Phase identification method and device based on intelligent fusion terminal, equipment and medium

Abstract

The application provides a phase identification method and device based on an intelligent fusion terminal, equipment and a medium, and belongs to the technical field of data analysis. The method comprises the following steps: acquiring single-phase voltage time series data through a smart meter of a target user, and acquiring three-phase voltage time series data through a total meter at the outlet of a transformer in a transformer area; determining a user load type based on historical power consumption data of the target user; if the user load type is a stable load type, calculating a correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on a first attention weight coefficient; the first attention weight coefficient comprises a first stable period weight; if the user load type is a fluctuating load type, performing a first operation to obtain the correlation coefficient; and determining a target phase corresponding to the target user based on the correlation coefficient. The application can improve the accuracy of user phase identification and adapt to scenarios with a large number of transformer area users and frequent changes.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of data analysis, and more particularly to a phase recognition method and device based on an intelligent fusion terminal, equipment and a medium. BACKGROUND

[0002] A low-voltage distribution area refers to a unit directly supplying power to users at the end of a power distribution network, usually taking a distribution transformer as the core and covering residents, commercial or industrial users in a certain area; phase recognition is a technical process for determining that a user's smart meter is actually connected to a certain phase of the A, B or C three-phase distribution transformer, and its result is the core data basis for achieving three-phase load balancing control, reducing line loss and ensuring power supply quality; an intelligent fusion terminal is a core device for collecting and processing transformer outlet total meter and user meter data in the area;

[0003] In the operation and maintenance of a low-voltage distribution area, due to incomplete early records, frequent on-site changes such as user capacity expansion and modification, and other reasons, the actual power supply phase of a large number of users is inconsistent with the marketing system archives, which leads to ineffective balancing of three-phase loads, causes abnormal increase of line loss and decline of power supply voltage quality, and when a fault occurs, the fault phase user cannot be quickly located, which seriously affects the operation and maintenance efficiency.

[0004] At present, the existing phase recognition technology cannot adapt to the scene of a large number of users and frequent changes in the area, has low accuracy in recognizing the phase of a user, and cannot meet the high-precision operation and maintenance requirements of the area. SUMMARY

[0005] The purpose of the present application is to provide a phase recognition method and device based on an intelligent fusion terminal, equipment and a medium to improve the accuracy of user phase recognition and adapt to the scene of a large number of users and frequent changes in the area.

[0006] The first aspect of the embodiment of the present application provides a phase recognition method based on an intelligent fusion terminal, comprising:

[0007] Obtaining single-phase voltage time series data through a smart meter of a target user and three-phase voltage time series data through an area transformer outlet total meter;

[0008] Determining a user load type based on historical power consumption data of the target user;

[0009] If the user load type is a stable load type, calculating a correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on a first attention weight coefficient; the first attention weight coefficient includes a first stable period weight; the first stable period weight is the weight of stable period data in the single-phase voltage time series data;

[0010] If the user load type is a fluctuating load type, performing a first operation to obtain the correlation coefficient;

[0011] determine the target phase corresponding to the target user based on the correlation coefficient;

[0012] The first operation comprises:

[0013] extract the voltage fluctuation frequency and the voltage fluctuation amplitude based on the single-phase voltage time series data;

[0014] If the voltage fluctuation frequency exceeds the first fluctuation frequency threshold and / or the voltage fluctuation amplitude exceeds the first fluctuation amplitude threshold, determine the non-fluctuation period based on the single-phase voltage time series data; extract the target single-phase voltage data corresponding to the non-fluctuation period from the single-phase voltage time series data, and extract the target three-phase voltage data corresponding to the non-fluctuation period from the three-phase voltage time series data; and calculate the correlation coefficient of the target single-phase voltage data and the target three-phase voltage data;

[0015] If the voltage fluctuation frequency does not exceed the first fluctuation frequency threshold and the voltage fluctuation amplitude does not exceed the first fluctuation amplitude threshold, calculate the correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on the second attention weight coefficient; the second attention weight coefficient comprises a second stationary period weight; the second stationary period weight is greater than the first stationary period weight.

[0016] In a second aspect, the embodiment of the application provides a phase recognition device based on an intelligent fusion terminal, comprising:

[0017] The data acquisition module is configured to acquire the single-phase voltage time series data through the smart meter of the target user, and acquire the three-phase voltage time series data through the outlet total meter of the transformer of the transformer area;

[0018] The user division module is configured to determine the user load type based on the historical power consumption data of the target user;

[0019] The first phase recognition module is configured to, if the user load type is the stationary load type, calculate the correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on the first attention weight coefficient; the first attention weight coefficient comprises a first stationary period weight; the first stationary period weight is the weight of the stationary period data in the single-phase voltage time series data.

[0020] The second phase recognition module is configured to, if the user load type is the fluctuation load type, perform the first operation to obtain the correlation coefficient; wherein the first operation comprises:

[0021] extract the voltage fluctuation frequency and the voltage fluctuation amplitude based on the single-phase voltage time series data;

[0022] if the voltage fluctuation frequency exceeds the first fluctuation frequency threshold and / or the voltage fluctuation amplitude exceeds the first fluctuation amplitude threshold, determine a non-fluctuation period based on the single-phase voltage time series data; extract target single-phase voltage data corresponding to the non-fluctuation period from the single-phase voltage time series data, and extract target three-phase voltage data corresponding to the non-fluctuation period from the three-phase voltage time series data; and calculate a correlation coefficient of the target single-phase voltage data and the target three-phase voltage data;

[0023] if the voltage fluctuation frequency does not exceed the first fluctuation frequency threshold and the voltage fluctuation amplitude does not exceed the first fluctuation amplitude threshold, calculate a correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on a second attention weight coefficient; the second attention weight coefficient includes a second stationary period weight; and the second stationary period weight is greater than the first stationary period weight.

[0024] a phase analysis module configured to determine a target phase corresponding to a target user based on the correlation coefficient.

[0025] In a third aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of the phase recognition method based on the intelligent fusion terminal are implemented.

[0026] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the phase recognition method based on the intelligent fusion terminal are implemented.

[0027] The phase recognition method and device based on the intelligent fusion terminal, the electronic device, and the computer readable storage medium provided by the embodiments of the present application have the following beneficial effects:

[0028] The embodiments of the present application improve the phase recognition accuracy of different users by distinguishing the user load types and then processing them in a targeted manner. For stationary load type users, the embodiments of the present application highlight the effective voltage data of the stationary period by attention weight, reducing the interference of daily small-scale power consumption fluctuations. For fluctuating load type users, the embodiments of the present application not only filter out more reliable non-fluctuation period data according to the voltage fluctuation, but also strengthen the effect of key data by a larger attention weight, effectively offsetting the influence of drastic fluctuations such as the start and stop of high-power equipment. In this way, the embodiments of the present application fully consider the characteristics of different load types, more accurately calculate the correlation coefficient of single-phase and three-phase voltage time series data, and then accurately determine the target phase corresponding to the target user.

[0029] The embodiment of the present application improves the accuracy of user phase recognition, can adapt to the scene of a large number of users in a transformer area and frequent changes, effectively solves the problems of three-phase load imbalance, high line loss, poor power supply quality and difficult fault positioning caused by inconsistent phases in the background technology, improves the operation and maintenance efficiency of the low-voltage distribution transformer area, and guarantees the stability and reliability of power supply. BRIEF DESCRIPTION OF DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0031] Figure 1 The flowchart of the phase recognition method based on the intelligent fusion terminal provided by an embodiment of the present application is shown in the figure.

[0032] Figure 2 The structural block diagram of the phase recognition device based on the intelligent fusion terminal provided by an embodiment of the present application is shown in the figure.

