Power transmission line icing early warning method

Abstract

The invention provides a power transmission line icing early warning method, which comprises the steps of obtaining historical tension data and historical micrometeorological data of a power transmission line, screening out non-icing tension data from the historical tension data, screening out a plurality of strong correlation factors causing tension change of the power transmission line from the historical micrometeorological data, and carrying out early warning on icing of the power transmission line. The non-icing tension data and the strong correlation factors are used for fitting and creating a tension correction model, the non-icing tension data are substituted into the tension correction model to obtain a tension reference value, the real-time tension data are substituted into the tension correction model to obtain a tension correction value, and icing early warning is performed according to the tension reference value and the tension correction value. According to the power transmission line icing early warning method, the tension reference value is calculated according to the environmental factor change, so that the tension when the power transmission line is at the icing critical point is obtained, the accuracy of line icing detection is improved, and icing early warning can respond more quickly and timely.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric power transmission, and in particular relates to an ice-covering early warning method for a power transmission line. Background Art

[0002] Under the influence of meteorological and geographical conditions, icing phenomena such as rime or rime may form on the surface of overhead transmission lines. When icing occurs, the risk of line failure increases significantly, which may lead to accidents such as insulator flashover, line tripping, line breakage, tower collapse, conductor dancing and communication interruption, thus causing huge economic losses and serious social impact. In order to improve the safety and reliability of overhead transmission lines, local power grid companies usually monitor the icing situation of the lines in real time through icing monitoring systems, and issue icing alarm signals according to preset benchmark values, so as to timely grasp the icing status of the lines.

[0003] However, the current ice tension reference value is usually set based on the average tension value during the non-icing period. When the tension exceeds the reference value, the line is considered to be iced, and the part exceeding the reference value is the ice load. However, in actual monitoring, in addition to the weight of the conductor itself and the weight of ice, external factors such as internal tension fluctuations caused by temperature changes and wind pressure will also affect the conductor tension. If these changing factors are not taken into account, even if there is no ice or the amount of ice does not reach the alarm standard, the line tension may increase to a certain level and trigger the ice alarm, causing a false alarm. Especially during the ice period, false alarms will lead to a waste of manpower and time, and delay the investigation of hidden dangers of real ice-covered lines.

[0004] In addition, most of the existing methods for setting the ice pull reference value according to meteorological conditions directly set fixed reference meteorological factors for ice pull calculation, and then establish a prediction model based on historical data and these reference meteorological factors. Due to the different environments in different places, fixed reference meteorological factors cannot widely adapt to the complex meteorological conditions in different places. In some areas, a certain meteorological condition has little effect on the ice pull reference value, but in another area, the meteorological condition is quite different from other areas, and it may become an important factor affecting the ice pull reference value.

[0005] Therefore, it is necessary to provide a transmission line icing warning method to improve the accuracy of icing monitoring, reduce the waste of manpower and material resources caused by false alarms, and at the same time improve the response speed of icing alarms to ensure safer and more stable operation of the power system. Summary of the invention

[0006] One of the objects of the present invention is to solve at least one or more of the above-mentioned problems existing in the prior art. In other words, one of the objects of the present invention is to provide a transmission line icing warning method that meets one or more of the above-mentioned needs.

[0007] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a transmission line icing early warning method, the method comprising: S1. Obtain historical tension data and historical micro-meteorological data of the transmission line; S2, filtering out non-icing tension data from historical tension data, and removing ice-covered tension data and abnormal tension data; S3. Select several strongly correlated factors that cause the change of transmission line tension from historical micro-meteorological data; S4. Perform multivariate linear regression fitting on the non-icing tension data and the strongly correlated factors to create a tension correction model, which is used to normalize the influence of the strongly correlated factors on the tension of the transmission line; S5, substituting the non-icing tension data and the historical micro-meteorological data into the tension correction model to obtain the tension reference value; S6, collecting real-time tension data and real-time micro-meteorological data, substituting the real-time tension data and real-time micro-meteorological data into a tension correction model to obtain a tension correction value; S7. Provide ice cover warning based on the tension reference value and the tension correction value.

[0008] As a preferred implementation, performing ice cover warning according to the tension reference value and the tension correction value includes: The tension increase is calculated based on the tension baseline value and the tension correction value, and an ice cover warning is issued based on the tension increase.

[0009] As a further preferred implementation, different levels of ice cover warnings are issued according to the amount of tension increase.

