A Photovoltaic Power Distribution Operation Early Warning Method, System, Device and Storage Medium

Abstract

The present application discloses a photovoltaic power distribution operation early warning method, system, device and storage medium, relating to the technical field of photovoltaic power distribution. The method includes: removing interference data from the monitoring data of a photovoltaic power station; after removing the interference data, determining abnormal data in the monitoring data of the photovoltaic power station through time series test analysis; setting early warning measurement indicators and error weighting indicators for the abnormal data, and calculating the fault early warning thresholds of the two indicators through variable transformation; and performing early warning on the operation of the photovoltaic power station according to the abnormal data and its fault early warning thresholds. By removing interference data, the present application improves the data quality, then uses time series analysis technology to quickly and accurately identify abnormal data, and sets early warning indicators for the abnormal data, making the early warning more intelligent and flexible, capable of adjusting according to different situations, avoiding unnecessary false early warnings and missed early warnings, and improving the accuracy and reliability of the early warning.

Description

Technical Field

[0001] The present application relates to the technical field of photovoltaic power distribution, and particularly relates to a method, system, device and storage medium for early warning of photovoltaic power distribution operation. Background Art

[0002] With the rapid development and large-scale application of photovoltaic power generation technology, the stable operation and efficient management of photovoltaic power stations have become particularly important. However, due to the influence of complex environmental conditions such as meteorological changes and equipment aging during the operation of photovoltaic power stations, as well as the interference of data acquisition devices, noise, anomalies or distortions often occur in the monitoring data, which makes it impossible to carry out accurate fault early warning and poses a hidden danger to the operation safety of the power station. Summary of the Invention

[0003] In order to achieve accurate fault early warning for photovoltaic power stations and ensure the operation safety of photovoltaic power stations, the present application proposes a method, system, device and storage medium for early warning of photovoltaic power distribution operation. By removing interference data, the present application improves the data quality, and then uses time series analysis technology to quickly and accurately identify abnormal data and sets early warning metrics for abnormal data, making the early warning more intelligent and flexible, capable of adjusting according to different situations, avoiding unnecessary false early warnings and missed early warnings, and improving the accuracy and reliability of early warning.

[0004] The present application is realized through the following technical solutions:

[0005] A method for early warning of photovoltaic power distribution operation, the early warning method comprising:

[0006] Removing interference data from the monitoring data of the photovoltaic power station;

[0007] After removing the interference data, determining the abnormal data in the monitoring data of the photovoltaic power station through time series test analysis;

[0008] Setting early warning measurement metrics and error weighting metrics for the abnormal data, and calculating the fault early warning thresholds of the two metrics through variable transformation;

[0009] Performing early warning of the operation of the photovoltaic power station according to the abnormal data and its fault early warning threshold.

[0010] In some embodiments, the process of removing interference data includes:

[0011] Identifying interference data from the monitoring data of the photovoltaic power station, and the process of identifying interference data includes: first calculating the average value of the absolute values of the data waveform segments of the monitoring data; then, based on the set determination criterion, identifying the interference data; wherein, the set determination criterion is: if the absolute value of the data is greater than or equal to the product of the average value and the discrimination threshold, then the data is interference data, otherwise it is not interference data;

[0012] The identified interference data is suppressed by using the median filtering method, specifically: the interference data is replaced with the median value of the data values within the filtering window.

[0013] In some embodiments, the time series test analysis process includes:

[0014] When there are only two time series variables in the monitoring data, the E-G two-step test method based on regression residuals is used to perform a stationarity test analysis on the time series variables, specifically including:

[0015] The unit root test method is used to perform a stationarity test on the time series variables to determine the integration order;

[0016] A stationarity regression model is constructed to obtain a stationary residual sequence;

[0017] The unit root test is performed on the stationary residual sequence. If the stationarity hypothesis test is not valid, it indicates that the data is abnormal.

[0018] In some embodiments, the time series test analysis process includes:

[0019] When the number of time series variables in the monitoring data is greater than 2, the Johansen test method is used to perform a stationarity test analysis on the time series variables, specifically including:

[0020] Determine the number of mutually independent stationary vectors;

[0021] Construct a vector error correction model for the considered time series variables;

[0022] The maximum likelihood method is used to estimate the parameters of the vector error correction model;

[0023] The maximum eigenvalue test is performed on the vector error correction model. If the stationarity hypothesis test is not valid, it indicates that the data is abnormal.

