Method for identifying electric energy quality trend of responsibility subject in typical interference scene

Abstract

The invention discloses a responsibility subject electric energy quality trend identification method in a typical interference scene, and belongs to the field of electric energy quality analysis, and the method comprises the steps: obtaining electric energy quality time sequence data, decomposing the electric energy quality time sequence data, and obtaining a steady-state time sequence trend component; checking the steady-state time sequence trend component through a trend checking method to obtain a trend type; and calculating the steady-state time sequence trend component according to the trend type to obtain a trend span index so as to carry out quantitative identification on the power quality trend, and through the above technical scheme, quantitative effective identification can be carried out on the power quality trend.

Description

Technical Field

[0001] The present invention belongs to the technical field of power quality trend analysis, and particularly relates to a method for identifying the power quality trend of the responsible entity under typical interference scenarios. Background Art

[0002] With the continuous acceleration of the industrialization process and the continuous improvement of the electrification level, the fluctuations of power loads and power quality problems have become increasingly prominent. In urban planning and power system design, more and more attention has been paid to power quality problems. At the same time, with the wide application of electronic devices and information technology, the requirements for power quality have become higher and higher, and power quality problems have gradually become an important issue affecting social and economic development and human health. Therefore, the research on the evolution law of power quality trends has become increasingly important.

[0003] The research on the evolution law of power quality trends requires the use of a variety of methods and technologies, including: (1) Measurement and analysis method: By real-time monitoring various indicators of electric energy, analyze and diagnose the power quality, and master the evolution law of power quality trends through the analysis results. (2) Data mining and processing method: By mining and processing a large amount of historical data, extract the key indicators and characteristics of power quality, and analyze the evolution law of trends. (3) Mathematical modeling method: Establish a mathematical model to fit and predict power quality data, so as to reveal its evolution law of trends.

[0004] Among them, on the one hand, it mainly focuses on the measurement, analysis and modeling research of indicators such as grid voltage, current, and harmonics; on the other hand, it pays more attention to the research on flicker, intermittent interference and nonlinear loads in power quality. At the same time, in the research methods, a large number of modern methods such as statistical analysis, time series analysis, and neural networks are adopted. However, the above-mentioned existing technologies are not accurate enough in identifying the evolution law of power quality trends. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes a method for identifying the power quality trend of the responsible entity under typical interference scenarios to solve the problems existing in the above-mentioned existing technologies.

[0006] To achieve the above object, the present invention provides a method for identifying the power quality trend of the responsible entity under typical interference scenarios, including:

[0007] Obtain power quality time series data, decompose the power quality time series data to obtain a steady-state time series trend component;

[0008] Test the steady-state time series trend component through a trend test method to obtain a trend type;

[0009] Calculate the steady-state time-series trend component according to the trend type to obtain a trend span index for quantitatively identifying the power quality trend.

[0010] Optionally, decompose the power quality time-series data by the seasonal extraction decomposition method to obtain the steady-state time-series trend component.

[0011] Optionally, the process of decomposing the power quality time-series data includes:

[0012] Construct an ARIMA model based on the power quality time-series data and obtain the seasonal component through the ARIMA model;

[0013] Decompose the power quality time-series data through the Hodrick-Prescott filter to obtain the trend component and the disturbance component;

[0014] Remove the trend component and the seasonal component from the power quality time-series data to obtain the residual component;

[0015] Test the ARIMA model, the seasonal component, and the trend component. When the test is passed, use the trend component as the steady-state time-series trend component.

[0016] Optionally, test the steady-state time-series trend component by the Mann-Kendall trend test method.

[0017] Optionally, the process of testing the steady-state time-series trend component includes:

[0018] Calculate the statistic according to the steady-state time-series trend component, where:

[0019]

[0020] where S represents the statistic, n represents the number of data in the power quality index time series, T j and T k represent the data of any two time points in the trend component respectively, where j > k, and the subscripts j and k represent the corresponding data point labels;

[0021] Calculate the variance of the statistic;

[0022] V ar (S) = n(n - 1)(2n - 5) / 18;

[0023] where V ar (S) represents the variance of the statistic;

[0024] Calculate the trend test statistic value according to the variance of the statistic;

[0025]

[0026] Among them, Z represents the trend test statistic value;

[0027] Judge the trend test statistic value to obtain the trend type.