[0010] As a preferred implementation, step S2 determines whether each micro-meteorological data is a strong correlation factor of the tension data by the following method; The Pearson correlation coefficient method was used to detect the linear correlation between each micro-meteorological data and the tension data to determine whether it was a strong correlation factor; The Spearman rank correlation coefficient was used to detect various micrometeorological data that did not conform to the normal distribution or had nonlinear relationships to determine whether they were strong correlation factors; Grey correlation analysis is used to detect various micrometeorological data to determine whether they are strong correlation factors.

[0011] As a preferred implementation, step S2 uses a K-shape clustering algorithm to filter out normal tension data and ice-covered tension data from the tension data and remove abnormal tension data, including: S21, standardizing the historical tension data using z standardization; S22, determining the number of clusters K; S23, using K-shape algorithm to compare and analyze the lengths of different time periods; S24, comparing the output visual clustering results with the trends of each group of data, and dividing each group of data into its corresponding cluster; S25. Retain the non-icing tension data clusters, and remove the ice-covered tension data clusters and abnormal tension data clusters.

[0012] As a preferred implementation, step S4 includes: S41, importing the non-ice tension data and strong correlation factors into the multivariate linear regression model; S42, calculating the regression coefficient of each independent variable in the multiple linear regression model; S43. Generate a tension correction model according to the regression coefficient.

[0013] As a preferred implementation, the micrometeorological data includes at least air temperature, humidity, air pressure, maximum wind speed, average wind speed, standard wind speed, maximum wind speed angle, deflection angle, wind direction, and precipitation.

[0014] As a preferred implementation, after step S1 and before step S2, the method further includes step S11, preprocessing the historical tension data and the historical micro-meteorological data.

[0015] As a further preferred implementation, the preprocessing includes: deleting duplicate data, interpolating blank data, and eliminating isolated outliers. As a preferred embodiment, it also includes: S8. Verify the accuracy of the tension correction model through historical tension data.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The transmission line icing warning method of the present invention calculates the influence of micro-meteorological factors on the conductor tension by a linear regression analysis fitting method, and the tension reference value calculated using historical data and the tension correction value calculated using real-time data both include the same micro-meteorological factors, so that the difference between the two depends only on the influence of ice gravity on the transmission line tension, thereby avoiding the false alarm caused by the increase of line tension due to factors such as wind and temperature, improving the accuracy of line icing detection, and solving the problem of insufficient accuracy faced by the prior art when detecting the icing condition of the transmission line. At the same time, the transmission line icing warning method of the present invention is optimized for data anomalies and missing problems. By adding a link to process data anomalies and missing, the problem of inaccurate warning criteria caused by data anomalies is overcome, and the warning accuracy of the model is effectively improved, thereby enhancing the practicality of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1The figure is a flow chart of a power transmission line icing warning method according to an embodiment of the present application; Figure 2 Shown is a flowchart for executing the K-shape clustering algorithm; Figure 3 Shown is a classification diagram of tension data; Figure 4 Shown is a schematic diagram of the relationship between the tension of the transmission line and micro-meteorological factors; Figure 5 The figure shows a flow chart for determining ice coverage on a transmission line; Figure 6 Shown is the tensile force data after pretreatment of the experimental line; Figure 7 Shown is the corrected tension data of the experimental line. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0019] The following description provides examples and does not limit the scope, applicability or examples set forth in the claims. Changes may be made to the functions and arrangements of the elements described without departing from the scope of the present application. Various processes or components may be appropriately omitted, substituted or added to each example. For example, the described method may be performed in an order different from the order described, and various steps may be added, omitted or combined. In addition, features described in some examples may be combined in other examples.

[0020] The embodiment of the present invention provides a transmission line icing early warning method, the flow chart of which is as follows: Figure 1 As shown, the method comprises the following steps: Step S1: Collect historical tension data and historical micro-meteorological data of local power transmission lines. These data are the tension data of power transmission lines and micro-meteorological data corresponding to time and location collected by tension sensors and monitoring equipment. They can be obtained and organized into data sets through historical data saved by local meteorological or power grid units. The organized data sets include ice tension data of transmission lines under different environmental conditions and corresponding time and location marking information.

[0021] In one embodiment of the present application, step S1 further includes step S11 , pre-processing the historical tension data and the historical micro-meteorological data.

