Low-voltage user phase identification method and device based on user feature screening

Abstract

The invention discloses a low-voltage user phase identification method and device based on user feature screening. The method comprises the following steps: acquiring load data acquired by a user and a terminal every day, archive data of the user and the terminal and power utilization data of the user from a power utilization system; preprocessing the load data acquired by the user and the terminal every day, the archive data of the user and the terminal and the power consumption data of the user to obtain the current variation of the user and the terminal; calculating a current change direction coefficient according to the current variation of the user and the terminal meeting a preset condition, calculating an improved Spearman correlation coefficient according to the current change direction coefficient, and outputting a first identification phase; and carrying out voltage feature screening on the load data of the user and the terminal which do not meet the preset condition, calculating a determination coefficient of three phases of the user and the terminal according to the screened voltage features, and outputting a second identification phase. According to the invention, the technical effect of improving the recognition efficiency and reliability by integrating the current and voltage characteristics and optimizing the algorithm can be achieved.

Description

Technical Field

[0001] The present invention relates to the field of electric power technology application, and in particular to a low-voltage user phase identification method and device based on user feature screening. Background Art

[0002] With the acceleration of industrialization and urbanization, the load scale of the power system is expanding. Especially in urban areas, with the increase in population density and the growth of economic activities, the demand for electricity continues to rise. This trend has led to a rapid increase in the number of users in the substation area, and the load management of the substation area power grid faces unprecedented challenges.

[0003] Traditionally, the management of substations in power systems mainly relies on manual inspections and manual operations, but with the expansion of substation scale and the increase in complexity, this method has become inefficient and unreliable. At the same time, traditional substation management methods often have many problems, such as incomplete information, difficulty in judging three-phase imbalance, and inability to identify complex topological relationships.

[0004] Existing phase identification technology usually only considers voltage fluctuations. At the same terminal, the A / B / C phase voltages often have a high degree of fluctuation consistency. Identification based on voltage may have large deviations. In addition, directly using the Spearman correlation coefficient to calculate the correlation between two variables will ignore the impact of the variable fluctuation direction on the data results.

[0005] With regard to the problem of low phase recognition accuracy in the prior art, no effective solution has been proposed yet. Summary of the invention

[0006] In order to solve the above technical problems, the embodiments of the present invention are intended to provide a method and device for low-voltage user phase identification based on user feature screening, so as to at least solve the problem of low accuracy of phase identification in the prior art.

[0007] The technical solution of the present invention is achieved in this way:

[0008] The embodiment of the present invention provides a low-voltage user phase identification method based on user feature screening, comprising: acquiring load data collected daily from users and terminals, archive data of users and terminals, and power consumption data of users from a power consumption system; preprocessing the load data collected daily from users and terminals, archive data of users and terminals, and power consumption data of users to obtain current changes of users and terminals; calculating a current change direction coefficient for the current changes of users and terminals that meet preset conditions, and calculating an improved Spearman correlation coefficient based on the current change direction coefficient, and outputting a first identification phase; performing voltage feature screening on the load data of users and terminals that do not meet the preset conditions, and calculating a determination coefficient of three phases of users and terminals based on the screened voltage features, and outputting a second identification phase.

[0009] Optionally, preprocessing is performed on the load data collected daily from users and terminals, the archive data of users and terminals, and the electricity usage data of users to obtain the current change of users and terminals, including: deduplicating duplicate data in the load data collected daily from users and terminals, the archive data of users and terminals, and the electricity usage data of users, to obtain the deduplicated load data collected daily from users and terminals, the archive data of users and terminals, and the electricity usage data of users; identifying abnormal load values ​​by using the 3sigma rule for the deduplicated load data collected daily from users and terminals, the archive data of users and terminals, and the electricity usage data of users, and calculating the mean and standard deviation of the three-phase current and voltage, determining that the current and voltage greater than the threshold are abnormal values, and repairing them by using a random forest model to obtain the repaired load value; determining the current change based on the repaired load value to generate the current change of users and terminals.

[0010] Optionally, the preset conditions include: determining the user's electricity consumption data to filter users whose average daily electricity consumption is greater than a first threshold, associating the user's current change value and the terminal current change value according to the load time, and filtering whether the user's current change is greater than the user's current change and the terminal's current change in the second threshold period.

[0011] Optionally, the current change direction coefficient is calculated based on the current change of the user and the terminal that meet the preset conditions, and an improved Spearman correlation coefficient is calculated based on the current change direction coefficient, and outputting the first identification phase includes: calculating the current change direction weight coefficient based on the current change of the user and the current change of the terminal; sorting the current changes to obtain the first rank of the user's current change and the second rank of the terminal's current change; calculating the rank difference based on the first rank and the second rank; calculating the improved Spearman correlation coefficient based on the current change direction weight coefficient and the rank difference; and outputting the first identification phase based on the improved Spearman correlation coefficient.