[0033] Figure 3 The schematic block diagram of the electronic device provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0034] In the following description, specific details such as specific system structures, techniques, etc. are presented in order to thoroughly understand the embodiments of the present application. However, it should be clear to those skilled in the art that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits and methods are omitted to avoid unnecessary details that hinder the description of the present application.

[0035] It can be understood that in the embodiments of the present application, data related to user information is involved, and when the embodiments of the present application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards.

[0036] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application and the drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0037] Reference Figure 1 , Figure 1A flowchart of a phase recognition method based on an intelligent fusion terminal is provided in an embodiment of the present application. The method can be executed by an electronic device. Specifically, the method can include S101-S105.

[0038] S101: Obtain single-phase voltage time series data from a smart meter of a target user, and obtain three-phase voltage time series data from a total meter at an outlet of a transformer in a transformer area.

[0039] In the present embodiment, the smart meter of the target user refers to an electric meter used by a single user in the transformer area to be identified in phase, which has data acquisition and communication functions. The smart meter can acquire voltage data of the phase accessed by the user. The single-phase voltage time series data refers to a sequence of voltage data continuously acquired by the smart meter of the target user at a fixed time interval. The total meter at the outlet of the transformer in the transformer area refers to an electric meter installed at the low-voltage side outlet of the distribution transformer to monitor the overall power consumption of the transformer area, which can acquire three-phase voltage data. The three-phase voltage time series data refers to a sequence of A, B, and C three-phase voltage data continuously acquired by the total meter at the outlet of the transformer in the transformer area at a fixed time interval, i.e., the three-phase voltage time series data includes a sequence of A-phase voltage data, a sequence of B-phase voltage data, and a sequence of C-phase voltage data.

[0040] The target of the present embodiment is to provide high-quality, time-aligned basic data for phase recognition. Phase recognition needs to determine the belonging phase by comparing the voltage of the user with the three-phase voltage of the total meter, so it is necessary to obtain two types of key data first. The present embodiment selects to obtain single-phase voltage time series data from the smart meter of the target user, because the electric meter directly acquires the voltage of the phase accessed by the user, which can truly reflect the voltage characteristics at the user end. The present embodiment obtains three-phase voltage time series data from the total meter at the outlet of the transformer in the transformer area, because the voltage of the total meter represents the three-phase voltage reference at the power supply side of the transformer area, which can be used as a reference standard for phase comparison. At the same time, both types of data are in time series form, which can ensure the analysis of voltage variation law under the same time dimension, avoid comparison deviation caused by time misalignment, and lay a data foundation for subsequent accurate calculation of correlation and judgment of phase.

[0041] For example, the present embodiment can implement data acquisition according to the following steps:

[0042] Firstly, the intelligent fusion terminal establishes a communication connection with the smart meter of the target user. The intelligent fusion terminal sends a data acquisition instruction to the smart meter of the target user through the power line carrier or the 4G / 5G communication module in the transformer area, and the instruction includes a preset acquisition period and data acquisition duration. After receiving the instruction, the smart meter of the target user retrieves the single-phase voltage data of the corresponding period from the data storage unit according to the instruction, associates each data point with a unique timestamp, forms single-phase voltage time series data, and uploads the data to the intelligent fusion terminal through the original communication link.

[0043] Secondly, the intelligent fusion terminal establishes a communication connection with the transformer outlet total meter. The intelligent fusion terminal sends a data acquisition instruction to the transformer outlet total meter in the same way as the user meter, and the acquisition period and acquisition duration in the instruction are completely consistent with those sent to the user meter, ensuring that the two types of data are time-aligned. After receiving the instruction, the transformer outlet total meter retrieves the A, B, and C three-phase voltage data stored therein, and each phase of the voltage data is associated with a time stamp in the same format as the user meter to form three-phase voltage time series data, which is then uploaded to the intelligent fusion terminal.

[0044] Thirdly, the intelligent fusion terminal can also perform time sequence verification on the two types of data. Specifically, the intelligent fusion terminal compares the time stamps of the single-phase voltage time series data and the three-phase voltage time series data of the target user, and eliminates data points that do not match in time. For example, if the user meter is missing some time stamp data, the corresponding three-phase voltage data of the total meter at the same time stamp is also eliminated synchronously, ensuring that the time dimensions of the remaining data are completely consistent, and finally obtaining single-phase voltage time series data and three-phase voltage time series data that can be used for subsequent phase recognition.

[0045] S102: Determine the user load type based on the historical power consumption data of the target user.

[0046] In this embodiment, the historical power consumption data of the target user refers to the power consumption data continuously collected and stored by the smart meter of the target user in the past period of time, which is used to reflect the long-term power consumption load variation characteristics of the user. The user load type refers to the classification of the user according to the power consumption load fluctuation characteristics of the user. The user load type includes a stable load type and a fluctuating load type.

[0047] The target of this embodiment is to achieve differentiated and accurate processing of phase recognition by dividing the user load type. Considering that the power consumption load characteristics of different users are significantly different, the power consumption fluctuation of stable load type users (such as residents) is small, and the power consumption fluctuation of fluctuating load type users (such as industries) is large. If a unified recognition strategy is used, it is easy to cause recognition deviation due to the mismatch of load characteristics. This embodiment determines the load type based on the historical power consumption data of the target user, which can master the user load fluctuation law in advance and provide a basis for selecting an appropriate phase recognition algorithm, avoiding the defects of unified algorithm application, improving the accuracy of phase recognition from the source, and ensuring that the transformer users with different load characteristics are adapted.

[0048] For example, this embodiment can determine the user load type based on the historical power consumption data of the target user according to the following steps:

[0049] The first step is to obtain historical power consumption data. Specifically, the intelligent fusion terminal establishes a stable communication connection with the smart meter of the target user through the power line carrier or wireless communication module in the transformer area, sends a data retrieval instruction to the smart meter, and specifies the data range as the past 30 days and the collection period as 15 minutes / time. After receiving the instruction, the smart meter extracts the historical power consumption data of the corresponding period from the local storage unit, which contains the timestamp and active power value of each collection time point, and then feeds the data back to the intelligent fusion terminal through the original communication link.

[0050] The second step is to calculate the load fluctuation characteristics. Specifically, the intelligent fusion terminal processes the historical power consumption data of the target user on a daily basis: the intelligent fusion terminal can calculate the maximum, minimum and average active power of all collection points in each day; the intelligent fusion terminal can calculate the daily load fluctuation amplitude by the ratio of (maximum active power - minimum active power) to the average active power; the intelligent fusion terminal can take the arithmetic mean of the daily load fluctuation amplitude of 30 days to obtain the average load fluctuation amplitude of the target user in 30 days, and count the number of times the active power changes more than 500W in each day, and take the 30-day average as the average load fluctuation frequency.

[0051] The third step is to determine the user load type. Specifically, the intelligent fusion terminal has built-in load type determination criteria, and the intelligent fusion terminal can preset the average load fluctuation amplitude threshold of the stable load class as 20% and the average load fluctuation frequency threshold as 2 times / hour. If the average load fluctuation amplitude of the target user is ≤20% and the average load fluctuation frequency is ≤2 times / hour, it is determined as a stable load class; if the average load fluctuation amplitude is >20% or the average load fluctuation frequency is >2 times / hour, it is determined as a fluctuating load class. The intelligent fusion terminal can bind the determination result with the meter number of the target user to provide a basis for subsequent phase recognition algorithm selection.

[0052] S103: If the user load type is a stable load class, calculate the correlation coefficient between the single-phase voltage time series data and the three-phase voltage time series data based on a first attention weight coefficient; the first attention weight coefficient includes a first stable period weight; the first stable period weight is the weight of the stable period data in the single-phase voltage time series data.