[0028] Optionally, the process of judging the trend test statistic value includes:

[0029] When the trend test statistic value is greater than 0, the power quality time series data is in an upward trend; when the trend test statistic value is less than 0, the power quality time series data is in a downward trend.

[0030] Optionally, when the trend type is an upward trend, calculate the trend span index according to the steady-state time series trend component:

[0031]

[0032] Among them, P c represents the trend span index, max(T vc ) and min(T vc ) are respectively the maximum value and the minimum value of the trend component of the power quality time series data in the c-th group, T vc is the steady-state time series trend component of the power quality time series data in the c-th group, and u c is the national standard limit interval corresponding to this index.

[0033] On the other hand, the present application provides a responsible entity power quality trend recognition system under a typical interference scenario, which is used to execute the above method.

[0034] Compared with the prior art, the present invention has the following advantages and technical effects:

[0035] 1. The SEATS decomposition method proposed by the present invention is a time series decomposition method, which decomposes the time series into three parts: trend, seasonality and residuals. It can adaptively adjust the parameters of the ARIMA model according to different time series to best capture the trend and seasonal changes in the time series. The results output by SEATS are very clear and can clearly indicate the changes in the trend, seasonality and residual parts of the time series, which is convenient for analysts to conduct further analysis and interpretation. Since SEATS uses the ARIMA model for time series decomposition, it can be implemented using existing ARIMA model tools, and the processing speed is relatively fast. The SEATS method has a high reliability in time series decomposition, can provide relatively accurate decomposition results, and can also test the ARIMA model, seasonal and trend models to ensure the reliability of the decomposition.

[0036] 2. The MK trend test method proposed by the present invention is a commonly used non-parametric statistical method, which does not require sample data to follow a certain distribution and has good processing effects on data sets with non-normal distributions.

[0037] 3. The meaning of the trend span index proposed by the present invention is clear and definite, and the calculation method is simple and fast. It is applicable to time series of different types of power quality indicators and is convenient for analyzing a large amount of power quality monitoring data. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings constituting a part of this application are used to provide a further understanding of this application. The schematic embodiments and descriptions thereof of this application are used to explain this application and do not constitute an improper limitation of this application. In the drawings:

[0039] Figure 1 It is the basic principle diagram of the power quality trend recognition method based on the SEATS decomposition method in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The following will refer to the drawings and describe the application in detail with reference to the embodiments.

[0041] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0042] The present invention discloses a power quality trend identification method based on the seasonal extraction (SEATS) decomposition method of the autoregressive integrated moving average (ARIMA) time series, comprising the following steps: decomposing the power quality time series data by using the SEATS decomposition method to extract the characteristic components reflecting the overall trend change of the time series; using the Mann-Kendall (MK) trend test method to test the trend components of the steady-state power quality index time series obtained from the SEATS decomposition, judging the significance of the trend change of the power quality index, and determining the trend type of the time series data; calculating the trend span index to quantitatively describe the degree of trend change of different power quality indexes, so as to realize the research on the trend evolution law of power quality. Compared with the conventional trend identification method, the SEATS decomposition method proposed by the present invention is a time series decomposition method, which decomposes the time series into three parts: trend, seasonality and residual. The parameters of the ARIMA model can be adaptively adjusted according to different time series to best capture the trend and seasonal changes in the time series. Since SEATS uses the ARIMA model for time series decomposition, the ready-made ARIMA model tools can be used for implementation, and the processing speed is relatively fast.

[0043] Aiming at the deficiencies of the prior art, the present invention proposes a power quality trend identification method based on the seasonal extraction (SEATS) decomposition method of the autoregressive integrated moving average (ARIMA) time series. This method uses the SEATS method to decompose the power quality time series data, extracts the characteristic components reflecting the overall trend change of the time series, uses the MK trend test method to test the trend components of the steady-state power quality index time series, calculates the trend span index, and quantitatively describes the degree of trend change of different power quality indexes, and finally completes the research on the trend evolution law of power quality.

