the user side, and calculates the system harmonic impedance by measuring the harmonic voltage and current data at the PCC. The equivalent model is as follows: Figure 2 As shown, based on the equivalent Norton circuit, the following relationship is obtained:

[0043] in, At the point of common coupling (PCC) h Subharmonic voltage For PCC h Second harmonic current for h Subsystem harmonic impedance for h Sub-background harmonic current for h Sub-background harmonic voltage.

[0044] Depend on Figure 1 It is known that only the harmonic voltage and current data at point PCC are measurable. Therefore, the data at point PCC is collected. hSubharmonics at discrete times k The voltage and current amplitudes and phase angle data are represented as follows:

[0045]

[0046] in, and These represent the amplitudes of the harmonic voltage and current, respectively. and These are the corresponding phase angles.

[0047] S2: Based on the harmonic voltage and current vector data, construct a harmonic data feature vector that reflects the characteristics of harmonic amplitude, phase and time variation, and use this vector to select periods when the background harmonic voltage is relatively stable for system harmonic impedance estimation.

[0048] The harmonic data feature vector is represented as follows:

[0049] The change between adjacent time points is defined as follows:

[0050]

[0051]

[0052]

[0053] The feature vector reflects both the harmonic amplitude level and short-term fluctuations, and is used to distinguish between background harmonic voltage quasi-steady-state data and abnormal disturbance data.

[0054] S3: Divide the harmonic data into multiple overlapping sliding time windows according to the preset time window length and sliding step size; like Figure 3 As shown, in this embodiment, let the sliding time window be L and the sliding step size be s, then the m-th sliding window is:

[0055] in, m This is the starting index of the sliding window. L The entire data sequence is divided into multiple sliding time windows, with the window length determined by the total sliding time window. Within the sliding time window, for h Subharmonic constructs a set of harmonic characteristic samples:

[0056] in, This is the characteristic vector of the harmonic at that moment.

[0057] This time window limits the time constraints of anomaly detection and avoids the impact of changes in global data distribution on anomaly determination results.

[0058] S4: Within each sliding time window, construct an isolated forest model based on the harmonic data feature vector within the corresponding time window, and calculate the anomaly score for each harmonic data sample.

[0059] Specifically, such as Figure 4 As shown, suppose an isolated forest consists of T random isolated trees, and the sample... exist t The path length in the isolated tree is:

[0060] Its average path length is:

[0061] sample In the sliding time window The abnormality score within is:

[0062] in, The average path length normalization factor is expressed as follows:

[0063] in, This represents the number of samples within the time window. is Euler's constant.

[0064] The larger the anomaly score, the easier it is for the harmonic data sample to be isolated, and the more likely it is to belong to background harmonic voltage fluctuations or abnormal operating conditions. In this way, local anomaly scores can be performed on the samples within each sliding window.

[0065] S5: Due to overlapping sliding windows, at the same time... The data may be contained in multiple sliding windows. To comprehensively consider the anomaly characteristics in the time dimension, the anomaly scores at the same time point are averaged and fused to obtain a comprehensive anomaly score.

[0066]

[0067] By fusing anomaly scores across multiple time windows and introducing time consistency constraints, misjudgments caused by instantaneous noise can be avoided.

[0068] S6: Based on the comprehensive anomaly score, the harmonic data is filtered or weighted to obtain an effective dataset for non-interventional system harmonic impedance estimation. Specifically, the filtering process is as follows: Set an abnormal threshold Define the data retention criteria:

[0069] Only when At that time, the corresponding harmonic data is used for subsequent impedance estimation.

[0070] The weighting process is as follows: Construct weighting coefficients based on the comprehensive anomaly score:

[0071] in, This is the weight adjustment parameter.

[0072] Weighting coefficients at each time point The corresponding harmonic data samples are weighted and labeled to obtain a weighted harmonic dataset, which is a weighted effective dataset; each sample in the weighted harmonic dataset contains a weight coefficient and its corresponding harmonic voltage and harmonic current.

[0073] This continuous weighting approach weakens the impact of outliers, rather than simply removing them, thus improving the stability of impedance estimation.

[0074] S7: Based on the filtered harmonic data, the system harmonic impedance is estimated using a regression method, as detailed below: Based on the selected harmonic data, a harmonic impedance regression model is established:

[0075] The harmonic impedance is estimated using the weighted least squares method: .

[0076] Based on the above method, in this embodiment, Figure 5 The standard 14-node system shown is illustrated here. This system is often used to verify the performance of harmonic impedance estimation methods. The sampling frequency is set to 6400 Hz, and the number of samples is 2000.