[0053] In the embodiment, the first attention weight coefficient further includes a first non-stationary period weight, and the first stationary period weight is greater than the first non-stationary period weight. The first attention weight coefficient refers to a coefficient set for giving different influences to voltage data in different periods in phase identification of the stationary load type user. The first stationary period weight is a weight specially set for the stationary period data in the single-phase voltage time series data. The first non-stationary period weight is a weight specially set for the non-stationary period data other than the stationary period data in the single-phase voltage time series data. The correlation coefficient refers to an index for measuring the consistency of the change trend of the single-phase voltage time series data and the three-phase voltage time series data.

[0054] The target of the embodiment is to improve the accuracy of phase identification of the stationary load type user. Considering that the stationary load type user has small overall power fluctuation, but still has small interference in the non-stationary period, such as turning on the light at night and using household appliances, if the voltage data in all periods are treated equally, the real phase correlation signal is easy to be diluted by the interference data. Therefore, the embodiment highlights the first stationary period weight through the first attention weight coefficient, so as to strengthen the influence of the low interference voltage data in the stationary period (such as early morning) and weaken the influence of the small interference in the non-stationary period, so as to ensure that the correlation coefficient calculation can more accurately reflect the real voltage correlation of the user and the corresponding phase of the total meter, avoid phase misjudgment caused by small interference, adapt to the power consumption characteristics of the stationary load type user, and lay a foundation for accurately determining the target phase.

[0055] For example, the embodiment can calculate the correlation coefficient based on the first attention weight coefficient according to the following steps:

[0056] Firstly, the first attention weight coefficient parameters are determined. Specifically, the intelligent fusion terminal retrieves the first attention weight coefficient preset parameters from the local configuration library. It is assumed that the first stationary period weight is set to 1.5, and the non-stationary period weight is set to 1; at the same time, the stationary period determination standard is preset as 0:00-5:00 every day, and the proportion of the sampling points with voltage difference ≤0.5V in the adjacent sampling points in this period is ≥90%, and the non-stationary period is 5:00-24:00 every day.

[0057] Secondly, the voltage data period is screened and marked. Specifically, the intelligent fusion terminal can screen the stationary period data of 0:00-5:00 and the non-stationary period data of 5:00-24:00 from the preprocessed single-phase voltage time series data according to the time stamp, and add period marks to the two types of data respectively; at the same time, the A-phase, B-phase and C-phase voltage data in the corresponding period are screened from the three-phase voltage time series data according to the same time stamp, so as to ensure that the periods of the user and the total meter data are completely aligned.

[0058] Thirdly, the weighted voltage data is calculated. Specifically, the intelligent fusion terminal can multiply the single-phase voltage data of the stationary period by the first stationary period weight 1.5, and multiply the single-phase voltage data of the non-stationary period by the non-stationary period weight 1. After obtaining the weighted voltage data of each period, the intelligent fusion terminal can combine the weighted voltage data to form complete weighted single-phase voltage time series data.

[0059] Fourthly, the correlation coefficient is calculated. Specifically, the intelligent fusion terminal can use the correlation analysis method to calculate the correlation coefficients of the weighted single-phase voltage time series data and the A-phase, B-phase and C-phase data in the three-phase voltage time series data, respectively, to obtain three correlation coefficient results, such as the correlation coefficient with the A-phase is 0.92, the correlation coefficient with the B-phase is 0.35, and the correlation coefficient with the C-phase is 0.41, which are used for subsequent target phase determination.

[0060] S104: If the user load type is a fluctuating load type, a first operation is performed to obtain a correlation coefficient. The first operation includes:

[0061] extracting a voltage fluctuation frequency and a voltage fluctuation amplitude based on the single-phase voltage time series data;

[0062] If the voltage fluctuation frequency exceeds the first fluctuation frequency threshold and / or the voltage fluctuation amplitude exceeds the first fluctuation amplitude threshold, a non-fluctuation period is determined based on the single-phase voltage time series data; target single-phase voltage data corresponding to the non-fluctuation period is extracted from the single-phase voltage time series data, and target three-phase voltage data corresponding to the non-fluctuation period is extracted from the three-phase voltage time series data; and a correlation coefficient of the target single-phase voltage data and the target three-phase voltage data is calculated.

[0063] If the voltage fluctuation frequency does not exceed the first fluctuation frequency threshold and the voltage fluctuation amplitude does not exceed the first fluctuation amplitude threshold, a correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data is calculated based on a second attention weight coefficient; the second attention weight coefficient includes a second stationary period weight; and the second stationary period weight is greater than the first stationary period weight.

[0064] In this embodiment, the non-fluctuation period is determined based on the single-phase voltage time series data, and specifically includes:

[0065] dividing the single-phase voltage time series data into a plurality of single-phase voltage data segments based on a first step length;

[0066] For each single-phase voltage data segment, a voltage segment fluctuation frequency and a voltage segment fluctuation amplitude of the single-phase voltage data segment are calculated; if the voltage segment fluctuation frequency exceeds a second fluctuation frequency threshold and / or the voltage segment fluctuation amplitude exceeds a second fluctuation amplitude threshold, a period corresponding to the single-phase voltage data segment is determined as a non-fluctuation period.

[0067] The second fluctuation frequency threshold is greater than the first fluctuation frequency threshold, and the second fluctuation amplitude threshold is greater than the first fluctuation amplitude threshold.

[0068] In the embodiment, the second attention weight coefficient further includes a second non-stationary period weight; the second stationary period weight is greater than the second non-stationary period weight.

[0069] The voltage fluctuation frequency refers to the number of times that the voltage change in unit time exceeds the basic fluctuation range in single-phase voltage time series data, and is used to represent the frequency of voltage fluctuation. The voltage fluctuation amplitude refers to the maximum difference of single voltage fluctuation in single-phase voltage time series data, and is used to represent the intensity of voltage fluctuation. The first fluctuation frequency threshold and the first fluctuation amplitude threshold refer to the critical value for judging whether the overall voltage fluctuation of the fluctuation load class user needs to screen the non-fluctuation period. The non-fluctuation period refers to the period with lower voltage fluctuation screened from the single-phase voltage time series data of the fluctuation load class user. The second attention weight coefficient refers to the weight set for calculating the correlation coefficient of the low fluctuation user in the fluctuation load class. The second stationary period weight refers to the weight of the stationary period data in the second attention weight coefficient. The second non-stationary period weight refers to the weight of the non-stationary period data in the second attention weight coefficient. The first step refers to the preset time interval for dividing the single-phase voltage time series data into segments. The voltage segment fluctuation frequency and the voltage segment fluctuation amplitude refer to the fluctuation characteristics in a single voltage data segment. The second fluctuation frequency threshold and the second fluctuation amplitude threshold refer to the critical value for judging whether the voltage data segment is a non-fluctuation period.

[0070] The target of the embodiment is to realize the accurate adaptation of the phase recognition of the fluctuation load class user, and to avoid the recognition deviation caused by the interference of the fluctuation load. Considering the complex electricity consumption characteristics of the fluctuation load class user, some users have intense fluctuation (such as frequent start and stop of industrial equipment), and some users have mild fluctuation, which needs to be processed in different scenarios. Therefore, by extracting the voltage fluctuation frequency and amplitude and comparing them with the first threshold, the overall fluctuation degree of the user can be judged. If it exceeds the threshold, it means that the overall fluctuation is intense, and the data segment needs to be split to screen the non-fluctuation period with the second fluctuation frequency threshold and the second fluctuation amplitude threshold, so as to ensure that the correlation is calculated with low interference data; if it does not exceed the threshold, it means that the fluctuation is mild, and the influence of the stationary period data is strengthened through the second attention weight (the second stationary period weight is greater than the first stationary period weight), so as to offset the slight fluctuation interference.