[0012] Further, optionally, calculating the current change direction weight coefficient includes:

[0013]

[0014] Among them, λ is the weight coefficient of the current change direction, the if() condition means that if the user's current change value multiplied by the terminal's phase current change value is greater than 0, and the absolute value of the user's current change value is greater than or equal to k; k is the threshold; α is the specified weight coefficient.

[0015] Optionally, the rank difference includes:

[0016] d i =r user -r ter ;

[0017] Among them, r user is the first rank, r ter is the second rank, d i is the rank difference.

[0018] Optionally, calculate the modified Spearman correlation coefficient by:

[0019]

[0020] Among them, r is the improved Spearman correlation coefficient, and n is the number of rank differences.

[0021] Optionally, the load data of users and terminals that do not meet the preset conditions are screened for voltage characteristics, and the determination coefficient of the three phases of the users and terminals is calculated based on the screened voltage characteristics, and the second identification phase is output, including: calculating the voltage difference between the users and the terminals based on the load data of users and terminals that do not meet the preset conditions; calculating the average value and standard deviation of the voltage difference; screening the voltage characteristics based on the average value and the standard deviation; correcting the user's voltage based on the average value to obtain the corrected user's voltage; calculating the determination coefficient of the three phases of the users and the terminals based on the corrected user's voltage; and outputting the second identification phase based on the determination coefficient of the three phases of the users and the terminal.

[0022] The embodiment of the present invention provides a low-voltage user phase identification device based on user feature screening, comprising: an acquisition module, used to obtain load data collected daily by users and terminals, archive data of users and terminals, and power consumption data of users from a power consumption system; a preprocessing module, used to preprocess the load data collected daily by users and terminals, archive data of users and terminals, and power consumption data of users, to obtain the current change of users and terminals; a first calculation module, used to calculate the current change direction coefficient of the current change of users and terminals that meet preset conditions, and calculate the improved Spearman correlation coefficient based on the current change direction coefficient, and output a first identification phase; a second calculation module, used to perform voltage feature screening on the load data of users and terminals that do not meet the preset conditions, and calculate the determination coefficient of the three phases of users and terminals based on the screened voltage features, and output a second identification phase.

[0023] Optionally, the preprocessing module includes: a deduplication unit, which is used to deduplicate duplicate data in the load data collected daily by users and terminals, the archive data of users and terminals, and the electricity usage data of users, so as to obtain the deduplicated load data collected daily by users and terminals, the archive data of users and terminals, and the electricity usage data of users; an abnormality identification unit, which is used to identify abnormal load values ​​by using the 3sigma rule for the deduplicated load data collected daily by users and terminals, the archive data of users and terminals, and the electricity usage data of users, and calculate the mean and standard deviation of the three-phase current and voltage, determine that the current and voltage greater than the threshold are abnormal values, and repair them through a random forest model to obtain the repaired load value; a generation unit, which is used to determine the current change based on the repaired load value, and generate the current change of the user and the terminal.

[0024] The embodiment of the present invention provides a low-voltage user phase identification method and device based on user feature screening, which obtains load data collected daily by users and terminals, archive data of users and terminals, and power consumption data of users from the power consumption system; pre-processes the load data collected daily by users and terminals, archive data of users and terminals, and power consumption data of users to obtain the current change of users and terminals; calculates the current change direction coefficient of the current change of users and terminals that meet preset conditions, and calculates the improved Spearman correlation coefficient based on the current change direction coefficient, and outputs a first identification phase; performs voltage feature screening on the load data of users and terminals that do not meet the preset conditions, and calculates the determination coefficient of the three phases of users and terminals based on the screened voltage features, and outputs a second identification phase, thereby achieving the technical effect of improving identification efficiency and reliability by integrating current and voltage features and optimization algorithms. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0026] Figure 1 A flowchart of a low-voltage user phase identification method based on user feature screening provided by an embodiment of the present invention;

[0027] Figure 2 A flow chart of another low-voltage user phase identification method based on user feature screening provided by an embodiment of the present invention;

[0028] Figure 3 A schematic diagram of a low-voltage user phase identification device based on user feature screening provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first", "second" and the like in the specification, claims and drawings of the present invention are used to distinguish different objects rather than to limit a specific order.

[0031] It should also be noted that the following embodiments of the present invention may be executed separately, or in combination with each other, and the embodiments of the present invention do not impose any specific limitations on this.