[0044] The object of the present invention is achieved by the following technical solutions:

[0045] A power quality trend identification method based on the seasonal extraction (SEATS) decomposition method of the autoregressive integrated moving average (ARIMA) time series, comprising the following steps:

[0046] S1: Decomposing the power quality time series data by using the SEATS decomposition method to extract the characteristic components reflecting the overall trend change of the time series;

[0047] Among them, the power quality time series (time series) data is the time series data of different power quality indexes, where the indexes include the voltage quality, frequency quality, waveform quality, etc. of the electric energy;

[0048] S2: Use the Mann-Kendall (MK) trend test method to test the trend component of the steady-state power quality index time series obtained from the SEATS decomposition, judge the significance of the trend change of the power quality index, and determine the trend type of the time series data;

[0049] S3: Calculate the trend span index to quantitatively describe the degree of trend change of different power quality indicators.

[0050] Furthermore, in S1, the SEATS method is used to decompose the original power quality time series data into three additive components, namely the trend component, the seasonal component, and the irregular residual component: Y v = T v + S v + R v , where Y v is the original numerical value of the time series at time v, and T v , S v , R v correspond to the trend component value, the seasonal component value, and the residual component value at time v respectively. It includes the following working steps:

[0051] (1) Establish an ARIMA model: First, convert the time series into an ARIMA model, which can well capture the trends and seasonal variations in the time series.

[0052] (2) Estimate seasonality: Use the ARIMA model to estimate the seasonal part in the time series, that is, the seasonal component value, and remove this part from the original power quality time series data.

[0053] (3) Estimate the trend: Use the Hodrick-Prescott (HP) filter to estimate the trend part in the time series. The HP filter is a very popular filter that can decompose the time series into two parts, namely the trend and the perturbation, that is, the trend component value and the perturbation component value.

[0054] (4) Estimate the residual: Subtract the trend and seasonal parts, that is, the trend component value and the seasonal component value, from the original power quality time series to obtain the residual part, that is, the residual component value.

[0055] (5) Conduct diagnostic tests: Test the ARIMA model, the estimated seasonality, and the trend model to ensure the reliability of the decomposition.

[0056] Furthermore, in S2, the MK method is used to test the trend component of the steady-state power quality index time series obtained from the SEATS decomposition, that is, the trend component value T v , which includes the following working steps:

[0057] (1) Assume that the time series data T of the power quality indexv =(T1, T2, …, T n ) are randomly independent, and a statistic S is constructed: where n is the number of data in the time series of power quality indicators, S represents the statistic, and T j and T k respectively represent the data at any two time points in the trend component (j > k). The subscripts j and k represent the corresponding data point labels. For the above n total data points, (n - 1)(n) / 2 comparisons need to be calculated to clarify that the MK test constructs the statistic S by traversing all data pairs of j > k (not only adjacent points) to ensure the comprehensiveness of trend judgment.

[0058] (2) When n > 8, the variance Var(S) of the statistical variable S: V ar (S) = n(n - 1)(2n - 5) / 18;

[0059] (3) The trend test statistic value Z: When Z > 0, the time series of power quality indicators has an upward trend. When Z < 0, the time series of power quality indicators has a downward trend.

[0060] Furthermore, the S3 proposes a trend span index to quantitatively describe the degree of trend change of different power quality indicators. Suppose there are a total of C groups of time series of power quality indicators whose trend types are determined to be significantly upward after the MK trend test in S2. The c-th group of sequences is T vc , then the trend span index P c of the c-th group of time series is: where max(T vc ) and min(T vc ) are the maximum and minimum values of the trend component of the c-th group of power quality indicator time series respectively, and u c is the national standard limit interval corresponding to this indicator.

[0061] Among them, the MK method test can identify both significantly upward (Z > 0) and downward trends (Z < 0) simultaneously, but this method focuses on significantly upward trends (because an increase in power quality indicators such as harmonic distortion rate usually indicates deterioration and needs to be pre-warned first). If it is necessary to analyze the downward trend (such as improvement in frequency stability), a negative span index can be symmetrically defined. In this invention, it is mainly a relevant review of the possible deterioration risks, such as the one-way control requirement of the national standard for "exceeding the standard risk".

[0062] Based on the trend span index, the change degree of the trend component of the power quality indicator time series within the corresponding national standard limit interval is studied. The higher the index, the greater the upward amplitude of the c-th group of power quality indicator time series, and the more serious the deterioration trend of the power quality indicator. Thus, the research on the evolution law of power quality trends is completed.