[0024] In some embodiments, the fault warning threshold calculation process includes:

[0025] Variable transformation is used to process the set original index data and convert it into data that conforms to the standard normal distribution; for the transformation parameters in the variable transformation, the maximum likelihood estimation method is used for calculation;

[0026] The probability density function of the data that conforms to the normal distribution is obtained, and the threshold of the fault warning index is solved in combination with the confidence interval;

[0027] The solved threshold is subjected to variable inverse transformation to obtain the fault warning threshold corresponding to the original index data.

[0028] In some embodiments, the operation warning process of the photovoltaic power station includes:

[0029] Compare the identified abnormal data with its corresponding fault warning threshold. If the abnormal data is greater than the fault warning threshold, a warning is issued.

[0030] In a second aspect, the present application proposes a photovoltaic power distribution operation warning system, which includes:

[0031] An interference removal module, which is used to remove the interference data in the monitoring data of the photovoltaic power station;

[0032] An abnormal identification module, which determines the abnormal data in the monitoring data of the photovoltaic power station after removing the interference data through time series test analysis;

[0033] A warning index module, which is used to set the warning measurement index and error weighting index of the abnormal data, and calculate the fault warning threshold of the two indexes through variable transformation;

[0034] And a comparison module, which conducts the operation warning of the photovoltaic power station according to the abnormal data and its fault warning threshold.

[0035] In some embodiments, the warning index module further includes:

[0036] A variable transformation unit, which performs variable transformation on the original index data and converts it into data that conforms to the standard normal distribution; the transformation parameters in the variable transformation are calculated by the maximum likelihood estimation method;

[0037] A threshold solving unit, which obtains the probability density function of the data that conforms to the normal distribution, and solves the threshold of the fault warning index in combination with the confidence interval;

[0038] And a variable inverse transformation unit, which performs variable inverse transformation on the solved threshold to obtain the fault warning threshold corresponding to the original index data.

[0039] In a third aspect, the present application proposes a computer device, which includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the above-mentioned photovoltaic power distribution operation warning method is implemented.

[0040] In a fourth aspect, the present application proposes a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned photovoltaic power distribution operation warning method is implemented.

[0041] A photovoltaic power distribution operation warning method, system, device and storage medium proposed by this application first removes interference data in the monitoring data of a photovoltaic power station, which can effectively filter out invalid or irrelevant data, reduce false alarms and errors, and ensure the quality of the monitoring data. Then, time series analysis is used to quickly and accurately identify potential problems or abnormal situations in the operation of the photovoltaic power station, so as to timely capture abnormal situations that may affect the operation of the power station, improving the sensitivity of fault detection. Finally, warning indicators for abnormal data are set, and warning thresholds are calculated to trigger warnings, making the warnings more intelligent and flexible, capable of being adjusted according to different situations, avoiding unnecessary false warnings and missed warnings, and further improving the reliability and accuracy of the warnings.

[0042] A photovoltaic power distribution operation warning method, system, device and storage medium proposed by this application enables potential problems in a photovoltaic power station to be discovered and processed in advance, thereby reducing downtime and the impact of faults, and improving the overall reliability and operation safety of the photovoltaic power station system. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings described herein are used to provide a further understanding of the embodiments of this application, form a part of this application, and do not constitute a limitation on the embodiments of this application. In the drawings:

[0044] Figure 1 is a flowchart of the warning method proposed by an embodiment of this application;

[0045] Figure 2 is a schematic block diagram of the warning system proposed by an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] To make the objectives, technical solutions and advantages of this application clearer and more understandable, the following further details this application in conjunction with the embodiments and the drawings. The illustrative embodiments and descriptions of this application are only used to explain this application and do not limit this application.

[0047] Embodiment: During the operation of a photovoltaic power station, it is affected by complex environmental conditions and interference from data acquisition equipment, resulting in inaccurate fault warnings and posing safety hazards to the operation of the power station. In response to this, this embodiment proposes a photovoltaic power distribution operation warning method.

[0048] As Figure 1 shown, the warning method proposed by this embodiment includes the following steps:

[0049] Step 100, removing interference data in the monitoring data of the photovoltaic power station.