[0022] Preprocessing includes deleting duplicate data, using linear interpolation to fill missing values ​​for blank values, and removing isolated outliers in normal waveforms. After preprocessing the original data, the new data set obtained has improved data quality, removed noise, missing values ​​and outliers in the data, corrected potential data errors, deleted unreasonable values, and avoided the accuracy of the calculation model due to erroneous data when the calculation model was generated later, thus improving the credibility of the analysis results.

[0023] Step S2: The tension data of the transmission line during the period without ice cover are screened out by the K-shape clustering algorithm, and the tension data during the period of ice cover and abnormal tension data are eliminated. The K-shape clustering algorithm is a shape-based time series clustering method, which is used to classify similar time series into the same category. The basic idea is to achieve clustering by automatically preprocessing and transforming the time series, and then using the shape-based distance (SBD) to calculate the similarity between time series. Through the clustering algorithm, the distinguishing features between the data can be extracted, so as to identify whether a certain historical tension data was collected when the transmission line was covered with ice.

[0024] The flowchart for executing the K-shape clustering algorithm is as follows Figure 2 As shown, the specific steps are as follows: S21. Use z-standardization to standardize the historical tension data.

[0025] First, calculate the mean of each time series and standard deviation , the calculation formula is as follows:

[0026]

[0027] In the formula, is the length of the time series, It is Tension data at a certain time point.

[0028] Then z normalizes each time series, calculated as follows:

[0029] In the formula, is the first The tensile force data at each time point is obtained, thereby ensuring that each dimension contributes equally to the distance calculation and avoiding the influence of different numerical ranges on the distance measurement.

[0030] S22. Determine the number of clusters K. Determine the value of K based on the change of ice tension on the transmission line according to manual experience. Take K=6, that is, divide the tension data into 6 categories, such as Figure 3As shown, they are represented by K1-K6 respectively. K1-K6 are described in detail as follows. K1: No data or significantly fewer data points. K2: Rapid periodic changes. K3: Frequent rapid mutations in the waveform. K4: Data remains unchanged for a long time. K5: Data rises and falls slowly. K6: Data fluctuates slightly.

[0031] S23. Use K-shape algorithm to compare and analyze the lengths of different time periods.

[0032] First select the time period length Segment the data to form a time series input dataset:

[0033] In the formula, So A set of time series, express The tensile force data at the moment, and the other groups are similar. The number of groups after data segmentation is , The total number of data.

[0034] Then select the initial cluster center, i.e. the initial cluster center. Randomly select the index sequence from the segment as the initial cluster center, calculate the cross-correlation measure of each pair of time series and index sequence at different time lags, and compare the degree of overlap. That is, assume that the time series and index sequence , the sample The time window is appropriately shifted to match Global alignment is used to compare the global shape features of two sequences. The formula is as follows:

[0035] In the formula, for Time series after time window shift; for All possible translations of the time window, where .like ,but The time window moves to the right units; if ,but The time window moves to the left units.

[0036] This gives a length of The cross-correlation sequence , defined as follows:

[0037] In the formula, is the number of all moves, The calculation of is based on the following formula:

[0038] The goal of the above calculation is to find The position of makes the cross-correlation sequence maximum. , The optimal displacement is Then normalization is performed to eliminate inherent distortion so that the fluctuation range is between -1 and 1. The larger the value of the mutual correlation sequence coefficient is, the higher the positive correlation between the two sequences is. The specific calculation is as follows:

[0039] Where: It is the corresponding mutual correlation value when completely similar time series do not undergo relative displacement.

[0040] Then calculate the time series similarity judgment The distance measurement, after normalization, is calculated as follows:

[0041] Where: The distance value representing the similarity between two time series ranges from 0 to 2, where 0 represents that the time series samples are completely similar. For each time series, find the corresponding minimum value in the distance value, find the time series closest to the cluster center, and assign the time series to the cluster corresponding to the minimum distance; Then calculate the average shape of all members of each cluster, calculate the average value of all cluster members at each time point, and get a time series of average shape. The calculated average shape is used as the updated cluster center of all time series in the cluster. There are k clusters of time series, each with a length of m. For each time point t, calculate the value of the average shape as follows:

[0042] In the formula, Indicates The cluster members at time The value at Finally, the distance between each time series and the new cluster center is recalculated, and the time series is reallocated to the nearest cluster according to the minimum distance; the cluster center is updated again; step S23 is repeated, and the iteration is performed until the convergence condition is met, that is, the cluster center no longer changes or the predetermined number of iterations is reached; S24, comparing the output visual clustering results with the trends of each group of data, and dividing each group of data into its corresponding cluster. S25, retaining the non-icing tension data cluster, and eliminating the ice-covered tension data cluster and the abnormal tension data cluster. Specifically, K1-K4 are abnormal tension data, K5 is ice-covered tension data, and K6 is non-icing tension data.