[0063] Among them, "Harmonics in Public Power Grids of Power Quality" (GB / T 14549-93) clearly stipulates the harmonic voltage limits. When the trend span index is calculated based on this limit interval, clauses such as "warning is required when exceeding the standard by 10% and mandatory treatment is required when exceeding the standard by 30%" in the standard can be directly used to define the severity level.

[0064] In some examples of the present invention, a power quality trend identification method based on the seasonal extraction (SEATS) decomposition method of the autoregressive integrated moving average (ARIMA) time series is disclosed, as Figure 1 shown, including the following steps:

[0065] S1: Use the SEATS decomposition method to decompose the power quality time series data and extract the characteristic components reflecting the overall trend change of the time series;

[0066] Using the SEATS method, the original power quality time series data is decomposed into three additive components, namely the trend component, the seasonal component, and the irregular residual component: Y v = T v + S v + R v , where Y v is the original numerical value of the time series at time v, and T v , S v , R v correspond to the trend component value, the seasonal component value, and the residual component value at time v respectively. It includes the following working steps:

[0067] (1) Establish an ARIMA model: First, convert the time series into an ARIMA model, which can well capture the trends and seasonal variations in the time series.

[0068] (2) Estimate seasonality: Use the ARIMA model to estimate the seasonal part in the time series and remove this part from the original series.

[0069] (3) Estimate the trend: Use the Hodrick-Prescott (HP) filter to estimate the trend part in the time series. The HP filter is a very popular filter that can decompose the time series into a trend and a disturbance part.

[0070] (4) Estimate the residuals: Subtract the trend and seasonal parts from the original series to obtain the residual part.

[0071] For the residual part, although the subsequent main participant in the operation is the trend component, it is necessary to test the extraction process of its trend component according to the residual part, that is, the residual component. Among them, the residual component is mainly used for the reliability verification of the ARIMA model: verify whether it is white noise through the Ljung-Box test (if not passed, it indicates that the ARIMA has not fully extracted the pattern); at the same time, its mutation component can locate abnormal events such as lightning strikes (such as the residual peak matching the fault time recorded by SCADA). Even if the trend component is used subsequently, the residual still provides "traceability" for the decomposition process, meeting the data integrity requirements of IEC 61000-4-30.

[0072] (5) Conduct diagnostic tests: Test the ARIMA model, estimated seasonal and trend components to ensure the reliability of the decomposition.

[0073] Among them, the following methods are used to conduct relevant tests on the above content: The test of the ARIMA model is carried out through the residual white noise test (such as the p-value of the Ljung-Box test > 0.05) and normality verification (linear distribution of the Q-Q plot), while the tests of the seasonal component and the trend component respectively point to the rationality evaluation of the decomposition results - the seasonal component needs to verify that there is no significant periodic energy residue in the residual through spectral analysis (such as the main frequency amplitude decreasing by 90% after Fourier transform), and the trend component needs to conduct the HP filter parameter sensitivity test (such as comparing the trend smoothness differences between λ = 14400 and λ = 129600) and the ADF stationarity test (ensuring that the trend component sequence has no unit root). Finally, the reliability of the model is comprehensively determined through the "decomposition - reconstruction" error rate (such as RMSE < 5%). When the reliability test is passed, the above-mentioned trend component is used for subsequent data processing. When the test is not passed, the ARIMA model and filter are readjusted or redesigned for re-component extraction and estimation.

[0074] S2: Use the Mann-Kendall (MK) trend test method to test the trend component of the steady-state power quality index time series obtained from the SEATS decomposition, judge the significance of the trend change of the power quality index, and determine the trend type of the time series data;

[0075] Use the MK method to test the trend component T of the steady-state power quality index time series obtained from the SEATS decomposition v , which includes the following working steps:

[0076] (1) Assume that the time series data T of the power quality index v =(T1, T2,..., T n ) is randomly independent, and construct the statistic S: where n is the number of data in the power quality index time series;

[0077] (2) When n > 8, calculate the variance Var(S) of the statistical variable S: V ar (S) = n(n - 1)(2n - 5) / 18;

[0078] (3) Trend test statistic Z: When Z > 0, the time series of power quality indicators has an upward trend; when Z < 0, the time series of power quality indicators has a downward trend.