[0032] The embodiment of the present invention provides a low-voltage user phase identification method based on user feature screening. Figure 1 A flow chart of a low-voltage user phase identification method based on user feature screening provided by an embodiment of the present invention; Figure 1 As shown, the low-voltage user phase identification method based on user feature screening provided in the embodiment of the present application includes:

[0033] Step S102, acquiring load data collected daily by users and terminals, archive data of users and terminals, and power consumption data of users from the power consumption system;

[0034] Specifically, the embodiment of the present application obtains the load data collected daily by users and terminals in the electricity consumption information collection system in the past N days, the archive data of users and terminals, and the electricity consumption data of users.

[0035] User load and profile are shown in Table 1:

[0036] Table 1

[0037] Terminal Number Energy meter asset number time Meter current Meter voltage Meter ct Meter pt 100217***2 000100010020****203 20240110033000 11.094 226.6 1 1 100217***2 000100010020****203 20240110034500 3.42 227.7 1 1 100217***2 000100010020****203 20240110071500 10.34 226.4 1 1 100217***2 000100010020****203 20240110073000 4.245 226.3 1 1 100217***2 000100010020****203 20240110074500 9.914 226.3 1 1 100217***2 000100010020****203 20240110080000 3.369 225.1 1 1 100217***2 000100010020****203 20240110090000 2.714 224.1 1 1 100217***2 000100010020****203 20240110110000 7.758 222 1 1

[0038] The terminal load and files are shown in Table 2:

[0039] Table 2

[0040]

[0041] The user power consumption data is shown in Table 3:

[0042] Table 3

[0043] Energy meter asset number date Power consumption (kwh) 000100010038****019 20230807 129 000100010038****019 20230819 116 000100010038****019 20230813 124 000100010038****019 20230806 148 000100010038****019 20230821 112 000100010038****019 20230801 102 000100010038****019 20230822 112 000100010038****019 20230814 115

[0044] Step S104, pre-processing the load data collected daily from the user and the terminal, the archive data of the user and the terminal, and the power consumption data of the user to obtain the current change of the user and the terminal;

[0045] Optionally, in step S104, the load data collected daily from users and terminals, the archive data of users and terminals, and the electricity usage data of users are preprocessed to obtain the current change of users and terminals, including: deduplicating duplicate data in the load data collected daily from users and terminals, the archive data of users and terminals, and the electricity usage data of users, to obtain the deduplicated load data collected daily from users and terminals, the archive data of users and terminals, and the electricity usage data of users; identifying abnormal load values ​​by the 3sigma rule for the deduplicated load data collected daily from users and terminals, the archive data of users and terminals, and the electricity usage data of users, and calculating the mean and standard deviation of the three-phase current and voltage, determining that the current and voltage greater than the threshold are abnormal values, and repairing them by the random forest model to obtain the repaired load value; determining the current change based on the repaired load value to generate the current change of users and terminals.

[0046] Specifically, during the power data collection process, it may be affected by various interference factors, resulting in missing or abnormal data, which will have an adverse effect on the phase identification results. Therefore, it is necessary to preprocess the data, including data cleaning, identification and repair of missing and abnormal data, to ensure the quality of the original data.

[0047] First, identify duplicate data in the load. Since abnormalities in the energy meter may cause duplicate voltage and current data, this will directly affect the accuracy of the subsequent model. Therefore, it is necessary to identify and delete duplicate current and voltage records.

[0048] Secondly, identify and repair outliers in the load data. Use the 3sigma rule to identify abnormal load values, calculate the mean and standard deviation of A / B / C phase current and A / B / C phase voltage, identify current and voltage values ​​greater than the mean + 3*standard deviation as outliers, and use the random forest model to repair them.

[0049] According to the 3 sigma rule, most of the data (about 99.7%) are concentrated within three standard deviations around the mean, and data outside this range are considered outliers or extreme cases.

[0050] In the embodiment of the present application, the current abnormal value is: current value>current mean+3*current standard deviation;

[0051] In the embodiment of the present application, the voltage abnormal value is: voltage>voltage mean+3*voltage standard deviation.

[0052] As shown in Table 4:

[0053] Table 4

[0054] Terminal Number time Phase A current average value Standard Deviation Is it abnormal? Correction value 100215***4 20240110033000 32 57.8 54.9 no 100215***4 20240110034500 46 57.8 54.9 no 100215***4 20240110071500 56 57.8 54.9 no 100215***4 20240110073000 267 57.8 54.9 yes 53 100215***4 20240110074500 40 57.8 54.9 no 100215***4 20240110080000 42 57.8 54.9 no 100215***4 20240110090000 54 57.8 54.9 no 100215***4 20240110110000 56 57.8 54.9 no

[0055] Third, feature derivation is performed for later model construction. Based on the current change, the current change of the user and the terminal is generated. The calculation formula is as follows:

[0056] user_current = user's current current - user's last current;

[0057] ter_a = the current at terminal A at the current moment - the current at terminal A at the previous moment;

[0058] ter_b = the current at terminal B at the current moment - the current at terminal B at the previous moment;

[0059] ter_c = the current at terminal C at the current moment - the current at terminal C at the previous moment.