[0071] The hierarchical logic of the embodiment not only adapts to different fluctuation states of the fluctuation load class user, but also ensures that the effective data dominates the calculation through threshold difference and weight adjustment, which fundamentally improves the accuracy of the phase recognition of the fluctuation load class user, and can solve the defects of poor adaptability of the prior art to fluctuation users.

[0072] For example, the first operation of calculating the correlation coefficient can be performed according to the following steps:

[0073] The first step is to extract the voltage fluctuation frequency and amplitude. Specifically, the intelligent fusion terminal can count the number of times the voltage changes more than 0.8V per hour (i.e. voltage fluctuation frequency) from the pre-processed fluctuation load class user single-phase voltage time series data, and calculate the maximum voltage difference per hour (i.e. voltage fluctuation amplitude), obtaining 24 groups of hour-level voltage fluctuation frequency and voltage fluctuation amplitude. The 24 groups of voltage fluctuation frequency are summed to obtain the overall voltage fluctuation frequency of the single-phase voltage time series data, and the maximum voltage fluctuation amplitude is calculated based on the 24 groups of voltage fluctuation amplitude as the overall voltage fluctuation amplitude of the single-phase voltage time series data.

[0074] The second step is to compare with the first threshold value. Specifically, the intelligent fusion terminal can call the preset first fluctuation frequency threshold value 50 times / day and the first fluctuation amplitude threshold value 2V. Assuming that the voltage fluctuation frequency of a user for 24 hours is 60 times / day (exceeding the first frequency threshold), or the average fluctuation amplitude is 2.5V (exceeding the first amplitude threshold), the third step is executed; if the average fluctuation frequency of a user is 40 times / day and the average amplitude is 1.8V (both of which do not exceed the first threshold value), the fifth step is executed.

[0075] The third step is to divide the voltage data segments. Specifically, the intelligent fusion terminal can divide the 24-hour single-phase voltage time series data into 24 voltage data segments with a first step of 1 hour, each segment containing 60 voltage sampling points, and label each segment with a number (such as segment 1 corresponding to 0:00-1:00).

[0076] The fourth step is to screen the non-fluctuation period and calculate the correlation. Specifically, the intelligent fusion terminal can call the second fluctuation frequency threshold value 8 times / hour and the second fluctuation amplitude threshold value 3V, and calculate the voltage segment fluctuation frequency and voltage segment fluctuation amplitude for each segment. If the segment 3 (2:00-3:00) fluctuation frequency is 2 times / hour (does not exceed the second frequency threshold) and the amplitude is 0.6V (does not exceed the second amplitude threshold), it is determined that this period is a non-fluctuation period. The intelligent fusion terminal can extract the data corresponding to all non-fluctuation periods from the single-phase voltage data as the target single-phase voltage data, and extract the data corresponding to all non-fluctuation periods from the three-phase voltage data as the target three-phase voltage data, and then use the correlation analysis method to calculate the correlation coefficient of the two.

[0077] In the fifth step, the correlation is calculated based on the second attention weight. Specifically, the intelligent fusion terminal can call the second attention weight coefficient, assuming that the second stationary period weight is 2.0 (greater than the first stationary period weight 1.5) and the non-stationary period weight is 1.0; the intelligent fusion terminal can filter the stationary period data corresponding to 0:00-6:00 in the single-phase voltage data according to the preset stationary period (0:00-6:00), multiply by the weight 2.0, and multiply the non-stationary period data by the weight 1.0; the same weighting processing is performed on the three-phase voltage data, and the correlation coefficient of the weighted single-phase data and the three-phase data is calculated, which is used for subsequent target phase determination.

[0078] In the fifth step, the correlation is calculated based on the second attention weight. Specifically, the intelligent fusion terminal can call the second attention weight coefficient, assuming that the second stationary period weight is 2.0 (greater than the first stationary period weight 1.5) and the non-stationary period weight is 1.0; the intelligent fusion terminal can filter the stationary period data corresponding to 0:00-6:00 in the single-phase voltage data according to the preset stationary period (0:00-6:00), multiply by the weight 2.0, and multiply the non-stationary period data by the weight 1.0; the same weighting processing is performed on the three-phase voltage data, and the correlation coefficient of the weighted single-phase data and the three-phase data is calculated, which is used for subsequent target phase determination.

[0079] In the fifth step, the correlation is calculated based on the second attention weight. Specifically, the intelligent fusion terminal can call the second attention weight coefficient, assuming that the second stationary period weight is 2.0 (greater than the first stationary period weight 1.5) and the non-stationary period weight is 1.0; the intelligent fusion terminal can filter the stationary period data corresponding to 0:00-6:00 in the single-phase voltage data according to the preset stationary period (0:00-6:00), multiply by the weight 2.0, and multiply the non-stationary period data by the weight 1.0; the same weighting processing is performed on the three-phase voltage data, and the correlation coefficient of the weighted single-phase data and the three-phase data is calculated, which is used for subsequent target phase determination.

[0080] In the fifth step, the correlation is calculated based on the second attention weight. Specifically, the intelligent fusion terminal can call the second attention weight coefficient, assuming that the second stationary period weight is 2.0 (greater than the first stationary period weight 1.5) and the non-stationary period weight is 1.0; the intelligent fusion terminal can filter the stationary period data corresponding to 0:00-6:00 in the single-phase voltage data according to the preset stationary period (0:00-6:00), multiply by the weight 2.0, and multiply the non-stationary period data by the weight 1.0; the same weighting processing is performed on the three-phase voltage data, and the correlation coefficient of the weighted single-phase data and the three-phase data is calculated, which is used for subsequent target phase determination.

[0081] The closer the correlation coefficient is to 1, the more synchronized the user voltage and the corresponding phase voltage of the total meter are, and the higher the probability that the user accesses the phase is. Selecting the phase corresponding to the maximum value as the target phase can directly lock the phase that best meets the actual power supply of the user, avoid complex judgment logic, ensure that the determination result is intuitive and reliable, and meet the core needs of phase recognition.

[0082] In the fifth step, the correlation is calculated based on the second attention weight. Specifically, the intelligent fusion terminal can call the second attention weight coefficient, assuming that the second stationary period weight is 2.0 (greater than the first stationary period weight 1.5) and the non-stationary period weight is 1.0; the intelligent fusion terminal can filter the stationary period data corresponding to 0:00-6:00 in the single-phase voltage data according to the preset stationary period (0:00-6:00), multiply by the weight 2.0, and multiply the non-stationary period data by the weight 1.0; the same weighting processing is performed on the three-phase voltage data, and the correlation coefficient of the weighted single-phase data and the three-phase data is calculated, which is used for subsequent target phase determination.

[0083] The intelligent fusion terminal can extract the correlation coefficient data calculated for the target user, which includes the target user number, the correlation coefficient with the total meter A phase, the correlation coefficient with the total meter B phase, and the correlation coefficient with the total meter C phase. For example, the coefficients corresponding to a certain user are 0.91, 0.33, and 0.42, respectively. The intelligent fusion terminal can compare the values of the three groups of correlation coefficients and find the maximum value. In the above example, the maximum value is 0.91, and the maximum value is associated with the correlation coefficient of the total meter A phase. The intelligent fusion terminal can determine that the target phase of the target user is the total meter A phase according to the total meter phase associated with the maximum value; then, the target user number and the target phase (A phase) are bound to generate a corresponding record of user-phase, and uploaded to the substation marketing system to complete the update of the target user power supply phase.