[0043] Existing methods rely on collected data to evaluate the icing condition of transmission lines, but do not consider the anomalies and missing data collected. The online monitoring system of transmission lines is affected by events such as strong winds, ice shedding, and uncertain factors such as sunshine fluctuations and sensor failures. The monitoring data will have singular points, outliers, noise and data missing, which directly affect the accuracy of the pre-set reference meteorological factors and the prediction results of the model. The above steps remove abnormal data through a clustering algorithm, and divide normal data into tension data in the icing state and tension data in the non-icing state, so that the tension data in the non-icing state is used alone for linear regression analysis in the subsequent steps.

[0044] Step S3, select several strongly correlated factors that cause the change of tension of local transmission lines from the historical micro-meteorological data. The strongly correlated factors can be analyzed by correlation analysis method, micro-meteorological data is analyzed according to tension data, and data items with strong correlation with tension data are found from many data items of micro-meteorological data, and then these data items are selected as strongly correlated factors affecting tension data.

[0045] Specifically, step S3 determines whether each micro-meteorological data is a strong correlation factor of the tension data by using the Pearson correlation coefficient method, the Spearman rank correlation coefficient detection method and the grey correlation analysis method.

[0046] Among them, the Pearson correlation coefficient is used to measure the correlation (linear correlation) between two variables X and Y, and its value is between -1 and 1. The Pearson correlation coefficient between two variables is defined as the quotient of the covariance and standard deviation between the two variables. The formula defines the overall correlation coefficient, and the Greek lowercase letter ρ is often used as a representative symbol. The calculation formula is as follows:

[0047] In the formula, and are the expectations of random variables X and Y, respectively. and To distinguish their standard deviations, is their covariance.

[0048] The value range of the Pearson correlation coefficient is between -1 and 1. The closer the value is to 1 or -1, the stronger the linear correlation is, and 0 indicates no correlation.

[0049] If the historical tension data and micro-meteorological data do not conform to the normal distribution or have a nonlinear relationship, the Spearman rank correlation coefficient can be used. The Pearson correlation is a statistical measure of the strength of the linear relationship between two random variables, while the Spearman correlation examines the strength of the monotonic relationship between the two, that is, the extent to which the two keep pace in the trend of increasing or decreasing, even if the proportional relationship is not maintained. The data sample value itself is used to calculate the Pearson correlation coefficient, while the data sample ranking value is used to calculate the Spearman correlation coefficient. The formula for calculating the Spearman correlation coefficient is as follows:

[0050] In the formula, It represents the difference in the rank value of the i-th data pair, and n is the total number of observed samples.

[0051] The Spearman rank correlation coefficient also has a value range between -1 and 1 and has the same meaning as the Pearson correlation coefficient, but is applicable to data with non-normal distribution or nonlinear relationships.

[0052] If the historical tension data or micro-meteorological data is incomplete or the information is incomplete, the degree of connection between them is determined by using the gray correlation analysis method, that is, the degree of correlation between the curves. The gray correlation analysis method is used to analyze the influencing factors of the comprehensive ice load in the transmission line. The basic steps are as follows: First, determine the analysis sequence. According to the analysis target, determine the comprehensive load of the transmission line as the reference sequence, set as , , is the sample size. The influencing factors of tension are the comparison sequence, set as , is the number of feature quantities.

[0053] Then the data is normalized. The order of magnitude and unit of the reference sequence and the comparison sequence are usually different. In order to facilitate analysis, each sequence needs to be normalized. The formula is as follows:

[0054] Where: is the original value of a certain feature quantity; and are the maximum and minimum values ​​of a certain characteristic quantity respectively; It is the normalized value of a certain feature.

[0055] Then calculate the correlation coefficient, which refers to the degree of correlation between the comparison sequence and the reference sequence at each moment. and In the Correlation coefficient of time for:

[0056] Where: is the resolution coefficient, , usually taken =0.5, is the minimum absolute difference between all compared sequences and the reference sequence. It is the maximum absolute difference between all compared sequences and the reference sequence. ,when <0.5, indicating a weak correlation; when 0.5≤ When <0.7, it indicates a strong correlation; When ≥ 0.7, it indicates a strong correlation.