[0060] Step S106, calculating the current change direction coefficient based on the current change amount of the user and the terminal that meet the preset conditions, and calculating the improved Spearman correlation coefficient based on the current change direction coefficient, and outputting the first identification phase;

[0061] Optionally, the preset conditions include: determining the user's electricity consumption data to filter users whose average daily electricity consumption is greater than a first threshold, associating the user's current change value and the terminal current change value according to the load time, and filtering whether the user's current change is greater than the user's current change and the terminal's current change in the second threshold period.

[0062] Specifically, the change relationship of current is usually more obvious than that of voltage. In actual verification, the accuracy of current identification is also higher than that of voltage identification. Therefore, for users with large power consumption and large current fluctuations, the current identification model based on the improved spearman coefficient is used for identification. For the remaining users, the current identification accuracy is not high, and the voltage identification model based on the determination coefficient is used for identification.

[0063] When the current fluctuation is small, due to the influence of other users on the same line, the terminal may not experience current fluctuations in the same direction. Including the current in this period in the calculation will interfere with the accuracy of the current correlation calculation. Therefore, only the data of users with larger electricity consumption during the period of larger current fluctuations are screened for current dimension identification.

[0064] According to the user's electricity consumption data, users whose average daily electricity consumption is greater than μ (i.e., the first threshold in the embodiment of the present application) are screened, and the user current change value and the terminal current change value of such users are associated according to the load time, and the load data in the period when user_current is greater than β (i.e., the second threshold in the embodiment of the present application) is screened.

[0065] The screening results are shown in Table 5:

[0066] Table 5

[0067] Terminal Number Energy meter asset number time user_current ter_a ter_b ter_c 100217***2 000100010007****046 20240110003000 -3.599 0 14 4 100217***2 000100010007****046 20240110010000 4.939 2 2 10 100217***2 000100010007****046 20240110011500 -5.122 -38 4 -18 100217***2 000100010007****046 20240110181500 1.329 -14 2 0 100217***2 000100010007****046 20240110221500 3.271 26 78 -8 100217***2 000100010007****046 20240110223000 -3.513 -20 16 34 100217***2 000100010007****046 20240110224500 2.956 -10 -26 -2 100217***2 000100010007****046 20240110230000 -3.716 36 10 -16

[0068] Optionally, in step S106, the current change direction coefficient is calculated based on the current change of the user and the terminal that meet the preset conditions, and an improved Spearman correlation coefficient is calculated based on the current change direction coefficient, and outputting the first identification phase includes: calculating the current change direction weight coefficient based on the current change of the user and the current change of the terminal; sorting the current changes to obtain a first rank of the user's current change and a second rank of the terminal's current change; calculating the rank difference based on the first rank and the second rank; calculating the improved Spearman correlation coefficient based on the current change direction weight coefficient and the rank difference; and outputting the first identification phase based on the improved Spearman correlation coefficient.

[0069] Further, optionally, calculating the current change direction weight coefficient includes:

[0070]

[0071] Among them, λ is the weight coefficient of the current change direction, the if() condition means that if the user's current change value multiplied by the terminal's phase current change value is greater than 0, and the absolute value of the user's current change value is greater than or equal to k; k is the threshold; α is the specified weight coefficient.

[0072] Optionally, the rank difference includes:

[0073] d i =r user -r ter ;

[0074] Among them, r user is the first rank, r ter is the second rank, d i is the rank difference.

[0075] Optionally, calculate the modified Spearman correlation coefficient by:

[0076]

[0077] Among them, r is the improved Spearman correlation coefficient, and n is the number of rank differences.

[0078] Specifically, since the users in the substation and the main meter are on the same transmission line, the user current fluctuations cause the main meter current changes. When the user current in the substation increases significantly, while other users' electricity consumption is stable, the main meter current will also increase accordingly. By analyzing the relationship between the user current fluctuations and the current fluctuations of each phase of the substation main meter, the terminal phase corresponding to the user can be determined.

[0079] In the relevant technology, the spearman correlation coefficient is a non-parametric indicator used to measure the monotonic relationship between two variables. If one variable increases, the other variable will also increase (monotonic positive correlation), or if one variable increases, the other variable will decrease (monotonic negative correlation). Its value range is between -1 and 1. The closer to 1 or -1, the stronger the monotonic relationship between the two variables. 0 means that there is no monotonic relationship between the two variables. The calculation of the spearman correlation coefficient is based on the rank transformation of the data. Therefore, compared with the Pearson correlation coefficient, it is not sensitive to outliers and does not depend on the distribution of the data. It has a wider range of applicability to data. At the same time, the Pearson correlation coefficient can only calculate the linear relationship between two variables, but the spearman correlation coefficient is not limited by this, so the calculation using the spearman correlation coefficient is more practical and adaptable.