[0079] S3: Calculate the trend span index to quantitatively describe the degree of trend change of different power quality indicators.

[0080] Propose a trend span index to quantitatively describe the degree of trend change of different power quality indicators. Assume that there are C groups of time series of power quality indicators whose trend types are determined to be significantly upward after the MK trend test in S2. The c-th group of sequences is T vc , then the trend span index P c of the c-th group of time series is: where max(T vc ) and min(T vc ) are the maximum and minimum values of the trend component of the c-th group of power quality indicator time series respectively, and u c is the national standard limit interval corresponding to this indicator. Based on the trend span index, study the degree of change of the trend component of the power quality indicator time series within the corresponding national standard limit interval. The higher the index, the greater the upward amplitude of the c-th group of power quality indicator time series, and the more serious the deterioration trend of the power quality indicator. Thus, the research on the trend evolution law of power quality is completed.

[0081] The present invention proposes a power quality trend identification method based on the SEATS decomposition method. The SEATS decomposition method is used to decompose the power quality time series data to extract the characteristic components reflecting the overall trend change of the time series. It can adaptively adjust the parameters of the ARIMA model according to different time series to best capture the trends and seasonal variations in the time series. The MK method is used to test the trend component T v of the steady-state power quality indicator time series obtained from the SEATS decomposition, judge the significance of the trend change of the power quality indicator, determine the trend type of the time series data, does not require the sample data to follow a certain distribution, has a good processing effect on non-normal distribution data sets, proposes a trend span index, which is applicable to different types of power quality indicator time series, is convenient to be applied to analyze a large amount of power quality monitoring data, and is used to quantitatively describe the degree of trend change of different power quality indicators.

[0082] The above are only the preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present application should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for identifying power quality trends of responsible entities under typical interference scenarios, characterized in that: include: Acquire power quality time series data, decompose the power quality time series data, and obtain steady-state time series trend components; The trend component of steady-state time series is tested by trend test method to obtain the trend type; The steady-state time series trend component is calculated according to the trend type to obtain a trend span index to quantitatively identify the power quality trend.

2. The method according to claim 1, characterized in that The power quality time series data is decomposed by a seasonal extraction decomposition method to obtain a steady-state time series trend component.

3. The method according to claim 1, characterized in that The process of decomposing the power quality time series data includes: Build an ARIMA model based on the power quality time series data and obtain the seasonal component through the ARIMA model; Decomposing the power quality time series data by using a Prescott filter to obtain a trend component and a disturbance component; The trend component and the seasonal component are removed from the power quality time series data to obtain the residual component; The ARIMA model, seasonal component and trend component are tested. When they meet the test, the trend component is taken as the steady-state time series trend component.

4. The method according to claim 1, characterized in that: The steady-state time series trend component was tested by the Mann-Kendall trend test.

5. The method according to claim 1, characterized in that The process of testing the steady-state time series trend component includes: According to the steady-state time series trend component, a statistic is calculated, where: Among them, S represents the statistic, n represents the number of data in the power quality index time series, T j and T k Respectively represent any two time point data in the trend component, where j>k, and the subscripts j and k represent the corresponding data point numbers; calculating the variance of the statistic; V ar (S)=n(n-1)(2n-5) / 18; Among them, V ar (S) represents the variance of the statistic; Calculate the trend test statistic according to the variance of the statistic; Where Z represents the trend test statistic; The trend test statistic is judged to obtain the trend type.

6. The method according to claim 1, characterized in that The process of judging the trend test statistic value includes: When the trend test statistic is greater than 0, the power quality time series data is in an upward trend, and when the trend test statistic is less than 0, the power quality time series data is in a downward trend.

7. The method according to claim 1, characterized in that When the trend type is an upward trend, the trend span index is calculated according to the steady-state time series trend component: Among them, P c Represents the trend span index, max(T vc ) and min(T vc ) are the maximum and minimum values ​​of the trend component of the cth group of power quality time series data, T vc is the steady-state temporal trend component of the cth group of power quality time series data, u c is the corresponding national standard limit range.

8. The responsible party power quality trend identification system under typical interference scenarios is characterized by: Used to execute the method described in any one of claims 1 to 7.

Images