[0050] In the monitoring of a photovoltaic power station, invalid or irrelevant data caused by external environmental interference can easily lead to false warnings for the photovoltaic power station. Therefore, it is necessary to effectively filter out invalid or irrelevant data, reduce false alarms and errors, and ensure the quality of monitoring data, so as to provide a more accurate data basis for subsequent analysis. In this embodiment, the specific process of removing interference data is as follows:

[0051] Step 101: Identify interference data from the monitoring data of the photovoltaic power station.

[0052] First, calculate the average value of the absolute values of the data in the waveform segment of the monitoring data; the average value calculation formula is as follows:

[0053]

[0054] where, represents the data value of the th sampling point in this waveform segment, is the number of sampling points, is the mean value

[0055] Then, based on the set judgment criterion, identify the interference data; the judgment criterion is as follows:

[0056]

[0057] where, is the outlier flag. If the value of is then it is interference data; otherwise, it is not interference data, is the given discrimination threshold.

[0058] Step 102: Suppress the identified interference data.

[0059] In this embodiment, median filtering method can be used but not limited to suppress the interference data. The principle of median filtering is:

[0060]

[0061] where, is the data value of the th sampling point in the filtering window, is the output result of median filtering, is the length of the median filter (usually set to an odd number), represents the median value of the data values within this window length.

[0062] Thus, the interference data monitored in the photovoltaic power station can be identified and suppressed, thereby avoiding false warnings caused by interference data and providing a more accurate and reliable data basis for the subsequent process.

[0063] Step 200: After removing the interference data, abnormal data in the photovoltaic power station monitoring data is determined through time series test analysis.

[0064] This embodiment uses time series to determine whether there are abnormal data in the monitored data, and then issues an early warning based on the abnormal data.

[0065] Among them, the central idea of time series is that even if two or more time series variables are non-stationary individually, their combination may produce a new stationary relationship. This stationary linear relationship is defined as stationarity, which means that multiple variables can be expressed as a long-term equilibrium stable relationship. If there is a stationary linear combination between non-stationary series, then they can be said to be stationary.

[0066] In the definition of stationarity, it is assumed that there is a stationary relationship between two sequences, where the input time variable sequence is The corresponding response time variable sequence is , the regression model between them can be expressed as:

[0067]

[0068] Where: is the stationary vector: is the regression residual, t Indicates the duration of the time series, a sequence with a stationary relationship and The regression residual series between It can be seen from the above formula that the ultimate goal of the stationary process is to identify the stationary vector to ensure the stationary residual sequence.

[0069] For a stationary time series variable , if the variable itself is stationary, it is called 0th order integration, denoted as , Indicates 0th order integration. For non-stationary time series variables, if it passes through After the second difference operation, it becomes a stationary time series variable, which is called Order single, indicating time series variables Existence Unit roots, denoted as , express d The stationary analysis process is only valid for non-stationary variables with the same stationary order. Therefore, in order to obtain the stationary vector , we first need to determine the non-stationary order of the variables. Then, the stationary vector It can be calculated during the stationarity test analysis.

[0070] The stationarity test can be divided into methods based on regression parameters and methods based on regression residuals. In addition, four key points need to be noted during the stationarity analysis process: First, the stationarity vector representing the relationship between non-stationary time series variables is not unique: Second, each time series variable should have the same non-stationarity, that is, the same order of stationarity can the stationarity analysis process be carried out; Third, for time series data with the number of variables being , there are at most linearly independent stationarity vectors: Fourth, each time series variable follows a similar sequence trend.

[0071] In this embodiment, the unit root test method is used to perform the time series stationarity test. Among them, the unit root test method is that when it is proved that a certain time series data has a unit root, then the series is non-stationary. The stationarity test process is as follows:

[0072] Starting from the form of the first-order autoregressive model, the first-order autoregressive model can be expressed as:

[0073]

[0074] In the formula: is the value at the current moment; is the value at the previous moment; is an independent normal random distribution Gaussian white noise sequence with a mean of 0 and a variance of ; is a real number, and its value determines the root of the characteristic equation of the above model. The root of the characteristic equation of any time series process will determine its stationarity. When is greater than or equal to 1, the time series process is non-stationary. When is less than 1, the time series process is stationary. When , there is a unit root in the process, and the above formula can be transformed into:

[0075]