[0080] The formula for calculating the Spearman correlation coefficient is as follows:

[0081] Where n is the sample size, d i Represents the corresponding rank difference in the sample.

[0082] In the original Spearman correlation coefficient calculation process, only the monotonic relationship of the data rank size is captured, but the impact of the data change direction on the relationship between the two is not considered. However, in actual applications, the direction of the user's power change is closely related to the direction of the terminal change. For example, when the user's current increases, but the terminal current decreases, from the perspective of actual business, the correlation between the user and the terminal should be low, but the original calculation formula will ignore this part of information, resulting in deviation.

[0083] Therefore, the low-voltage user phase identification method based on user feature screening provided in the embodiment of the present application improves the algorithm effect by introducing a weight coefficient of the data change direction.

[0084] The improved Spearman correlation coefficient calculation formula is as follows:

[0085]

[0086] Among them, user_current is the change in user meter current, and ter is the change in terminal current.

[0087] The specific calculation steps are as follows:

[0088] Step 1: Calculate the weight coefficient of the current change direction.

[0089] The current change direction weight coefficient λ is calculated based on the user's current change and the terminal's current change. The formula is as follows:

[0090]

[0091] Step 2: Calculate the improved spearman coefficient.

[0092] By sorting the current changes, the rank r of the user current change value is obtained user (i.e., the first rank in the embodiment of the present application) and the rank r of the terminal A / B / C phase current change value ter (That is, the second rank in the embodiment of the present application). Then, the rank difference is obtained according to the ranks of the two: d i =r user -r ter .

[0093] The improved spearman coefficient is calculated according to the following formula:

[0094]

[0095] The calculation results are shown in Table 6:

[0096] Table 6

[0097] Energy meter asset number ra rb rc 000100010038****999 0.5619 0.9328 -0.0123 000100010038****019 0.9464 0.2285 -0.0947 000100010038****169 -0.0656 0.1883 0.9539 000100010038****705 -0.1174 0.4279 0.9360 000100010039****430 0.2865 0.4368 0.9099 000108450116****173 -0.3600 0.6137 0.9467 000108450116****019 0.1932 0.8944 -0.0341 000108450116****990 0.3277 0.8904 -0.3533 000108450119****086 0.9321 0.0184 -0.1105

[0098] Step 3: Output identification phase.

[0099]

[0100] The output results are shown in Table 7:

[0101] Table 7

[0102] meterno ra rb rc max_r Phase_c 000100010038****999 0.5619 0.9328 -0.0123 0.9328 B 000100010038****019 0.9464 0.2285 -0.0947 0.9464 A 000100010038****169 -0.0656 0.1883 0.9539 0.9539 C 000100010038****705 -0.1174 0.4279 0.9360 0.9360 C 000100010039****430 0.2865 0.4368 0.9099 0.9099 C 000108450116****173 -0.3600 0.6137 0.9467 0.9467 C 000108450116****019 0.1932 0.8944 -0.0341 0.8944 B 000108450116****990 0.3277 0.8904 -0.3533 0.8904 B 000108450119****086 0.9321 0.0184 -0.1105 0.9321 A

[0103] Step S108, screening the load data of users and terminals that do not meet the preset conditions by voltage characteristics, and calculating the determination coefficients of the three phases of the users and terminals according to the screened voltage characteristics, and outputting a second identification phase.

[0104] Optionally, in step S108, the load data of users and terminals that do not meet the preset conditions are screened for voltage characteristics, and the determination coefficients of the three phases of the users and terminals are calculated based on the screened voltage characteristics, and the second identification phase is output, including: calculating the voltage difference between the users and the terminals based on the load data of the users and terminals that do not meet the preset conditions; calculating the average value and standard deviation of the voltage difference; screening the voltage characteristics based on the average value and the standard deviation; correcting the user's voltage based on the average value to obtain the corrected user's voltage; calculating the determination coefficients of the three phases of the users and the terminals based on the corrected user's voltage; and outputting the second identification phase based on the determination coefficients of the three phases of the users and the terminals.

[0105] Specifically, for users with low or zero electricity, they are not selected in the user feature screening due to their low daily average electricity consumption and low current, so the voltage recognition model based on the decision system is used. In addition, for users whose maximum spearman coefficient calculated by the current recognition model of the improved spearman coefficient is lower than the threshold k, the recognition effect is poor, so the voltage recognition model based on the decision system is also used.