[0057] Finally, the grey correlation degree is determined according to the above correlation coefficient. The grey correlation degree refers to the overall correlation degree between the comparison sequence and the reference sequence, that is, the average value of the correlation coefficient at each moment is taken. The calculation formula is as follows:

[0058] Where: is the grey relational degree, and its value and meaning are the same as .

[0059] Figure 4 The figure shows the influence relationship between each data item in the micro-meteorological data and the historical tension data obtained after the correlation analysis according to the three methods of the above step S3. After the above correlation analysis, it is determined that the comprehensive tension load has a strong correlation with the four factors of temperature, humidity, air pressure, and maximum wind speed; it is moderately correlated with the average wind speed, standard wind speed, maximum wind speed wind deflection angle, and deflection angle; it is weakly correlated with wind direction and precipitation intensity. Among them, the relationship between temperature and comprehensive tension load is negatively correlated, that is, when the temperature rises, the conductor expands and contracts, and the tension increases; the relationship between wind speed and comprehensive tension load is positively correlated, that is, the greater the wind speed, the greater the tension.

[0060] S4. Perform multivariate linear regression fitting on the non-ice-covered tension data selected in step S2 and several strongly correlated factors to create a tension correction model.

[0061] Multiple linear regression is a statistical method used to analyze the relationship between two or more independent variables (explanatory variables) and one dependent variable (response variable). In the multiple linear regression model, the dependent variable Y is assumed to be related to the linear combination of the independent variables X, and there is a random error term ϵ, which represents the part that cannot be explained by the model.

[0062] This embodiment provides a specific method for fitting and analyzing non-ice-covered tension data and strongly correlated factors through a multivariate linear regression model and creating a tension correction model, including the following steps: S41. The non-icing tension data and the strong correlation factors are introduced into the multivariate linear regression model, and the non-icing tension data and the strong correlation factors are fitted. The formula is as follows:

[0063] Where Y is the tension, X1, X2, …, Xn are the strongly correlated factors of the tension data, β0 is the intercept term, which represents the expected value of the dependent variable when all independent variables are zero. β1, β2, …, βn are regression coefficients, which represent the slope corresponding to the strongly correlated factors, that is, the expected average change in tension for each unit change in the strongly correlated factors. is the error term, which represents the influence of factors other than the strongly correlated factors on the tension.

[0064] S42. Calculate the regression coefficient β1, β2, ..., βn of each independent variable in the multivariate linear regression model.

[0065] S43. Substitute the regression coefficient of each independent variable into the multiple linear regression model to obtain the final tension correction model.

[0066] The method of this embodiment further includes step S44, re-substituting the non-icing tension data and the strong correlation factors to evaluate the tension correction model to ensure that the tension correction model is accurate.

[0067] Specifically, step S44 uses four evaluation indicators, namely, mean absolute percentage error (MAPE), root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R2), to comprehensively evaluate the tension correction model. Among them, MAPE provides a percentage form of prediction error and is sensitive to outliers; RMSE takes into account the size of the prediction error, but is not as intuitive as MAPE; MAE is similar to RMSE, but does not consider the direction of the error and is less sensitive to outliers; R2 measures the explanatory power of the model, and the closer the value is to 1, the stronger the explanatory power of the model. The formula is as follows: ; ; ; .

[0068] in, is the true value, is the sample mean, is the predicted value.

[0069] Step S5: Substitute the non-icing tension data and the historical micro-meteorological data into the tension correction model to obtain a tension reference value.

[0070] The above step S4 has established a linear relationship between the strong correlation factor and the tension of the transmission line in the tension correction model. Therefore, by controlling the strong correlation factor to a certain specified value through the tension correction model, the influence of the strong correlation factor on the tension of the transmission line can be normalized and adjusted, so that the difference between the tension of the transmission line under different meteorological conditions only includes the part affected by the weight of ice.

[0071] Specifically, since the intercept represents the expected value of the dependent variable when all independent variables are zero, step S5 takes the new intercept of the tension correction model after substituting the non-icing tension data as Y in the linear model and the historical micrometeorological data as X1, X2, ..., Xn in the linear model into the tension correction model. It is the reference value of tension after normalization and adjustment of strong correlation factors in non-icing conditions.