[0076] Through 1-time difference operation, the non-stationary time series process is transformed into a stationary process, that is , is integrated of order 1. When the time series process is fitted to (1) model (i.e., the first-order autoregressive model), the stationarity information of the time series process can be obtained from the defined characteristic root parameters, which is usually achieved by testing the null hypothesis. When the time series data is fitted to a more complex time series model, the stationarity test follows the above principle, and the more complex time series model is expressed as:

[0077]

[0078] Among them, is the difference operator, defined as ; is the coefficient of a real number: is the model coefficient: is the lag number, which ensures that becomes a white noise sequence. To transform the model (i.e., the p order autoregressive model) into the above form, the following substitutions are required:

[0079]

[0080] Among them, is the coefficient of the model. By performing the above substitution process on the characteristic equation of the model, the characteristic equation can be obtained, that is:

[0081]

[0082] Among them, is the root of the above equation. In the characteristic equation of the time series model, there are independent roots. If at least one of them is a unit root, then from the above equation, can be obtained. When there is only one unit root, considering the remaining roots of the characteristic equation, the above equation can be expressed as:

[0083]

[0084] When has only one unit root, the values of the remaining roots are all less than 1. At the same time, the following is the characteristic equation of the difference time series model, that is:

[0085]

[0086] Since all the roots in the above equation are less than 1, the first-order difference is stationary. To make be integrated of order one, the condition in the time series model is necessary. If the form of the time series regression model includes a constant term or a trend term, additional assumptions and test statistics are required. The stationarity test process depends on the estimation of the following model:

[0087]

[0088] Among them, is the constant term, is the trend coefficient. In summary, the purpose of the stationarity test process is to estimate the parameters in the characteristic equation of the time series model and test the null hypothesis 0. The statistic The test process is as follows:

[0089]

[0090] where, is the least squares estimate of is the variance estimate of. The result of the hypothesis test is obtained by comparing with the critical value in the stationarity test statistical table. If the hypothesis is accepted, the time series ; if the hypothesis is rejected, continue to perform the stationarity test process on each order difference of until the integration order of is determined.

[0091] When performing time series analysis, the stationarity test is whether there is a long-term equilibrium (or stationary linear combination) among all elements of more than two variables. When there are only two time series variables, the E-G two-step test method based on regression residuals can be used; when the number of time series variables is greater than 2, the Johansen test method can be used.

[0092] where, after the non-stationary order of the time series variables is determined, the variables are transformed to the same order to obtain a stationary residual sequence. Therefore, using the E-G two-step test method based on regression residuals, the test process is as follows:

[0093] In the first step, estimate the stationarity regression model to obtain a stationary residual sequence. The stationarity regression model is expressed as:

[0094]

[0095] In the formula: In the formula: is the time series variable; is the number of variables in the regression model; is the regression coefficient of the variable; is the stationarity error. Assuming that all time series variables are non-stationary and conform to , then a stationary residual sequence can be obtained:

[0096]

[0097] where, is the stationary residual sequence.

[0098] When time series variables are stationary, they have a common trend of change, forming a long-term stationary relationship.

[0099] In the second step, perform a unit root test on the stationary residual series.

[0100] The result of the stationarity test of the residual series represents the result of the stationarity test of the time series variables. Therefore:

[0101]

[0102] In the formula: is the residual series at the current moment is the first-order difference of is the value of the residual series at the previous moment; represents the first-order difference of the residual series at the previous moment; is the model coefficient corresponding to the stationarity equation; k represents a moment within time t; is the residual series of the stationarity regression process. Once the residual series is confirmed to be a stationary series after passing the stationarity test, there is a stationary relationship between the time series variables. Under the assumption of non-stationarity, the estimated residual series is , and all parameters are 0. Therefore, it can be considered that the coefficient shows whether stationarity holds, and the test is transformed into:

[0103]

[0104]

[0105] Where: is the null hypothesis, indicating that the series is non-stationary; is the alternative hypothesis, indicating that the series is stationary. When the hypothesis holds, there is no stationarity between the original time series variables; otherwise, the stationary relationship holds.

