Zone area marker post line loss identification method based on ridge regression algorithm and double-hook curve

Abstract

The invention relates to a zone area benchmark line loss identification method based on a ridge regression algorithm and a double-hook curve in the field of electric power, and the method comprises the steps: in a data preprocessing stage, through defining a business logic rule and a quantile threshold value, eliminating obviously unreasonable and extreme fluctuation data points, and guaranteeing that subsequent analysis is based on a reliable data set; in the modeling stage, a polynomial fitting method is used for fitting line loss data of a transformer area, an optimal fitting curve is found, and model complexity is controlled through regularization in combination with a ridge regression algorithm; according to a fitting result, constructing a line loss double-hook curve to obtain a representative line loss level within a normal line loss range, and based on the constructed line loss double-hook curve, performing real-time preprocessing on newly collected data to provide a clear target and reference for transformer area line loss management; the method has the advantages of being finer in data processing, high in algorithm flexibility and high in transformer area adaptability, and the problems that an existing transformer area data processing method is insufficient in accuracy and low in efficiency are solved.

Description

Technical Field

[0001] The present invention relates to the field of electric power, and in particular to a method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double-hook curve. Background Art

[0002] The transformer district (full name: transformer district) is a basic management unit in the power system, which refers to an area consisting of a distribution transformer and the low-voltage power grid within its power supply range. The transformer district data refers to various operation and management information related to the power transformer area (i.e., transformer district). These data reflect the status, performance, and power consumption of the power system in the area. In the operation and management of the power system, transformer district data is an important basis for evaluating the operation status and efficiency of the transformer district. However, since the operation status of the transformer district is comprehensively analyzed based on a large amount of real-time data, the transformer district data is easily interfered by various factors during the actual collection process, such as load fluctuations, equipment aging, and sudden abnormal power consumption behavior of users, which easily introduces noise data. The accumulation of noise data may cause deviations in the analysis results and affect the judgment of transformer district efficiency, power safety, and line health. For example, noise data will cause abnormal fluctuations in the transformer district line loss calculation, which will affect the accuracy of the transformer district analysis and increase the operating cost of the system.

[0003] Therefore, in order to ensure the accuracy of line loss calculation, it is necessary to remove abnormal data from the substation data. Although the existing substation data processing methods have certain data processing capabilities, they mainly focus on the collection and preliminary processing of basic data, and there is still room for improvement in the accuracy and efficiency of intelligent identification and processing of abnormal data.

[0004] The above problems are worth solving. Summary of the invention

[0005] In order to overcome the shortcomings of the existing technology, the present invention provides a substation benchmark line loss identification method based on a ridge regression algorithm and a double-hook curve. By constructing a double-hook curve of the line loss interval and identifying the benchmark line loss, the substation line loss is intelligently analyzed and processed to identify and eliminate noise data. The method of the present invention can significantly improve the accuracy of the substation data by automatically identifying and processing the line loss data that exceeds a reasonable range, thereby providing a reliable data basis for substation analysis.

[0006] The technical solution of the present invention is as follows:

[0007] A method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve comprises the following steps:

[0008] Step 1: Obtain original area data from the power area database;

[0009] Step 2: preprocess the original data to remove unreasonable data points and extreme data points;

[0010] Step 3: Determine the reasonable distribution range of line loss data;

[0011] Step 4: Fit the data using polynomial fitting and ridge regression algorithm;

[0012] Use the polynomial fitting method to fit the data of the station area and find the best fitting curve; on the basis of polynomial fitting, add regularization terms to constrain the complexity of the polynomial;

[0013] Step 5: construct a line loss double hook curve according to the fitting result, wherein the line loss double hook curve includes an upper limit curve and a lower limit curve, wherein the upper limit curve and the lower limit curve are used to define the highest allowable value and the lowest allowable value of the line loss data of the transformer area, respectively;

[0014] Step 6: Identify the benchmark line loss through the line loss double-hook curve, and perform adaptive preprocessing on the substation data, and only output the data between the upper limit curve and the lower limit curve.

[0015] As a preferred technical solution of the present invention, in step 1, the original substation data includes the daily substation power supply and substation power supply.

[0016] As a preferred technical solution of the present invention, the step 2 comprises the following steps:

[0017] Step 201: Eliminate unreasonable data points;

[0018] Define the power substation line loss rules and remove abnormal data that does not conform to the substation business logic. The abnormal data includes data with a substation power input of 0, data with a substation power output of 0, and data with a substation line loss rate lower than 0%;

[0019] Step 202: Eliminate extreme value data points;

[0020] By analyzing the distribution of line loss data in the substation area, the upper and lower thresholds are set, and extreme values ​​outside the threshold range are eliminated.