[0084] From the above, it can be concluded that the embodiment synchronously acquires single-phase voltage time series data of the target user smart meter and three-phase voltage time series data of the transformer outlet total meter through the intelligent fusion terminal, and eliminates time mismatch data points through time series verification, ensuring that the time dimensions of the two types of data are completely aligned. Compared with the problem of asynchronous data collection, missing / correction data in the prior art, the high-quality, time-series consistent basic data provided by the method avoids the comparison deviation caused by time misalignment, and provides a premise guarantee for the accuracy of subsequent correlation calculation.

[0085] The embodiment is based on the historical power consumption data of the target user, and divides the users into smooth load class and fluctuation load class by calculating the average load fluctuation amplitude and frequency, and designs the identification strategy accordingly. For the smooth load class users, the influence of low interference data in the smooth period is strengthened through the first attention weight coefficient; for the fluctuation load class users, the high fluctuation (non-fluctuation period calculation) and low fluctuation (smooth period is strengthened by the second attention weight) scenes are processed. The embodiment avoids the defect that the existing technology uniformly adapts to all users, and adapts to users with different load characteristics, thereby fundamentally reducing the identification deviation caused by the mismatch of load characteristics.

[0086] The embodiment ensures that the correlation coefficient can truly reflect the voltage correlation between the user and the corresponding phase of the total meter through differential weight design and accurate screening of non-fluctuation period, and further guarantees the result reliability through the intuitive judgment logic of taking the maximum value of the correlation coefficient corresponding phase. The accurate phase identification result can effectively correct the inconsistency between the marketing system profile and the actual phase, provide data support for three-phase load balance regulation and control, reduce the abnormal occurrence rate of line loss; at the same time, the fault phase user can be quickly located when a fault occurs, the repair time is shortened, and the efficiency of substation operation and maintenance and the power supply quality are significantly improved.

[0087] In an embodiment of the present application, the load type of the target user is determined based on the historical power consumption data of the target user, comprising:

[0088] Extracting the daily average power fluctuation coefficient, peak period power proportion, and daily average power maximum value from the historical power consumption data of the target user;

[0089] Constructing a target feature vector based on the daily average power fluctuation coefficient, peak period power proportion, and daily average power maximum value;

[0090] Calculating the first matching degree of the target feature vector and the first data set; the first data set includes a plurality of first reference users each corresponding to a feature vector, and the user load types of the plurality of first reference users are all smooth load class;

[0091] a second matching degree of the target feature vector and a second data set is calculated; the second data set includes a plurality of second reference user feature vectors corresponding to a plurality of second reference users respectively, and the user load types of the plurality of second reference users are all fluctuation load types;

[0092] If the first matching degree is greater than or equal to the second matching degree, it is determined that the user load type is a stable load type.

[0093] If the first matching degree is less than the second matching degree, it is determined that the user load type is a fluctuation load type.

[0094] In the embodiment, the voltage fluctuation frequency and the voltage fluctuation amplitude are extracted based on the single-phase voltage time series data, specifically including:

[0095] The first voltage fluctuation frequency and the first voltage fluctuation amplitude of the electricity peak period are extracted based on the single-phase voltage time series data;

[0096] The second voltage fluctuation frequency and the second voltage fluctuation amplitude of the electricity valley period are extracted based on the single-phase voltage time series data; the electricity peak period and the electricity valley period are both calculated based on the historical single-phase voltage time series data of the plurality of second reference users;

[0097] The first voltage fluctuation frequency and the second voltage fluctuation frequency are weighted and summed to obtain the voltage fluctuation frequency;

[0098] The first voltage fluctuation amplitude and the second voltage fluctuation amplitude are weighted and summed to obtain the voltage fluctuation amplitude.

[0099] In the embodiment, the daily average power fluctuation coefficient refers to an index reflecting the daily power fluctuation degree extracted from the target user historical power consumption data, which can be calculated by the ratio of the daily power extreme value difference to the average value, and is used to describe the daily fluctuation characteristics of the user load. The peak period power ratio refers to the ratio of the total power of the target user in the daily electricity peak period to the total power of the whole day, which is used to distinguish the electricity concentration degree. The daily average power maximum value refers to the average value of the maximum value of the target user daily power consumption, which is used to distinguish the large power equipment users and the small power equipment users. The target feature vector refers to a vector composed of the daily average power fluctuation coefficient, the peak period power ratio, and the daily average power maximum value, which is used to quantify the load characteristics of the target user and serves as the basis for matching degree calculation.

[0100] The first matching degree refers to the similarity between the target feature vector and the feature vector in the first data set, and the higher the value, the closer the target user is to the steady load type user characteristics. The second matching degree refers to the similarity between the target feature vector and the feature vector in the second data set, and the higher the value, the closer the target user is to the fluctuating load type user characteristics. The first reference user refers to a user whose load type is known to be a steady load type, and the feature vector thereof is used to construct the first data set. The second reference user refers to a user whose load type is known to be a fluctuating load type, and the historical single-phase voltage time series data thereof is used to determine the power consumption peak and valley periods, and the feature vector thereof is used to construct the second data set. The first voltage fluctuation frequency and the first voltage fluctuation amplitude refer to the fluctuation frequency and the fluctuation amplitude extracted from the single-phase voltage time series data in the power consumption peak period, respectively, for reflecting the voltage fluctuation characteristics of the user in the peak period. The second voltage fluctuation frequency and the second voltage fluctuation amplitude refer to the fluctuation frequency and the amplitude extracted from the single-phase voltage time series data in the power consumption valley period, respectively, for reflecting the voltage fluctuation characteristics of the user in the valley period. The power consumption peak period refers to a period in which the second reference user generally has high power consumption load and frequent voltage fluctuation, which is calculated based on the historical single-phase voltage time series data of a plurality of second reference users. The power consumption valley period refers to a period in which the second reference user generally has low power consumption load and gentle voltage fluctuation, which is calculated based on the historical single-phase voltage time series data of a plurality of second reference users.

[0101] In the present embodiment, the target of the user load type determination step is to improve the classification accuracy through multi-feature matching. The present embodiment selects three dimensional features, i.e., the daily power fluctuation coefficient, the peak period power ratio, and the daily power maximum value, to comprehensively depict the user load characteristics. Meanwhile, the present embodiment introduces a reference user data set of known types, and replaces the single threshold judgment with matching degree comparison, so as to reduce the subjective threshold setting deviation and ensure that the classification result is more consistent with the real load properties of the user.

[0102] The target of the voltage fluctuation feature extraction step is to cover the fluctuation characteristics of the user in the complete power consumption period. Considering that the voltage fluctuation of the user in the power consumption peak period and the valley period is significantly different, the present embodiment determines the common period based on the second reference user data, and then performs weighted summation on the fluctuation characteristics of the two periods, so as to avoid the influence of the individual user period deviation and comprehensively reflect the fluctuation law of the user under different load states, thereby providing more comprehensive feature basis for subsequent scene processing of the fluctuating load type user and improving the accuracy of phase recognition.

[0103] For example, the step of determining the user load type based on the historical power consumption data can include:

[0104] The intelligent fusion terminal can retrieve the historical power consumption data of the target user smart meter for the past 30 days to calculate the features. The daily average power fluctuation coefficient = (daily maximum power - daily minimum power) / daily average power; the peak period power ratio = total power during 18:00-22:00 (preset resident peak period) / total daily power; the daily average maximum power = 30-day average of daily maximum power. The intelligent fusion terminal can combine the calculated 30-day average daily power fluctuation coefficient, the average peak period power ratio, and the average daily maximum power into a target feature vector, such as [0.18, 0.35, 2.8].