[0106] The Johansen test method is used to test whether time series variables are stationary. If so, it further determines the number of mutually independent stationary vectors and determines which stationary vector will produce the most stationary linear combination of time series variables. The premise of the Johansen test method is to estimate the parameters of the vector error correction model VECM (Vector error-correction model) of the considered time series variables using the maximum likelihood method. VECM can be expressed as:

[0107]

[0108] Where, is the first-order difference of the time series variable at time i ; is the value of the time series variable at time i -1; is the first-order difference of the time series variable at time i - j ; is a -dimensional vector containing the variables to be analyzed, where the subscript is related to the time series order; ; is the model order, i.e., the number of lag orders included in the model; is a white noise sequence with a normal distribution; is the deterministic trend term; are the model parameters. The existence of an error correction model implies that the included time series variables are stationary. If there is a true error correction model, the parameters in will describe the long-term balance between variables, while the parameters in will explain the short-term adjustments required to restore balance after any data drift. In the Johansen test, the maximum likelihood method is used to estimate the parameter matrix . In the basic principle of the Johansen test method, the problem is finally transformed into solving generalized eigenvalues. The eigenvalues can be interpreted as the squared canonical correlations between the time series variables. The larger the eigenvalue, the more stable the stationary relationship, and the optimal stationary vector corresponds to the largest eigenvalue.

[0109] By the above method, it is possible to determine whether there are abnormal data in the monitored data.

[0110] Step 300, set the early warning metric index and error weighting index for abnormal data, and calculate the fault early warning thresholds of the two indices through variable transformation.

[0111] When abnormal data is detected, this embodiment further sets the early warning metric index and error weighting index for abnormal data, and calculates the fault early warning indices of the two indices. Among them, under normal conditions, both the early warning metric index and the error weighting index are positive and do not conform to a certain specific distribution, so their probability density functions are unknown and it is impossible to use the confidence interval to determine the early warning threshold. To solve this problem, this embodiment uses variable transformation to process the index data in the normal state and transforms it into data that conforms to the standard normal distribution.

[0112] Variable transformation makes the transformed dependent variable linearly related to the independent variable through parametric transformation of the dependent variable, and the error components are independent of each other and have equal variance. After variable transformation, the normality, additivity, and homoscedasticity of the data can often be significantly improved. In addition, the calculation of variable transformation does not require prior information on the data distribution and the calculation process is relatively simple.

[0113] Variable transformation is a family of transformations, including many common data transformation functions such as logarithmic transformation, square root transformation, reciprocal transformation, etc. The calculation formula of variable transformation consists of the original data (index data), transformation parameter and the transformed data and is composed of three parts. The specific calculation formula is as follows:

[0114]

[0115] When the transformation parameter , the variable transformation is a logarithmic transformation; when , the variable transformation is a square root transformation; when , the variable transformation is a reciprocal transformation. When the original data , a translation operation needs to be performed on the data, that is, let , is the translation amount.

[0116] For the transformation parameter in variable transformation, the maximum likelihood estimation method is usually used for calculation. The specific derivation process is as follows: Let the transformed distribution , then for the transformation parameter , we can obtain and 's likelihood function as follows:

[0117]

[0118]

[0119] Among them: is the random variable after transformation; is the regression coefficient vector; is the variance, is the value at the current time with the index ; is the design matrix, which contains information about the independent variable; is the Jacobian determinant; is the value at the current time; is the value at the current time with the index ;

[0120] makes the likelihood function Regarding and the derivative function equals zero, we can obtain the maximum likelihood estimate of as follows:

[0121]

[0122] We obtain the maximum likelihood estimate of as follows:

[0123]

[0124]

[0125] wherein, RSS represents the sum of squared residuals.

[0126] In summary, the likelihood function has the following maximum value:

[0127]

[0128] Solving the maximum value of the likelihood function we can obtain the transformation parameter . We can take the logarithm of the above formula and then solve for the maximum value, as shown in the following formula:

[0129]

[0130] Simplifying the above formula, we get:

[0131]

[0132]

[0133] wherein, is the arithmetic mean of the transformed data , RSS represents the sum of squared residuals. When the original data is determined, the above formula is a one-dimensional function with respect to . By using common optimization methods such as the gradient descent method and the quasi-Newton method, it is easy to calculate the corresponding to the maximum value of the likelihood function.

[0134] After transforming the early warning metric indicators in the normal state, the data conforms to the normal distribution, and thus it is easy to obtain the data probability density function . Under the condition of determining the confidence interval , the threshold value of the fault early warning indicator is solved by using the following formula:

[0135]

[0136] Among them, is the fault warning index less than the threshold probability.

[0137] The threshold obtained by solving through the above process is the threshold of normally distributed data. In the online monitoring stage, the index of real-time data is not transformed by variables. Therefore, it is necessary to perform inverse variable transformation and then correspond to the distribution of the original index data, which is used as the warning threshold in the online monitoring stage.