[0106] The consistency of changes between user voltage and terminal voltage is caused by the design and operating conditions of the power system. This consistency helps to maintain the stability of the power system and the normal operation of user equipment. According to the similar voltage fluctuations between terminals and users under the same line and the inconsistent voltage fluctuations between terminals and users under different lines, a voltage identification model based on the determination coefficient is established for the remaining users.

[0107] The coefficient of determination is a statistical indicator that measures the degree of fit between the fitted curve and the actual data. It indicates the degree to which the fitted curve explains the total variation. The value is between 0 and 1. The closer it is to 1, the stronger the model's ability to explain the observed data. The formula for calculating the coefficient of determination is as follows:

[0108] Among them, SSR is the regression sum of squares, SST is the total sum of squares, and SSE is the residual sum of squares.

[0109] The corrected user voltage is regarded as the predicted value, the terminal voltage is regarded as the actual value, and the correlation between the terminal voltage and the user voltage fluctuation is determined by the determination coefficient.

[0110] The specific calculation steps are as follows:

[0111] Step 1: Voltage feature screening.

[0112] Due to special circumstances such as clock anomalies or abnormal electricity meter measurement, there may be a large difference between the terminal and the user voltage at some point in time, which will affect the determination coefficient effect. Therefore, feature screening is first performed on the user and terminal voltages.

[0113] The terminal voltage is recorded as ter_v, the user voltage is recorded as user_v, and the voltage difference between the terminal and the user is calculated: diff = ter_v-user_v

[0114] Calculate the average value avg_diff and standard deviation std_diff of diff, and select voltage data with diff≥avg_diff-3*std_diff and diff≤avg_diff+3*std_diff for modeling.

[0115] Step 2: User voltage correction.

[0116] Since there is a voltage drop between the substation terminal and the user, the user voltage needs to be corrected. The formula is as follows:

[0117] user_v_re=user_v+avg_diff

[0118] Step 3: Calculate the determination coefficient between the user and terminal A / B / C.

[0119] Residual error=ter_v-user_v_re

[0120] avg_ter_v=Avg(ter_v)

[0121] SST=sum((ter_v-avg_ter_v)*(ter_v-avg_ter_v))

[0122] SSE = sum(error*error)

[0123] Coefficient of determination R = 1-(SSE / SST).

[0124] The output results are shown in Table 8:

[0125] Table 8

[0126] Energy meter asset number Ra R Rc 000100010038****463 0.836 0.813 0.692 000100010038****591 0.473 0.633 0.414 000100010038****961 0.791 0.744 0.636 000100010038****132 0.849 0.846 0.768 000100010038****134 0.793 0.817 0.726 000100010038****296 0.844 0.583 0.311

[0127] Step 4: Output identification phase

[0128]

[0129] The output results are shown in Table 9:

[0130] Table 9

[0131] Energy meter asset number Ra R Rc max_R phase_v 000100010038****463 0.836 0.813 0.692 0.8363 A 000100010038****591 0.473 0.633 0.414 0.6329 B 000100010038****961 0.791 0.744 0.636 0.7910 A 000100010038****132 0.849 0.846 0.768 0.8495 A 000100010038****134 0.793 0.817 0.726 0.8170 B 000100010038****296 0.844 0.583 0.311 0.8437 A

[0132] In summary, in combination with step S102 to step S108, Figure 2 A flow chart of another method for identifying low-voltage user phase based on user feature screening provided by an embodiment of the present invention; Figure 2 As shown, the low-voltage user phase identification method based on user feature screening provided by an embodiment of the present invention effectively reduces the identification deviation caused by too small current fluctuation and data distribution based on the spearman coefficient calculated based on feature screening compared with the Pearson coefficient without feature screening. The following are the results of the same batch of electric energy meters identified based on the Pearson correlation coefficient without feature screening and the spearman coefficient based on feature screening.

[0133] The improved spearman coefficient recognition accuracy avoids the recognition bias caused by inconsistent fluctuation directions and improves the recognition accuracy. The recognition results based on the improved spearman coefficient and the original spearman coefficient are shown in Table 10:

[0134] Table 10

[0135]

[0136] Using the determination coefficient to identify the terminal phase effectively avoids the problem that the Pearson correlation coefficient cannot be identified because the terminal phase voltages are too similar. The results of determination coefficient identification and Pearson correlation coefficient identification are shown in Table 11:

[0137] Table 11

[0138]

[0139] The embodiment of the present invention provides a low-voltage user phase identification method based on user feature screening, which obtains load data collected daily by users and terminals, archive data of users and terminals, and power consumption data of users from the power consumption system; pre-processes the load data collected daily by users and terminals, archive data of users and terminals, and power consumption data of users to obtain current changes of users and terminals; calculates the current change direction coefficient of the current changes of users and terminals that meet preset conditions, and calculates the improved Spearman correlation coefficient based on the current change direction coefficient, and outputs a first identification phase; performs voltage feature screening on the load data of users and terminals that do not meet the preset conditions, and calculates the determination coefficient of the three phases of users and terminals based on the screened voltage features, and outputs a second identification phase, thereby achieving the technical effect of improving identification efficiency and reliability by integrating current and voltage features and optimization algorithms.