[0105] The intelligent fusion terminal can retrieve the first data set and the second data set from the local database. Assume that the first data set contains the feature vectors of 50 reference users of the stable load type (such as [0.15, 0.32, 2.5], etc.); the second data set contains the feature vectors of 50 reference users of the fluctuating load type (such as [0.85, 0.7, 15.2], etc.). The intelligent fusion terminal can use the cosine similarity algorithm to calculate the average similarity of the target feature vector with all vectors in the first data set (i.e., the first matching degree, such as 0.92) and the average similarity with all vectors in the second data set (i.e., the second matching degree, such as 0.38). Since the first matching degree 0.92 is greater than the second matching degree 0.38, the intelligent fusion terminal can determine that the target user is of the stable load type and store the result in association with the user number.

[0106] For example, the step of extracting the voltage fluctuation frequency and the voltage fluctuation amplitude based on the single-phase voltage time series data can include:

[0107] The intelligent fusion terminal can retrieve the historical single-phase voltage time series data of a plurality of second reference users (such as 20 industrial users) for the past month from the local database, and calculate that the power consumption peak period is 8:00-18:00 (voltage fluctuation is frequent), and the power consumption valley period is 0:00-6:00 (voltage fluctuation is gentle). The intelligent fusion terminal can count the number of times of voltage change exceeding 0.8V in the peak period (8:00-18:00) (first voltage fluctuation frequency, such as 45 times), and the maximum voltage difference in the peak period (first voltage fluctuation amplitude, such as 2.3V) from the single-phase voltage time series data of the target user (fluctuating load type); count the number of times of voltage change exceeding 0.8V in the valley period (0:00-6:00) (second voltage fluctuation frequency, such as 5 times), and the maximum voltage difference in the valley period (second voltage fluctuation amplitude, such as 0.6V). The intelligent fusion terminal can preset the peak period weight 0.7 and the valley period weight 0.3, and calculate: voltage fluctuation frequency = 45x0.7 + 5x0.3 = 33 times / day; voltage fluctuation amplitude = 2.3x0.7 + 0.6x0.3 = 1.79V, and store the result to the fluctuation feature library.

[0108] In the step of determining the user load type, the embodiment constructs a multi-dimensional target feature vector by extracting the daily average power fluctuation coefficient, the peak period power proportion, and the daily average power maximum value, and determines the type by comparing the matching degree with the reference user data set of the stationary / fluctuation load type, so as to avoid misjudgment caused by a single feature or subjective threshold and improve the load classification accuracy, thereby providing a reliable basis for the subsequent phase recognition algorithm selection.

[0109] In the step of extracting the voltage fluctuation feature, the embodiment determines the common electricity peak and valley period based on the second reference user data, and performs weighted summation on the fluctuation features of the two periods, so as to cover the fluctuation law of the complete electricity period of the user, avoid the influence of individual period deviation, provide comprehensive feature support for the scene processing of the fluctuation load type user, finally improve the overall accuracy of the phase recognition, adapt to the complex user characteristics of the low-voltage distribution area, and meet the high-precision operation and maintenance demand.

[0110] In an embodiment of the present application, the first matching degree of the target feature vector and the first data set is calculated, including:

[0111] If the number of feature vectors in the first data set is greater than the first number, the first number of feature vectors are randomly selected from the first data set; the matching degrees of the target feature vector and each feature vector are calculated respectively, and the first matching degree is calculated based on the minimum matching degree, the maximum matching degree and the average matching degree in all matching degrees;

[0112] If the number of feature vectors in the first data set is less than or equal to the first number, the matching degrees of the target feature vector and each feature vector in the first data set are calculated respectively, and the first matching degree is calculated based on the minimum matching degree, the maximum matching degree and the average matching degree in all matching degrees.

[0113] In an embodiment of the present application, the first matching degree of the target feature vector and the first data set is calculated, including:

[0114] The average feature vector is determined based on the first data set, and the matching degree of the target feature vector and the average feature vector is calculated as the first matching degree.

[0115] In the embodiment, the first quantity refers to a preset threshold for determining whether the number of feature vectors in the first data set needs to be sampled, used to balance the calculation efficiency and the result accuracy, which can be set as 20 for example. The average feature vector refers to a new feature vector formed by taking the average of the data in each dimension of all feature vectors in the first data set, used to simplify the matching degree calculation and represent the overall feature level of the first data set. The maximum matching degree refers to the maximum matching degree in the matching degree results of the target feature vector and the plurality of feature vectors in the first data set, used to highlight the most similar reference user feature of the target user. The minimum matching degree refers to the minimum matching degree in the matching degree results of the target feature vector and the plurality of feature vectors in the first data set, used to determine the lowest similarity level of the target user and the stable load type user or the fluctuating load type user. The average matching degree refers to the arithmetic average of all matching degree results.

[0116] The target of the embodiment is to balance the calculation efficiency, the result accuracy and the reliability. When the number of feature vectors in the first data set exceeds the first quantity, the sampling calculation can avoid the full calculation from occupying too much terminal computing power, ensuring the processing efficiency; and when the number is small, the full calculation can ensure that no data is missed. The embodiment combines the minimum, maximum and average matching degree calculation, which can comprehensively reflect the overall similarity and the extreme similar situation of the target user and the stable load type user, avoiding the deviation caused by a single mean or extreme value; the embodiment is based on the average feature vector calculation, which can simplify the process and quickly obtain the matching result, and the two methods can adapt to different computing power and accuracy requirements, providing a more comprehensive and flexible basis for load classification.

[0117] For example, the intelligent fusion terminal retrieves the first data set, and the number of feature vectors is 30, and the first quantity is preset as 20. Since 30>20, 20 feature vectors are randomly selected. The intelligent fusion terminal can use the cosine similarity algorithm to calculate the matching degree of the target vector (such as [0.18, 0.35, 2.8]) and the 20 feature vectors, and obtain the results (such as minimum 0.82, maximum 0.95, and average 0.90). The intelligent fusion terminal can use the preset weight (minimum 0.2, maximum 0.3, and average 0.5) to weight and sum, i.e. 0.82*0.2+0.95*0.3+0.90*0.5=0.89, which is the first matching degree; if the number of vectors is 15≤20, the full calculation is directly performed and then weighted.

[0118] For example, the intelligent fusion terminal can take the average of each dimension of all vectors in the first data set (such as daily average power fluctuation coefficient 0.16, peak proportion 0.33, and maximum power 2.6) to form the average vector [0.16, 0.33, 2.6]. The intelligent fusion terminal can use the Pearson correlation coefficient algorithm to calculate the matching degree of the target vector and the average vector (such as 0.88), which is the first matching degree.

[0119] The sampling calculation of the embodiment can reduce the computing power consumption and improve the processing efficiency; the embodiment combines multiple matching degree index calculations to avoid single index deviation and improve the result reliability; the embodiment is based on average feature vector calculation to simplify the process and quickly output the result. The two methods adapt to different scenarios to ensure that the load classification is efficient and accurate, provide reliable support for subsequent phase recognition algorithm selection, and further improve the overall performance of the phase recognition of the transformer area.

[0120] In an embodiment of the present application, the correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data is calculated based on the first attention weight coefficient, comprising:

[0121] The single-phase voltage time series data is divided into stationary single-phase voltage data and non-stationary single-phase voltage data;

[0122] The stationary single-phase voltage data and the non-stationary single-phase voltage data are respectively weighted based on the first attention weight coefficient to obtain the weighted single-phase voltage time series data;

[0123] The correlation coefficient of the weighted single-phase voltage time series data and the three-phase voltage time series data is calculated.