[0138] Step 400, according to the identified abnormal data and its fault warning threshold, perform operation warning for the photovoltaic power station.

[0139] If the abnormal data is greater than the fault warning threshold (one exceeds or both exceed), a warning is issued.

[0140] The warning method proposed in this embodiment first removes the interference data in the monitoring data of the photovoltaic power station, which can effectively filter out invalid or irrelevant data, reduce false alarms and errors, ensure the quality of the monitoring data, and thus provide a more accurate and reliable data basis for subsequent analysis. Then, time series analysis is used to analyze abnormal data, which can more quickly and accurately identify potential problems or abnormal situations in the operation of the photovoltaic power station, so as to timely capture abnormal situations that may affect the operation of the power station and improve the sensitivity of fault detection. Set the warning metric index and error weighting index of abnormal data, and trigger a warning by calculating the warning threshold, making the warning more intelligent and flexible, capable of being adjusted according to different situations, and avoiding unnecessary false warnings and missed warnings.

[0141] Through the warning mechanism proposed in this embodiment, the photovoltaic power station can detect and handle potential problems in advance, thereby reducing the downtime and fault impact, and improving the overall reliability and operation safety of the photovoltaic power station system.

[0142] Based on the same technical concept as above, this embodiment also proposes a photovoltaic power distribution operation warning system, as Figure 2 shown. The warning system proposed in this embodiment includes:

[0143] An interference removal module, which is used to remove the interference data in the monitoring data of the photovoltaic power station.

[0144] An abnormal identification module, which determines the abnormal data in the monitoring data of the photovoltaic power station after removing the interference data through time series test analysis.

[0145] An early warning index module, which is used to set the early warning measurement index and error weighting index of abnormal data, and calculate the fault early warning threshold of the two indexes through variable transformation.

[0146] And, a comparison module, which conducts early warning on the operation of the photovoltaic power station according to the abnormal data and its fault early warning threshold. Specifically, as described in step 400 above, it will not be elaborated here too much.

[0147] Furthermore, the interference removal module further includes:

[0148] An identification unit, which identifies interference data from the monitoring data of the photovoltaic power station. The specific process is as described in step 101 above, and it will not be elaborated here too much.

[0149] And, a filtering unit, which is used to suppress the identified interference data. The specific process is as described in step 102 above, and it will not be elaborated here too much.

[0150] Furthermore, the abnormal identification module further includes:

[0151] A first time series test and analysis unit, which uses the E-G two-step test method based on regression residuals to conduct a stationarity test and analysis on the time series variables when there are only two time series variables in the monitoring data.

[0152] And, a second time series test and analysis unit, which uses the Johansen test method to conduct a stationarity test and analysis on the time series variables when the number of time series variables in the monitoring data is greater than 2.

[0153] It should be noted that the specific implementation process of each functional unit of this abnormal identification module is as described in step 200 above, and it will not be elaborated here too much.

[0154] Furthermore, the early warning index module further includes:

[0155] A variable transformation unit, which conducts variable transformation on the original index data and converts it into data that conforms to the standard normal distribution; the transformation parameters in the variable transformation are calculated using the maximum likelihood estimation method.

[0156] A threshold solving unit, which obtains the probability density function of the data that conforms to the normal distribution, and combines the confidence interval to solve for the threshold of the fault early warning index.

[0157] And, a variable inverse transformation unit, which conducts variable inverse transformation on the solved threshold to obtain the fault early warning threshold corresponding to the original index data.

[0158] It should be noted that the specific implementation processes of the functional units of the warning index module are as described in step 300 above, and will not be elaborated here.