[0077] In this simulation, the load at node 11 is considered the user side, while the rest are considered the system side. Considering a multi-harmonic source scenario, harmonic current sources HS are added at nodes 3, 4, and 6 to model background harmonic voltage fluctuations, while the harmonic current source HL added at node 11 represents user-side fluctuations. The estimation process will now be illustrated using the fifth harmonic as an example.

[0078] To verify that the method described in this invention can effectively reduce the adverse effects of background harmonic voltage fluctuations on impedance estimation results, background harmonic voltage fluctuations were simulated in the system. Since the main harmonic sources in the system are the background harmonic source HS and the load-side harmonic source HL, different levels of background harmonic voltage fluctuations were simulated by adjusting the ratio R between the harmonic current sources HS and HL. Considering a relatively severe background harmonic voltage fluctuation scenario, harmonic data measured with R=1 were used, and the amplitude data obtained through Fourier transform are shown below. Figure 6 As shown.

[0079] Multiple experiments revealed that a sliding window width of 80 and a sliding step size of 25 effectively filtered harmonic data. Therefore, a sliding window combined with the isolated forest method was used to filter harmonic data, retaining those below the outlier threshold for subsequent system harmonic impedance estimation. A comparison of results before and after filtering is provided below. Figure 7 As shown, the method of the present invention can effectively identify and weaken the adverse effects of background harmonic voltage fluctuations on harmonic impedance estimation results by detecting anomalies in harmonic data within a local time range and introducing time consistency constraints.

[0080] The filtered harmonic data were substituted into the regression equation, and the system harmonic impedance was estimated based on the weight coefficients constructed according to the comprehensive anomaly score and the corresponding weights. The system harmonic impedance estimation results are shown in Table 1.

[0081] Table 1: System harmonic impedance estimation results

[0082] The estimated relative error of the fifth system harmonic impedance amplitude is 5.24%, and the relative error of the phase angle is 1.25%. It can be seen that the relative error is relatively small, which verifies the effectiveness of the method described in this invention in estimating system harmonic impedance.

[0083] In summary, the method described in this embodiment improves the accuracy and robustness of harmonic impedance estimation by detecting anomalies in harmonic data within a local time range and introducing time consistency constraints. This effectively identifies and weakens the adverse effects of background harmonic voltage fluctuations on harmonic impedance estimation results.

[0084] Example 2 Embodiment 2 of the present invention discloses a non-interventional system harmonic impedance estimation system. Based on the method described in Embodiment 1 above, the system includes a data acquisition module that acquires harmonic voltage and harmonic current vector data of the node to be analyzed in the power system at multiple discrete times. The feature construction module constructs a harmonic data feature vector that reflects the harmonic amplitude, phase, and time variation characteristics based on the harmonic voltage and current vector data. The sliding window partitioning module divides the harmonic data into multiple overlapping sliding time windows according to the preset time window length and sliding step size; The anomaly detection module establishes an isolated forest model based on the harmonic data feature vector within each sliding time window and calculates the anomaly score for each harmonic data sample. Specifically, the isolated forest consists of multiple random isolated trees. The anomaly score is calculated based on the average path length of the sample in the isolated tree. The larger the anomaly score, the more likely the sample is to belong to background harmonic voltage fluctuations or abnormal operating conditions. The scoring fusion module merges the anomaly scores corresponding to harmonic data samples at the same moment in multiple sliding time windows to obtain a comprehensive anomaly score. The data filtering module filters or weights the harmonic data based on the comprehensive anomaly score to obtain an effective dataset for non-interventional system harmonic impedance estimation. Specifically, the data filtering module includes a threshold filtering unit and / or a weighted processing unit; The threshold filtering unit performs binarization filtering on harmonic data according to a preset abnormal threshold, and outputs harmonic data with a comprehensive abnormal score less than or equal to the abnormal threshold as a valid dataset. The weighted processing unit is used to construct an exponentially decaying weight coefficient based on the comprehensive anomaly score, associate the weight coefficient with the harmonic voltage and harmonic current data at the corresponding time, and construct a weighted harmonic dataset as a valid dataset. The impedance estimation module uses a regression method to estimate the system harmonic impedance based on the filtered harmonic data.

[0085] This invention is not limited to the specific embodiments described above. The invention extends to any new feature or combination disclosed in this specification, as well as any new method or process step or combination disclosed herein.