[0124] In the embodiment, the first attention weight coefficient further includes a first non-stationary period weight, and the first stationary period weight is greater than the first non-stationary period weight; the stationary single-phase voltage data and the non-stationary single-phase voltage data are respectively weighted based on the first attention weight coefficient to obtain the weighted single-phase voltage time series data, specifically comprising:

[0125] The stationary single-phase voltage data is weighted based on the first stationary period weight;

[0126] The non-stationary single-phase voltage data is weighted based on the first non-stationary period weight;

[0127] The weighted single-phase voltage time series data is obtained based on the weighted stationary single-phase voltage data and the weighted non-stationary single-phase voltage data.

[0128] In the embodiment, the correlation coefficient includes a first correlation coefficient, a second correlation coefficient and a third correlation coefficient; the correlation coefficient of the weighted single-phase voltage time series data and the three-phase voltage time series data is calculated, specifically comprising: calculating the first correlation coefficient of the weighted single-phase voltage time series data and the A-phase voltage time series data; calculating the second correlation coefficient of the weighted single-phase voltage time series data and the B-phase voltage time series data; calculating the third correlation coefficient of the weighted single-phase voltage time series data and the C-phase voltage time series data.

[0129] In the embodiment, the first correlation coefficient of the weighted single-phase voltage time series data and the A-phase voltage time series data is calculated, specifically comprising:

[0130] For any one of the weighted single-phase voltage time series data and the A-phase voltage time series data, a third operation is performed to obtain a multi-dimensional feature vector; a correlation coefficient of the multi-dimensional feature vector corresponding to the weighted single-phase voltage time series data and the multi-dimensional feature vector corresponding to the A-phase voltage time series data is calculated, and the correlation coefficient is taken as a correlation coefficient of the weighted single-phase voltage time series data and the A-phase voltage time series data;

[0131] The third operation includes:

[0132] The basic voltage features include a voltage amplitude, a voltage mean value, and a voltage standard deviation, and the derived features include a voltage fluctuation rate, a voltage mutation slope, and a period voltage stability index.

[0133] The wavelet packet decomposition algorithm is used to perform multi-scale time-frequency feature extraction on the voltage time series data to obtain a low-frequency component and a plurality of high-frequency components; energy proportions of the low-frequency component and each high-frequency component are calculated respectively, and time-frequency entropies of the low-frequency component and each high-frequency component are calculated respectively; the energy proportions of the low-frequency component and each high-frequency component and the time-frequency entropies of the low-frequency component and each high-frequency component are taken as multi-scale time-frequency features.

[0134] The multi-dimensional feature vector is constructed based on the basic voltage features, the derived features, and the multi-scale time-frequency features.

[0135] In the embodiment, the smooth single-phase voltage data refers to voltage data corresponding to a period of small voltage fluctuation and gentle change in the single-phase voltage time series data, which is used to highlight the role of low-interference data, for example, the smooth period is preset as 0:00-5:00. The non-smooth single-phase voltage data refers to voltage data corresponding to a period of relatively large voltage fluctuation in the single-phase voltage time series data, which has relatively strong interference, for example, the non-smooth period is preset as voltage data of 5:00-24:00. The first non-smooth period weight refers to a weight set for the non-smooth single-phase voltage data in the first attention weight coefficient, which is smaller than the first smooth period weight and is used to weaken the interference of the non-smooth data. The weighted single-phase voltage time series data refers to complete voltage time series data formed by combining the smooth single-phase voltage data and the non-smooth single-phase voltage data after giving corresponding weights, which can better reflect the effective voltage change characteristics. The smooth correlation coefficient is used to reflect the voltage correlation in the low-interference period. The non-smooth correlation coefficient is used to reflect the voltage correlation in the interference period.

[0136] The low-frequency component refers to a signal component with a lower frequency and reflecting a long-term stable change trend of the voltage after wavelet packet decomposition of the voltage time series data, and is used to capture the basic change rule of the voltage. The high-frequency component refers to a signal component with a higher frequency and reflecting a short-term fluctuation detail of the voltage after wavelet packet decomposition of the voltage time series data, and is used to capture the instantaneous change characteristics of the voltage. The energy proportion refers to a proportion of energy of the low-frequency component or each high-frequency component in total energy of all decomposed components, and is used to quantify the contribution degree of different frequency components to the voltage time series data. The time-frequency entropy refers to an entropy value calculated based on the low-frequency component or each high-frequency component, and is used to represent the complexity and uncertainty of the voltage signal in the corresponding frequency band. The lower the entropy value is, the more stable the signal in the frequency band is. The multi-dimensional feature vector refers to a high-dimensional vector formed by fusing the basic voltage features, the derived features and the multi-scale time-frequency features, and is used to comprehensively quantify the characteristics of the voltage time series data, and lay a foundation for accurately calculating the correlation coefficient.

[0137] The target of the embodiment is to improve the accuracy of phase identification of the stable load type user. Considering that the stable load type user has small overall fluctuations, but there are still small disturbances in the non-stationary period, if the two types of data are equally weighted, the real phase correlation signal is easy to be diluted. Therefore, the embodiment distinguishes between stable and non-stationary data, and gives higher weight to stable data, so as to strengthen the influence of low-interference data. At the same time, when calculating the correlation coefficient, the A-phase voltage data is also split according to the corresponding period to ensure that the periods are aligned and avoid correlation deviation caused by time misalignment. The final coefficient is determined by combining the stable and non-stationary correlation coefficients, which not only retains effective information but also weakens interference, so as to ensure that the correlation coefficient can truly reflect the voltage correlation between the user and the corresponding phase of the total meter, and lay a foundation for accurately determining the phase.

[0138] The embodiment improves the accuracy of correlation coefficient calculation by multi-dimensional feature extraction. If the correlation is directly calculated based on the voltage time series data, it is easy to be affected by random noise and instantaneous interference, and cannot fully reflect the essential correlation of the voltage. Therefore, the embodiment extracts basic features (reflecting the basic attributes of the voltage), derived features (reflecting the dynamic changes of the voltage) and multi-scale time-frequency features (reflecting the characteristics of the voltage in different frequency bands), which can describe the characteristics of the voltage from three dimensions of static attributes-dynamic changes-time-frequency distribution, and avoid the limitations of single feature. The wavelet packet decomposition can capture time-frequency information in multiple scales, the energy proportion and the time-frequency entropy can quantify the contribution and stability of each frequency band, and the finally constructed multi-dimensional feature vector can more truly reflect the voltage correlation and ensure the accuracy of the first correlation coefficient, providing a reliable basis for phase identification of the stable load type user.

[0139] For example, the embodiment can calculate the correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on the first attention weight coefficient according to the following detailed steps:

[0140] The first step is to divide the stationary and non-stationary single-phase voltage data. Specifically, the intelligent fusion terminal can call the single-phase voltage time series data of the target user in the past 24 hours, and the data points are denoted as , , . The terminal presets the stationary period as 0:00-5:00 (assuming the corresponding data points - ), and the determination criterion is that the proportion of adjacent sampling points with voltage difference - ≤0.5V is ≥90% in this period. If it is satisfied, it is divided into stationary single-phase voltage data; the remaining 5:00-24:00 (assuming the corresponding data points - ) is divided into non-stationary single-phase voltage data.

[0141] The second step is to weight based on the first attention weight coefficient. The terminal calls the first attention weight coefficient: assuming that the first stationary period weight , the first non-stationary period weight . For each point (i=1-300) in the stationary single-phase voltage data, the weighting value is calculated according to '= × ; for each point (j=301-1440) in the non-stationary single-phase voltage data, the weighting value is calculated according to '= × ; merge ', ', ', ' in timestamp order, and get the weighted single-phase voltage time series data.