[0140] The embodiment of the present invention provides a low-voltage user phase identification device based on user feature screening, Figure 3 A schematic diagram of a low-voltage user phase identification device based on user feature screening provided by an embodiment of the present invention; Figure 3 As shown, the low-voltage user phase identification device based on user feature screening provided by the embodiment of the present invention includes: an acquisition module 32, which is used to obtain the load data collected daily by users and terminals, the archive data of users and terminals, and the power consumption data of users from the power consumption system; a preprocessing module 34, which is used to preprocess the load data collected daily by users and terminals, the archive data of users and terminals, and the power consumption data of users, to obtain the current change amount of users and terminals; a first calculation module 36, which is used to calculate the current change direction coefficient of the current change amount of users and terminals that meet the preset conditions, and calculate the improved Spearman correlation coefficient based on the current change direction coefficient, and output the first identification phase; a second calculation module 38, which is used to perform voltage feature screening on the load data of users and terminals that do not meet the preset conditions, and calculate the determination coefficient of the three phases of users and terminals based on the screened voltage features, and output the second identification phase.

[0141] Optionally, the preprocessing module 34 includes: a deduplication unit, which is used to deduplicate duplicate data in the load data collected daily by users and terminals, the archive data of users and terminals, and the electricity usage data of users, so as to obtain the deduplicated load data collected daily by users and terminals, the archive data of users and terminals, and the electricity usage data of users; an abnormality identification unit, which is used to identify abnormal load values ​​by using the 3sigma rule for the deduplicated load data collected daily by users and terminals, the archive data of users and terminals, and the electricity usage data of users, and calculate the mean and standard deviation of the three-phase current and voltage, determine that the current and voltage greater than the threshold are abnormal values, and repair them through the random forest model to obtain the repaired load value; a generation unit, which is used to determine the current change based on the repaired load value, and generate the current change amount of the user and the terminal.

[0142] The embodiment of the present invention provides a low-voltage user phase identification device based on user feature screening, which obtains load data collected daily by users and terminals, archive data of users and terminals, and power consumption data of users from the power consumption system; pre-processes the load data collected daily by users and terminals, archive data of users and terminals, and power consumption data of users to obtain current changes of users and terminals; calculates the current change direction coefficient of the current change of users and terminals that meet preset conditions, and calculates the improved Spearman correlation coefficient based on the current change direction coefficient, and outputs a first identification phase; performs voltage feature screening on the load data of users and terminals that do not meet the preset conditions, and calculates the determination coefficient of the three phases of users and terminals based on the screened voltage features, and outputs a second identification phase, thereby achieving the technical effect of improving identification efficiency and reliability by integrating current and voltage features and optimization algorithms.

[0143] The above are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention.

Claims

1. A low-voltage user phase identification method based on user feature screening, characterized in that: include: Obtaining load data collected daily by users and terminals, archive data of the users and terminals, and power consumption data of the users from the power consumption system; Preprocessing the load data collected daily from the user and the terminal, the archive data of the user and the terminal, and the power consumption data of the user to obtain the current change of the user and the terminal; Calculating a current change direction coefficient based on the current change amount of the user and the terminal that meets the preset conditions, and calculating an improved Spearman correlation coefficient based on the current change direction coefficient, and outputting a first identification phase; The load data of the user and the terminal that do not meet the preset conditions are screened for voltage characteristics, and the determination coefficients of the three phases of the user and the terminal are calculated based on the screened voltage characteristics, and a second identification phase is output.

2. The low-voltage user phase identification method based on user feature screening according to claim 1 is characterized in that: The preprocessing of the load data collected daily from the user and the terminal, the archive data of the user and the terminal, and the power consumption data of the user to obtain the current variation of the user and the terminal includes: Deduplication is performed on the load data collected daily by the user and the terminal, the archive data of the user and the terminal, and the power consumption data of the user to obtain the deduplication-free load data collected daily by the user and the terminal, the archive data of the user and the terminal, and the power consumption data of the user; The load data collected daily by the user and the terminal after deduplication, the archive data of the user and the terminal, and the power consumption data of the user are used to identify abnormal load values ​​by the 3sigma rule, and the mean and standard deviation of the three-phase current and voltage are calculated, and the current and voltage greater than the threshold are determined to be abnormal values, and they are repaired by the random forest model to obtain the repaired load value; A current change is determined according to the repaired load value, and a current change amount of the user and the terminal is generated.