Claims

1. A non-interventional method for estimating system harmonic impedance, characterized in that, include: S1: Collect harmonic voltage and harmonic current vector data of the node to be analyzed in the power system at multiple discrete moments; S2: Based on the harmonic voltage and current vector data, construct a harmonic data feature vector that reflects the characteristics of harmonic amplitude, phase and time variation; S3: Divide the harmonic data into multiple overlapping sliding time windows according to the preset time window length and sliding step size; S4: Within each sliding time window, based on the harmonic data feature vector within the corresponding time window, construct an isolated forest model and calculate the anomaly score for each harmonic data sample. S5: For harmonic data samples at the same moment, fuse their corresponding anomaly scores in multiple sliding time windows to obtain a comprehensive anomaly score; S6: Based on the comprehensive anomaly score, the harmonic data is filtered or weighted to obtain an effective dataset for non-interventional system harmonic impedance estimation. S7: Based on the filtered harmonic data, the system harmonic impedance is estimated using the regression method.

2. The non-interventional system harmonic impedance estimation method according to claim 1, characterized in that, Step S1 is as follows: Collection of the first h Subharmonics at discrete times k The voltage vector and the current vector; The voltage vector is represented as follows: The current vector is represented as follows: in, and These represent the amplitudes of the harmonic voltage and current, respectively. and These are the corresponding phase angles.

3. The non-interventional system harmonic impedance estimation method according to claim 2, characterized in that, Step S2, the harmonic data feature vector is represented as follows: The change between adjacent time points is defined as follows: 。 4. The non-interventional system harmonic impedance estimation method according to claim 1, characterized in that, The anomaly score for each harmonic data sample is calculated as follows: in, Indicates sample In the sliding time window Abnormal ratings within, Indicates sample Average path length in the isolated forest model This is the normalization factor for the average path length.

5. The non-interventional system harmonic impedance estimation method according to claim 4, characterized in that, The comprehensive anomaly score is calculated as follows: in, This indicates the overall anomaly score.

6. The non-interventional system harmonic impedance estimation method according to claim 1, characterized in that, The filtering process is as follows: Set an abnormal threshold The comprehensive anomaly score is compared with the anomaly threshold. Compare; If the overall anomaly score is not greater than the anomaly threshold If so, the corresponding harmonic data will be retained as valid data.

7. The non-interventional system harmonic impedance estimation method according to claim 1, characterized in that, The weighting process is as follows: The weighting coefficients are constructed based on the comprehensive anomaly score, and are expressed as follows: in, For weight adjustment parameters, This indicates the overall anomaly score; Weighting coefficients at each time point The corresponding harmonic data samples are weighted and labeled to obtain a weighted harmonic dataset.

8. The non-interventional system harmonic impedance estimation method according to claim 1, characterized in that, Step S7 is as follows: A harmonic impedance regression model was established based on the selected harmonic data. in, Indicates the first h Subharmonics at discrete times k Harmonic voltage, Indicates the first h Subharmonics at discrete times k Harmonic currents, for h Subsystem harmonic impedance for h Sub-background harmonic voltage.

9. A non-interventional system harmonic impedance estimation system, characterized in that, Includes a data acquisition module, which collects harmonic voltage and harmonic current vector data of the node to be analyzed in the power system at multiple discrete moments; The feature construction module constructs a harmonic data feature vector that reflects the harmonic amplitude, phase, and time variation characteristics based on the harmonic voltage and current vector data. The sliding window partitioning module divides the harmonic data into multiple overlapping sliding time windows according to the preset time window length and sliding step size; The anomaly detection module establishes an isolated forest model based on the harmonic data feature vector within each sliding time window and calculates the anomaly score for each harmonic data sample. The scoring fusion module merges the anomaly scores corresponding to harmonic data samples at the same moment in multiple sliding time windows to obtain a comprehensive anomaly score. The data filtering module filters or weights the harmonic data based on the comprehensive anomaly score to obtain an effective dataset for non-interventional system harmonic impedance estimation. The impedance estimation module uses a regression method to estimate the system harmonic impedance based on the filtered harmonic data.

10. The non-interventional system harmonic impedance estimation system according to claim 9, characterized in that, The data filtering module includes a threshold filtering unit and / or a weighted processing unit; The threshold filtering unit performs binarization filtering on harmonic data according to a preset abnormal threshold, and outputs harmonic data with a comprehensive abnormal score less than or equal to the abnormal threshold as a valid dataset. The weighted processing unit is used to construct an exponentially decaying weight coefficient based on the comprehensive anomaly score, associate the weight coefficient with the harmonic voltage and harmonic current data at the corresponding time, and construct a weighted harmonic dataset as a valid dataset.

Images