[0021] Furthermore, the upper threshold is set to the 95% quantile, and the lower threshold is set to the 5% quantile, and the daily data with the line loss electricity in the substation area falling below the 5% quantile and above the 95% quantile are eliminated.

[0022] Furthermore, the calculation formula of step 202 is as follows:

[0023]

[0024] Where n is the total number of days of data in the area, X is the line loss data set, and iQ5 and i Q95 are the positions of the 5% quantile and the 95% quantile respectively; It means to take the largest integer less than or equal to the number in the brackets. It means to find the smallest integer greater than or equal to the number in the brackets.

[0025] As a preferred technical solution of the present invention, the step 3 comprises the following steps:

[0026] Step 301, calculating the quantile value;

[0027] Based on the eliminated data, calculate the upper and lower quantiles of the data;

[0028] Step 302, calculating the observation interval;

[0029] The minimum and maximum observed values ​​are calculated and counted, and the station area data between the minimum and maximum observed values ​​are determined as the observation interval.

[0030] Furthermore, in step 301, the lower quantile of the data is set to 25%, and the upper quantile is set to 75%, and the calculation formula is as follows:

[0031]

[0032] Where n is the total number of days of data in the area, X is the line loss data set, i Q25 and i Q75 The positions of the 25% and 75% quantiles respectively; It means to take the largest integer less than or equal to the number in the brackets. It means taking the smallest integer greater than or equal to the number in the brackets;

[0033] In step 302, the calculation formula of the minimum observation value is:

[0034] R min =Q25-(Q75-Q25);

[0035] The formula for calculating the maximum observed value is:

[0036] R max =Q75+(Q75-Q25).

[0037] As a preferred technical solution of the present invention, step 4 comprises the following steps:

[0038] Step 401, polynomial fitting;

[0039] Select the data of the observation area within the observation interval and set X = {x i} is the line loss data set, T={t i} is the data time set, and the polynomial fitting method is used to fit the data in the selected interval; the polynomial fitting formula is:

[0040] f(t)=a n t n +a n-1 t n-1 +…+a 1 t+a 0 ;

[0041] Among them, a n , a n-1 , ..., a 0 is the fitting coefficient, and the fitting polynomial parameters are determined by the least squares method. The formula is as follows:

[0042]

[0043] Step 402, ridge regression fitting calculation;

[0044] Using the ridge regression fitting formula:

[0045]

[0046] Here, γ is the regularization parameter used to control the size of the polynomial system.

[0047] As a preferred technical solution of the present invention, the formula of the upper bound curve in step 5 is:

[0048] L up (t) = f(t) + ∈;

[0049] The formula of the lower bound curve is:

[0050] L low (t) = f(t)-∈;

[0051] Among them, ∈ is the noise tolerance, which is used to allow a certain fluctuation range.

[0052] Preferably, the value range of ∈ is [2, 3].

[0053] Furthermore, in step 6, the line loss double hook curve is used to identify the benchmark line loss, and the calculation formula is as follows:

[0054]

[0055] Among them, P x ' s (t) is the benchmark line loss.

[0056] Furthermore, the data located between the upper limit curve and the lower limit curve are output in step 6, and the specific steps are: eliminating the data falling outside the upper limit curve and the lower limit curve, that is, eliminating the substation line loss electricity that is less than the lower limit curve and the substation line loss electricity that is greater than the upper limit curve; the remaining data is output as substation preprocessing.

[0057] The present invention also provides a computer storage medium storing a readable program, and when the program is run, the method for identifying line loss of station area benchmark poles based on ridge regression algorithm and double hook curve described in the above scheme is executed.

[0058] The present invention also provides a device, comprising one or more processors and a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors execute the method for identifying line loss of substation benchmarks based on ridge regression algorithm and double-hook curve as described in the above scheme.

[0059] The present invention also provides a station area benchmark line loss identification system based on ridge regression algorithm and double hook curve, comprising:

[0060] A data acquisition module is used to obtain original area data from the database of the power area;

[0061] The data preprocessing module is used to preprocess the original station area data to eliminate unreasonable data points and extreme value data points;

[0062] Dynamic modeling module, used to determine the reasonable distribution range of line loss data, fit the data using polynomial fitting and ridge regression algorithms, and construct a line loss double hook curve based on the fitting results;

[0063] The substation data adaptive preprocessing module is used to identify the benchmark line loss through the line loss double-hook curve, adaptively preprocess the substation data, and only output the data between the upper limit curve and the lower limit curve.