[0072] Step S6, collect real-time tension data and real-time micro-meteorological data, substitute the real-time tension data and real-time micro-meteorological data into the tension correction model, and obtain the tension correction value. Similar to the above step S5, by controlling the strong correlation factor to a certain specified value through the tension correction model, the influence of the strong correlation factor on the tension of the transmission line can be normalized and adjusted. Therefore, after substituting the real-time tension data as Y in the linear model and the real-time micro-meteorological data as X1, X2, ..., Xn in the linear model into the tension correction model, the new intercept β0 of the tension correction model is taken as the tension correction value after normalization adjustment of the strong correlation factor under real-time conditions. Compared with the tension reference value, the part affected by the strong correlation factor has been normalized, so the difference between the two is the part affected by the weight of ice.

[0073] Step S7: Calculate the tension increase caused by icing based on the tension reference value and tension correction value obtained in steps S5 and S6, so as to judge the icing state and give a suitable warning signal. When the tension correction value exceeds the tension reference value, it means that the line has begun to be covered with ice. When the tension correction value reaches 120% of the tension reference value and the tension increase exceeds 20% of the tension reference value, it means that the line is seriously covered with ice, and the system can issue an alarm to notify the relevant departments to take ice melting measures in time.

[0074] Figure 5 The figure shows the process of determining whether a transmission line is iced after the tension reference value is calculated.

[0075] This embodiment takes the ice data monitored by a power transmission line in the south as an example to illustrate. According to the method of the above embodiment, the data is first preprocessed, that is, duplicate data is deleted, and linear interpolation is used to fill the missing values ​​for blank values, and isolated abnormal values ​​of normal waveforms are eliminated. Figure 6 shown.

[0076] Substituting the non-icing tension data of the experimental line into the multivariate linear regression model, the final tension correction model is obtained as follows: Y=22799-13.1 +2.43 +5.55 +703.94 -126.03

[0077] Where Y is the tension, is the temperature data, is the humidity data, is the average wind speed data, is the wind angle data, is the skew angle data.

[0078] The evaluation results of the tension correction model are shown in Table 1.

[0079] Table 1

[0080] From the perspective of the MAPE indicator, the MAPE of the experimental route is 0.001, and the low prediction error of the route indicates that the prediction is very close to the actual value. From the perspective of the RMSE and MAE indicators, the RMSE and MAE of the experimental route are 33.58 and 21.62 respectively, indicating that the prediction results of the model are relatively accurate. The R² indicator shows the model's ability to explain data variability. The R² of the experimental route is 0.985, indicating that the model can explain most of the data variability.

[0081] Obtain the tension data of the experimental line and substitute it into the tension correction model to obtain the corrected tension correction value, such as Figure 7 shown.

[0082] The tension value after eliminating the influence of strong correlation factors is taken as the tension reference value, that is, the tension value of 22799 is taken as the tension reference value. When the tension correction value of the transmission line exceeds 120% of this value, that is, 27358.8, an alarm is issued.

[0083] This embodiment also provides a verification step S8 to ensure the accuracy of the tension correction model. After calculating the tension reference value and the tension correction value according to the historical tension data, the accuracy of the tension correction model is verified according to the ice cover record of a certain period. The specific steps are as follows: S71. Calculate confusion matrices TP, TN, FP, and FN.

[0084] TP: the number of samples with actual ice cover and predicted ice cover; TN: number of samples that are actually not covered with ice and predicted to be not covered with ice; FP: the number of samples that are actually not covered with ice but are predicted to be covered with ice; FN: The number of samples with actual ice coverage but predicted to be non-ice coverage.

[0085] S72. Calculate evaluation indicators: Accuracy = (TP+TN) / total number of samples; Precision = TP / (TP+FP); Recall = TP / (TP+FN); F1 score = 2 × (precision × recall) / (precision + recall).

[0086] Taking the ice coverage data monitored by a power transmission line in the south as an example, the accuracy of the tension reference value is verified according to step S7, TP=136, TN=194, FP=6, FN=1.

[0087] Accuracy = (136 + 194) / 337 = 97.9%; Precision = 136 / (136+6) = 95.8%; Recall rate = 136 / (136+1) = 99.3%; F1 score = 2 × (0.958 × 0.993) / (0.958 + 0.993) = 0.975.

[0088] This shows that the set tension reference value can accurately identify the icing condition of the transmission line.