[0159] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0160] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0161] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0162] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Therefore, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0163] The specific embodiments described above further elaborate on the purpose, technical solution, and beneficial effects of the present application. It should be understood that the above description is only the specific embodiments of the present application and is not used to limit the protection scope of the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A photovoltaic power distribution operation warning method, characterized in that, The warning method includes: Removing the interference data from the monitoring data of the photovoltaic power station; After removing the interference data, determining the abnormal data in the monitoring data of the photovoltaic power station through time series test analysis; Setting the warning metric index and error weighting index for the abnormal data, and calculating the fault warning threshold of the two indexes through variable transformation; Performing operation warning for the photovoltaic power station according to the abnormal data and its fault warning threshold; The time series test analysis process includes: When there are only two time series variables in the monitoring data, using the E-G two-step test method based on regression residuals to perform stationarity test analysis on the time series variables, specifically including: Performing stationarity test on the time series variables using the unit root test method to determine the integration order; Constructing a stationarity regression model to obtain a stationary residual sequence; Performing unit root test on the stationary residual sequence. If the stationarity hypothesis test is not established, it indicates that the data is abnormal; When the number of time series variables in the monitoring data is greater than 2, using the Johansen test method to perform stationarity test analysis on the time series variables, specifically including: Determining the number of independent stationary vectors; Constructing a vector error correction model for the considered time series variables; Estimating the parameters of the vector error correction model using the maximum likelihood method; Performing a maximum eigenvalue test on the vector error correction model. If the stationarity hypothesis test is not established, it indicates that the data is abnormal; The fault warning threshold calculation process includes: Processing the set original index data using variable transformation to convert it into data that conforms to the standard normal distribution; for the transformation parameters in the variable transformation, using the maximum likelihood estimation method for calculation; Obtaining the probability density function of the data that conforms to the normal distribution, and solving the threshold of the fault warning index in combination with the confidence interval; Performing variable inverse transformation on the solved threshold to obtain the fault warning threshold corresponding to the original index data.

2. The photovoltaic power distribution operation warning method according to claim 1, characterized in that The interference data removal process includes: Identifying the interference data from the monitoring data of the photovoltaic power station. The interference data identification process includes: first calculating the average value of the absolute values of the data waveform segments of the monitoring data; then, based on the set determination criterion, identifying the interference data; where the set determination accuracy is: if the absolute value of the data is greater than or equal to the product of the average value and the discrimination threshold, then the data is interference data, otherwise it is not interference data; Suppressing the identified interference data using the median filtering method, specifically: replacing the interference data with the median value of the data values within the filtering window.

3. A photovoltaic power distribution operation warning method according to claim 1, characterized in that, The operation warning process of the photovoltaic power station includes: Comparing the identified abnormal data with its corresponding fault warning threshold. If the abnormal data is greater than the fault warning threshold, a warning is issued.

4. A photovoltaic power distribution operation early warning system, characterized in that, The warning system includes: An interference removal module, which is used to remove the interference data from the monitoring data of the photovoltaic power station; An abnormality identification module, which determines the abnormal data in the monitoring data of the photovoltaic power station after removing the interference data through time series test analysis; An early warning index module, which is used to set the early warning measurement index and error weighting index of abnormal data, and calculate the fault early warning threshold of the two indexes through variable transformation; And a comparison module, which conducts early warning on the operation of the photovoltaic power station according to the abnormal data and its fault early warning threshold; The abnormal identification module further includes: A first time series test and analysis unit, which adopts the E-G two-step test method based on regression residuals to conduct stationarity test and analysis on time series variables when there are only two time series variables in the monitoring data. Specifically, it includes: using the unit root test method to conduct stationarity test on time series variables to determine the integration order; constructing a stationarity regression model to obtain a stationary residual sequence; conducting a unit root test on the stationary residual sequence. If the stationarity hypothesis test is not established, it indicates that the data is abnormal; And a second time series test and analysis unit, which adopts the Johansen test method to conduct stationarity test and analysis on time series variables when the number of time series variables in the monitoring data is greater than 2. Specifically, it includes: determining the number of mutually independent stationary vectors; constructing a vector error correction model for the considered time series variables; using the maximum likelihood method to estimate the parameters of the vector error correction model; conducting a maximum eigenvalue test on the vector error correction model. If the stationarity hypothesis test is not established, it indicates that the data is abnormal; The early warning index module further includes: A variable transformation unit, which conducts variable transformation on the original index data and converts it into data conforming to the standard normal distribution; the transformation parameters in the variable transformation are calculated by the maximum likelihood estimation method; A threshold solving unit, which obtains the probability density function of the data conforming to the normal distribution, and combines the confidence interval to solve the threshold of the fault early warning index; And a variable inverse transformation unit, which conducts variable inverse transformation on the solved threshold to obtain the fault early warning threshold corresponding to the original index data.

5. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the photovoltaic power distribution operation early warning method described in any one of claims 1-3.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the photovoltaic power distribution operation early warning method described in any one of claims 1-3.

Images