[0064] The present invention according to the above scheme has the following beneficial effects:

[0065] In the data preprocessing stage, the present invention eliminates obviously unreasonable and extremely volatile data points by defining business logic rules and quantile thresholds, ensuring that subsequent analysis is based on a reliable data set; by calculating quantile values ​​and defining observation intervals, the real distribution characteristics of line loss data are captured, effectively reducing noise interference and providing an important reference standard for the model;

[0066] In the modeling stage, the present invention uses a polynomial fitting method to fit the substation line loss data to find the best fitting curve; combined with the ridge regression algorithm, the model complexity is controlled by regularization, the robustness of the model is enhanced, a smoother and more stable fitting result is obtained, and the reliability of the model is further improved;

[0067] In the data detection and processing stage, the present invention identifies the benchmark line loss based on the double-hook curve, obtains the line loss level that is within the normal line loss range and is representative, and provides a clear goal and reference for the line loss management of the substation area; based on the constructed line loss double-hook curve, the newly collected data is pre-processed in real time. If any new line loss data falls outside the double-hook curve, it is marked as abnormal. On the contrary, if the data is within the double-hook curve, it is accepted as valid data and added to the subsequent analysis;

[0068] Therefore, the present invention achieves more effective substation analysis data by fitting the substation data, constructing a double-hook curve, and adaptively screening the substation data. Compared with the existing technology, it has the characteristics of more sophisticated data processing, high algorithm flexibility and strong substation adaptability, which solves the problems of insufficient accuracy and low efficiency in existing substation data processing methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0070] In order to better understand the purpose, technical scheme and technical effect of the present invention, the present invention is further explained in conjunction with the accompanying drawings and embodiments. It should be noted that similar numbers and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, it is stated that the embodiments described below are only used to explain the present invention and are not used to limit the present invention.

[0071] It should be noted that when an element is referred to as being "fixed on" or "disposed on" another element, it may be directly on the other element or there may also be a central element, and when an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time.

[0072] like Figure 1 As shown in FIG. 1 , a method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve includes the following steps:

[0073] Step 1: Obtain original area data from the power area database;

[0074] In this embodiment, the original area data includes the daily area power supply p gr The power output of the area gc .

[0075] Step 2: preprocess the original data to remove unreasonable data points and extreme data points;

[0076] Specifically, step 2 includes the following steps:

[0077] Step 201, eliminate unreasonable data points;

[0078] Define the line loss rule of the power supply area, and eliminate abnormal data that does not conform to the business logic of the area. The abnormal data includes data with the power supply input of the area being 0 (i.e., p gr =0), data with the power supply output of the area being 0 (i.e., p gc =0), and data with the line loss rate p xsl below 0% (i.e., ); p xs is the line loss power of the area.

[0079] Step 202, eliminate extreme data points;

[0080] Through the distribution analysis of the line loss data of the area, set the upper threshold and the lower threshold, and eliminate the extreme values outside the threshold range. In this embodiment, the upper threshold is set to the 95% quantile, and the lower threshold is set to the 5% quantile, that is, eliminate the extreme values according to the 5% quantile Q5 and the 95% quantile Q95, and further exclude the abnormal data caused by accidental factors. The calculation formula is as follows:

[0081]

[0082] where n is the total number of days of the area data, X is the line loss data set, i Q5 and i Q95 are the positions of the 5% quantile and the 95% quantile respectively; means taking the largest integer less than or equal to the number in the parentheses, means taking the smallest integer greater than or equal to the number in the parentheses. Thus, this step eliminates the daily data that satisfies the line loss p xs <Q5 and p xs >Q95.

[0083] Step 3, determine the reasonable distribution range of the line loss data;

[0084] Specifically, step 3 includes the following steps:

[0085] Step 301, calculate the quantile values;

[0086] Based on the data eliminated in the previous steps, calculate the upper quantile and the lower quantile of the data; the lower quantile of the data is set to 25%, and the upper quantile is set to 75%. The calculation formula is as follows:

[0087]

[0088]

[0089] Where n is the total number of days of data in the area, X is the line loss data set, i Q25 and i Q75 The positions of the 25% and 75% quantiles respectively; It means to take the largest integer less than or equal to the number in the brackets. It means taking the smallest integer greater than or equal to the number in the brackets;

[0090] In the step 302, the observation interval is calculated;

[0091] Calculate and count the minimum observed value R min With the maximum observed value R max , and determine the station area data between the minimum observation value and the maximum observation value as the observation interval;

[0092] The formula for calculating the minimum observed value is:

[0093] R min =Q25-(Q75-Q25);

[0094] The formula for calculating the maximum observed value is:

[0095] R max =Q75+(Q75-Q25).