3. The low-voltage user phase identification method based on user feature screening according to claim 1 is characterized in that: The preset conditions include: determining the user's electricity consumption data to filter users whose average daily electricity consumption is greater than a first threshold, associating the user's current change value and the terminal current change value according to the load time, and filtering whether the user's current change is greater than the user's current change and the terminal's current change in the second threshold period.

4. The low-voltage user phase identification method based on user feature screening according to any one of claims 1 to 3, characterized in that: The step of calculating a current change direction coefficient based on the current change amount of the user and the terminal that meets the preset condition, calculating an improved Spearman correlation coefficient based on the current change direction coefficient, and outputting a first identification phase includes: Calculating a current change direction weight coefficient according to the current change amount of the user and the current change amount of the terminal; Sorting the current changes to obtain a first rank of the current change amount of the user and a second rank of the current change amount of the terminal; Calculating a rank difference according to the first rank and the second rank; Calculating the improved Spearman correlation coefficient according to the current change direction weight coefficient and the rank difference; The first identified phase is output according to the improved Spearman correlation coefficient.

5. The low-voltage user phase identification method based on user feature screening according to claim 4 is characterized in that: The calculation of the current change direction weight coefficient includes: Among them, λ is the weight coefficient of the current change direction, the if() condition means that if the user's current change value multiplied by the terminal's phase current change value is greater than 0, and the absolute value of the user's current change value is greater than or equal to k; k is the threshold; α is the specified weight coefficient.

6. The low-voltage user phase identification method based on user feature screening according to claim 5 is characterized in that: The rank difference includes: d i ＝r user -r ter ； Among them, r user is the first rank, r ter For the second rank, d i is the rank difference.

7. The low-voltage user phase identification method based on user feature screening according to claim 6 is characterized in that: The calculating of the improved Spearman correlation coefficient comprises: Wherein, r is the improved Spearman correlation coefficient, and n is the number of rank differences.

8. The low-voltage user phase identification method based on user feature screening according to claim 1 is characterized in that: The step of screening the load data of the user and the terminal that do not meet the preset conditions by voltage characteristics, calculating the determination coefficient of the three phases of the user and the terminal according to the screened voltage characteristics, and outputting the second identification phase includes: Calculating a voltage difference between the user and the terminal according to the load data of the user and the terminal that do not meet the preset condition; Calculating the mean and standard deviation of the voltage difference; screening voltage characteristics according to the mean value and the standard deviation; Correcting the voltage of the user according to the average value to obtain a corrected voltage of the user; Calculating the determination coefficient of the three phases of the user and the terminal according to the corrected voltage of the user; The second identified phase is outputted according to the determination coefficients of the three phases of the user and the terminal.

9. A low-voltage user phase identification device based on user feature screening, characterized in that: include: An acquisition module, used to acquire load data collected daily by users and terminals, archive data of the users and terminals, and power consumption data of the users from the power consumption system; A preprocessing module, used to preprocess the load data collected daily by the user and the terminal, the archive data of the user and the terminal, and the power consumption data of the user, to obtain the current change of the user and the terminal; A first calculation module, configured to calculate a current change direction coefficient based on the current change amount between the user and the terminal that meets a preset condition, calculate an improved Spearman correlation coefficient based on the current change direction coefficient, and output a first identification phase; The second calculation module is used to screen the load data of the user and the terminal that do not meet the preset conditions based on voltage characteristics, calculate the determination coefficient of the three phases of the user and the terminal based on the screened voltage characteristics, and output a second identification phase.

10. The low-voltage user phase identification device based on user feature screening according to claim 9, characterized in that: The preprocessing module comprises: a deduplication unit, configured to deduplicate duplicate data in the load data collected daily by the user and the terminal, the archive data of the user and the terminal, and the power consumption data of the user, to obtain deduplicated load data collected daily by the user and the terminal, the archive data of the user and the terminal, and the power consumption data of the user; an abnormality identification unit, used to identify abnormal load values ​​by using the 3sigma rule for the load data collected daily by the user and the terminal after deduplication, the archive data of the user and the terminal, and the power consumption data of the user, and calculate the mean and standard deviation of the three-phase current and voltage, determine that the current and voltage greater than the threshold are abnormal values, and repair them by using the random forest model to obtain the repaired load value; A generating unit is used to determine the current change according to the repaired load value, and generate the current change amount of the user and the terminal.

Images