[0089] The transmission line icing warning method of the present invention calculates the influence of micro-meteorological factors on the conductor tension by a linear regression analysis fitting method, and the tension reference value calculated using historical data and the tension correction value calculated using real-time data both include the same micro-meteorological factors, so that the difference between the two depends only on the influence of ice gravity on the transmission line tension, thereby avoiding the false alarm caused by the increase of line tension due to factors such as wind and temperature, improving the accuracy of line icing detection, and solving the problem of insufficient accuracy faced by the prior art when detecting the icing condition of the transmission line. At the same time, the transmission line icing warning method of the present invention is optimized for data anomalies and missing problems. By adding a link to process data anomalies and missing, the problem of inaccurate warning criteria caused by data anomalies is overcome, and the warning accuracy of the model is effectively improved, thereby enhancing the practicality of the method.

[0090] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0091] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0092] The above is only an exemplary embodiment of the present disclosure, and the scope of the present disclosure cannot be limited thereto. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure here, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field not recorded in the present disclosure. The description and examples are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A transmission line icing early warning method, characterized in that: include: S1. Obtain historical tension data and historical micro-meteorological data of the transmission line; S2, filtering out non-icing tension data from the historical tension data, and removing ice-covered tension data and abnormal tension data; S3, selecting several strongly correlated factors causing the change of transmission line tension from the historical micro-meteorological data; S4, performing multivariate linear regression fitting on the non-icing tension data and the strongly correlated factors to create a tension correction model, wherein the tension correction model is used to normalize the influence of the strongly correlated factors on the tension of the transmission line; S5, substituting the non-icing tension data and the historical micro-meteorological data into the tension correction model to obtain a tension reference value; S6, collecting real-time tension data and real-time micro-meteorological data, substituting the real-time tension data and the real-time micro-meteorological data into the tension correction model to obtain a tension correction value; S7: Perform ice cover warning according to the tension reference value and the tension correction value.

2. A power transmission line icing early warning method according to claim 1, characterized in that: The step of providing an ice warning according to the tension reference value and the tension correction value includes: The tension increase amount caused by icing is calculated according to the tension reference value and the tension correction value, and an icing warning is issued according to the tension increase amount.

3. A power transmission line icing early warning method as claimed in claim 2, characterized in that: According to the magnitude of the tension increase, different levels of ice cover warning are issued.

4. A power transmission line icing early warning method according to claim 1, characterized in that: The step S2 determines whether each micro-meteorological data is a strong correlation factor of the tension data by the following method; The Pearson correlation coefficient method is used to detect the linear correlation between each micro-meteorological data and the tension data to determine whether it is a strong correlation factor; The Spearman rank correlation coefficient was used to detect various micrometeorological data that did not conform to the normal distribution or had nonlinear relationships to determine whether they were strong correlation factors; Grey correlation analysis is used to detect various micrometeorological data to determine whether they are strong correlation factors.

5. The power transmission line icing early warning method according to claim 1, characterized in that: The step S2 selects normal tension data and ice-covered tension data from the tension data by using a K-shape clustering algorithm, and removes abnormal tension data, including: S21, performing standardization processing on the historical tension data by adopting z standardization; S22, determining the number of clusters K; S23, using K-shape algorithm to compare and analyze the lengths of different time periods; S24, comparing the output visual clustering results with the trends of each group of data, and dividing each group of data into its corresponding cluster; S25. Retain the non-icing tension data clusters, and remove the ice-covered tension data clusters and abnormal tension data clusters.

6. A power transmission line icing early warning method according to claim 1, characterized in that: The step S4 comprises: S41, importing the non-icing tension data and the strongly correlated factors into a multiple linear regression model; S42, calculating the regression coefficient of each independent variable in the multivariate linear regression model; S43: generating a tension correction model according to the regression coefficient.

7. The power transmission line icing early warning method according to claim 1, characterized in that: The micrometeorological data at least include temperature, humidity, air pressure, maximum wind speed, average wind speed, standard wind speed, maximum wind speed wind angle, deflection angle, wind direction, and precipitation.

8. The power transmission line icing early warning method according to claim 1, characterized in that: After step S1 and before step S2, the method further includes step S11 of preprocessing the historical tension data and historical micro-meteorological data.

9. A power transmission line icing early warning method as claimed in claim 8, characterized in that: The preprocessing includes: deleting duplicate data, interpolating blank data, and eliminating isolated outliers.

10. The power transmission line icing early warning method according to claim 1, characterized in that: The method also includes: S8. Verifying the accuracy of the tension correction model through the historical tension data.

Images