[0096] Step 4: Fit the data using polynomial fitting and ridge regression algorithm;

[0097] Use the polynomial fitting method to fit the data of the station area and find the best fitting curve; on the basis of polynomial fitting, add regularization terms to constrain the complexity of the polynomial;

[0098] Specifically, step 4 includes the following steps:

[0099] Step 401, polynomial fitting;

[0100] Select the data of the station area within the observation interval, that is, R min <Line loss power in the area p xs <R max The area data is set as X = {x i} is the line loss data set, T={t i} is the data time set, and the polynomial fitting method is used to fit the data in the selected interval; the polynomial fitting formula is:

[0101] f(t)=a n t n +a n-1 t n-1 +…+a 1 t+a 0 ;

[0102] Among them, an , a n-1 , ..., a 0 is the fitting coefficient, and the fitting polynomial parameters are determined by the least squares method. The formula is as follows:

[0103]

[0104] Step 402, ridge regression fitting calculation;

[0105] In order to avoid the overfitting problem in high-order polynomial fitting, ridge regression is used for regularized fitting. On the basis of polynomial fitting, regularization terms are added to constrain the complexity of the polynomial. The ridge regression fitting formula is:

[0106]

[0107] Here, γ is the regularization parameter used to control the size of the polynomial system.

[0108] Step 5: construct a line loss double hook curve according to the fitting result, wherein the line loss double hook curve includes an upper limit curve and a lower limit curve, wherein the upper limit curve and the lower limit curve are used to define the highest allowable value and the lowest allowable value of the line loss data of the transformer area, respectively;

[0109] Specifically, step 5 is to construct a line loss double hook curve of the station area data according to the results of polynomial fitting and ridge regression, and the curve consists of an upper limit curve and a lower limit curve; and the formula of the upper limit curve is:

[0110] L up (t) = f(t) + ∈;

[0111] The formula of the lower bound curve is:

[0112] L low (t) = f(t)-∈;

[0113] Wherein, ∈ is the noise tolerance, which is used to allow a certain fluctuation range. Preferably, the value range of ∈ is [2, 3].

[0114] Step 6: Identify the benchmark line loss through the line loss double-hook curve, and perform adaptive preprocessing on the substation data, and only output the data between the upper limit curve and the lower limit curve.

[0115] Specifically, the line loss of the benchmark pole is identified by the line loss double hook curve, and its calculation formula is as follows:

[0116]

[0117] Among them, P x ' s (t) is the benchmark line loss.

[0118] Adaptive preprocessing of the area data in step 6: Use the line loss double hook curve constructed by the area data to adaptively preprocess the area data. Any data falling outside the upper and lower boundary curves is regarded as line loss noise data, that is, the area line loss power p is eliminated. xs <L low (t) and p xs >L up (t) daily data, and the remaining data are used as the final output of the station area preprocessing.

[0119] In summary, unlike traditional line loss data processing methods, the present invention combines polynomial fitting with ridge regression algorithms, so that the processing process achieves a good balance between accuracy and robustness, avoiding the problems of overfitting and excessive model complexity. In traditional methods, the screening of line loss data often relies on fixed thresholds or simple empirical rules, while the present invention uses dynamically generated line loss double-hook curves to analyze the line loss interval boundaries and identify benchmark line losses, effectively improving the level of intelligence and accuracy of processing. When analyzing the conditions of different substations, compared with the traditional methods that require specific manual analysis, the present invention realizes full-process automation from data screening, fitting to preprocessing, greatly improving efficiency and ensuring the objectivity and consistency of the processing process.

[0120] It can be seen that the present invention can accurately identify abnormal fluctuations in the substation data by constructing a line loss double-hook curve, thereby effectively reducing the impact of noise data on line loss calculations. The method of the present invention is applicable to various types of substations and can be adaptively adapted according to the power consumption characteristics of different substations. It ensures that it can be effectively applied in large-scale substations or newly built substations, and improves compatibility and practicality. The present invention utilizes a data-driven intelligent detection and correction mechanism to reduce factors of human intervention and ensure the objectivity of data processing. The objectivity of data analysis is enhanced through automated analysis and processing based on the line loss double-hook curve.

[0121] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0122] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for those of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve, characterized in that: The following steps are involved: Step 1: Obtain original area data from the power area database; Step 2: preprocess the original data to remove unreasonable data points and extreme data points; Step 3: Determine the reasonable distribution range of line loss data; Step 4: Fit the data using polynomial fitting and ridge regression algorithm; Use the polynomial fitting method to fit the data of the station area and find the best fitting curve; on the basis of polynomial fitting, add regularization terms to constrain the complexity of the polynomial; Step 5: construct a line loss double hook curve according to the fitting result, wherein the line loss double hook curve includes an upper limit curve and a lower limit curve, wherein the upper limit curve and the lower limit curve are used to define the highest allowable value and the lowest allowable value of the line loss data of the transformer area respectively; Step 6: Identify the benchmark line loss through the line loss double-hook curve, and perform adaptive preprocessing on the substation data, and only output the data between the upper limit curve and the lower limit curve.

2. The method for identifying line loss of a station area benchmark based on a ridge regression algorithm and a double hook curve according to claim 1 is characterized in that: In step 1, the original area data includes the daily area power supply and area power supply.

3. The method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve according to claim 1, characterized in that: The step 2 comprises the following steps: Step 201: Eliminate unreasonable data points; Define the power substation line loss rules and remove abnormal data that does not conform to the substation business logic. The abnormal data includes data with a substation power input of 0, data with a substation power output of 0, and data with a substation line loss rate lower than 0%; Step 202: Eliminate extreme value data points; By analyzing the distribution of line loss data in the substation area, the upper and lower thresholds are set, and extreme values ​​outside the threshold range are eliminated.

4. The method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve according to claim 3 is characterized in that: The upper threshold is set to the 95% quantile, and the lower threshold is set to the 5% quantile, and the daily data with the line loss electricity in the substation area falling below the 5% quantile and above the 95% quantile are eliminated.

5. The method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve according to claim 1, characterized in that: The step 3 comprises the following steps: Step 301, calculating the quantile value; Based on the eliminated data, calculate the upper and lower quantiles of the data; Step 302, calculating the observation interval; The minimum and maximum observed values ​​are calculated and counted, and the area data between the minimum and maximum observed values ​​are determined as the observation interval.

6. The method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve according to claim 5, characterized in that: In step 301, the lower quantile of the data is set to 25%, and the upper quantile is set to 75%. The calculation formula is as follows: Where n is the total number of days of data in the area, X is the line loss dataset, and i Q25 and i Q75 The positions of the 25% and 75% quantiles respectively; It means to take the largest integer less than or equal to the number in the brackets. It means taking the smallest integer greater than or equal to the number in the brackets; In step 302, the calculation formula of the minimum observation value is: R min =Q25-(Q75-Q25); The formula for calculating the maximum observed value is: R max =Q75+(Q75-Q25)。 7. The method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve according to claim 1, characterized in that: The step 4 comprises the following steps: Step 401, polynomial fitting; Select the data of the observation area within the observation interval and set X = {x i } is the line loss data set, T={t i } is the data time set, and the polynomial fitting method is used to fit the data in the selected interval; the polynomial fitting formula is: f(t)=a n t n +a n-1 t n-1 +…+a1t+a0; Among them, a n , a n-1 , ..., a0 is the fitting coefficient, and the fitting polynomial parameters are determined by the least squares method. The formula is as follows: Step 402, ridge regression fitting calculation; Using the ridge regression fitting formula: Here, γ is the regularization parameter used to control the size of the polynomial system.

8. The method for identifying line loss of a station area benchmark based on a ridge regression algorithm and a double hook curve according to claim 1, characterized in that: In step 5, the formula of the upper bound curve is: L up (t)=f(t)+∈; The formula of the lower bound curve is: L low (t)=f(t)-∈; Among them, ∈ is the noise tolerance, which is used to allow a certain fluctuation range.

9. The method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve according to claim 8, characterized in that: In step 6, the line loss of the benchmark pole is identified by the line loss double hook curve, and the calculation formula is as follows: Among them, P x ′ s (t) is the benchmark line loss.

10. The method for identifying line loss of a substation benchmark based on a ridge regression algorithm and a double hook curve according to claim 9, characterized in that: In step 6, the data between the upper boundary curve and the lower boundary curve are outputted, and the specific steps are as follows: Eliminate the data that falls outside the upper and lower boundary curves, that is, eliminate the line loss electricity of the substation area that is less than the lower boundary curve and the line loss electricity of the substation area that is greater than the upper boundary curve; the remaining data is used as the substation preprocessing output.

Images