[0142] The third step is to perform the third operation to construct a multi-dimensional feature vector (taking the A-phase voltage time series data as an example):

[0143] Extract features: in the basic voltage feature, the voltage amplitude is the instantaneous value of each sampling point, the voltage mean , and the voltage standard deviation ; in the derived feature, the voltage fluctuation rate , where h is the hour sequence number, the voltage mutation slope , where N is the number of points with - ≥0.6V, and the period voltage stability index S= / 1440.

[0144] Wavelet packet decomposition: 3-level decomposition using db4 wavelet basis, resulting in 8 components, such as 1 low-frequency L, 7 high-frequency , component energy , energy proportion , time-frequency entropy .

[0145] Construct a vector: concatenate multi-dimensional basic derived features and multi-dimensional time-frequency features to form a multi-dimensional feature vector (denoted as V_user, V_A, corresponding to the weighted single-phase voltage and A-phase voltage data, respectively).

[0146] Fourth step, calculate the first correlation coefficient. Use the improved Pearson algorithm: first, do Z-score standardization on V_user and V_A; calculate the covariance , standard deviation product ; where is the value of the i-th feature of V_user after Z-score standardization, is the value of the i-th feature of V_A after Z-score standardization, is the mean of all feature values of V_user after standardization, is the mean of all feature values of V_A after standardization, is the standard deviation of all feature values of V_user after standardization, is the standard deviation of all feature values of V_A after standardization, n is the dimension of the multi-dimensional feature vector (i.e. the number of features), is the covariance of the feature vectors of V_user and V_A after standardization.

[0147] Introduce weights = 0.5 (time-frequency features), = 0.3 (basic features), = 0.2 (derived features), the first correlation coefficient , where , R_b, R_d are the correlation coefficients of time-frequency features, basic features, and derived features, respectively. The example result = 0.92. Similarly, calculate the second correlation coefficient (B phase), and the third correlation coefficient (C phase).

[0148] This embodiment distinguishes between stationary and non-stationary single-phase voltage data and differentiates the weighting, giving greater weight to low-interference stationary data and weakening small disturbances during non-stationary periods, avoiding dilution of the true phase correlation signal, and solving the recognition bias problem caused by stationary load class user disturbances during non-stationary periods.

[0149] Meanwhile, the embodiment extracts the basic, derived and multi-scale time-frequency features through the third operation, describes the voltage characteristics from the three dimensions of static attributes, dynamic changes and time-frequency distribution, captures the signal characteristics of different frequency bands through wavelet packet decomposition, quantifies the contribution and stability of each frequency band through energy proportion and time-frequency entropy, and then calculates the correlation coefficient through the improved Pearson algorithm and feature weight fusion, thereby significantly improving the coefficient accuracy, providing a reliable basis for phase recognition of smooth load type users, and effectively improving the overall accuracy of phase recognition.

[0150] In an embodiment of the present application, the second attention weight coefficient further includes a second non-stationary period weight, the second non-stationary period weight being less than the second stationary period weight; and the correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data is calculated based on the second attention weight coefficient, including:

[0151] The single-phase voltage time series data is divided into fluctuation single-phase voltage data and non-fluctuation single-phase voltage data;

[0152] The fluctuation single-phase voltage data and the non-fluctuation single-phase voltage data are respectively weighted based on the second attention weight coefficient to obtain weighted single-phase voltage time series data;

[0153] For any voltage time series data in the weighted single-phase voltage time series data and the three-phase voltage time series data, a second operation is performed to obtain a multi-dimensional feature vector; and the correlation coefficient of the weighted single-phase voltage time series data and the three-phase voltage time series data is calculated based on the multi-dimensional feature vectors corresponding to the weighted single-phase voltage time series data and the three-phase voltage time series data respectively;

[0154] The second operation includes:

[0155] The basic voltage features and the derived features are extracted from the voltage time series data; the basic voltage features include voltage amplitude, voltage mean value, voltage standard deviation and voltage mutation number, and the derived features include mutation duration proportion and mutation amplitude concentration;

[0156] The frequency domain signal is obtained by performing fast Fourier transform on the voltage time series data, and the low-order frequency domain features are extracted from the frequency domain signal; the low-order frequency domain features include fundamental phase angle and harmonic proportion;

[0157] The frequency domain signal is filtered through a narrowband filter to obtain filtered fundamental signal data, and the fluctuation variance of the fundamental amplitude is calculated based on the fundamental signal data as a fundamental stability feature;

[0158] The multi-dimensional feature vector is constructed based on the low-order frequency domain features, the fundamental stability feature, the basic voltage features and the derived features.

[0159] In the embodiment, the second non-stationary period weight is a weight set in the second attention weight coefficient for fluctuation single-phase voltage data other than non-fluctuation single-phase voltage data, and the value is less than the second stationary period weight, for weakening the interference of fluctuation data. The fluctuation single-phase voltage data refers to voltage data of a period with obvious fluctuation in the single-phase voltage time sequence, and the interference is strong; the non-fluctuation single-phase voltage data refers to voltage data of a period with gentle fluctuation, and the interference is weak, which is the key data of phase correlation.

[0160] The voltage mutation frequency refers to the number of times that the voltage change amount in the voltage time sequence data is greater than or equal to a preset threshold (such as 0.3V) per unit time, for quantifying the micro-mutation frequency of the fluctuation type user. The mutation duration proportion refers to the ratio of the total duration of voltage mutation events to the total duration of the period, for representing the time distribution characteristics of the micro-mutation of the fluctuation type user. The mutation amplitude concentration degree refers to the reciprocal of the variance (1 / Var) of the voltage mutation amplitude, and the smaller the variance, the higher the concentration degree, for quantifying the aggregation degree of the mutation amplitude. The low-order frequency domain feature refers to a feature extracted from the frequency domain signal, reflecting the influence of the load mutation of the fluctuation type user, focusing on the information of the low frequency band (fundamental wave and low-order harmonic wave). The fundamental wave phase angle refers to the phase angle of the 50Hz fundamental wave component in the frequency domain signal, for reflecting the phase correlation characteristics of the voltage signal. The harmonic proportion refers to the proportion of the amplitude of the 2-3th harmonic wave to the amplitude of the fundamental wave, for quantifying the influence of the low-order harmonic wave on the voltage signal. The narrowband filter refers to a filter with a center frequency of 50Hz and a bandwidth of 2Hz, for filtering high-frequency noise in the frequency domain signal and retaining the fundamental wave signal. The fundamental wave stability feature refers to the fundamental wave amplitude fluctuation variance calculated based on the filtered fundamental wave signal data, for representing the stability degree of the fundamental wave signal.

[0161] The target of the embodiment is to adapt to the characteristics of the low fluctuation user in the fluctuation load type. Such user still has fluctuation data interference, by dividing the fluctuation and non-fluctuation data, and using the second attention weight coefficient to give higher weight to the non-fluctuation data, the effective signal can be strengthened and the interference can be weakened, so as to ensure that the correlation coefficient can truly reflect the phase correlation and improve the calculation accuracy. At the same time, the load mutation of the fluctuation type user mainly affects the low-order harmonic wave and the stability of the fundamental wave, if the stationary type feature extraction mode is followed, the key fluctuation information is easy to be missed. Therefore, the embodiment extracts the basic features (including voltage mutation frequency, quantifying micro-mutation frequency) and derived features (focusing on mutation duration and amplitude distribution) through the second operation, which can capture the time domain fluctuation details; the low-order frequency domain feature is extracted by using the fast Fourier transform, which is consistent with the influence law of the load mutation on the low-order harmonic wave; the narrowband filter purifies the fundamental wave signal, and the fundamental wave stability feature supplements the fluctuation information of the fundamental wave, finally the multi-dimensional features are fused to construct the vector, avoiding the limitation of single feature, and ensuring that the voltage correlation of the fluctuation type user can be truly reflected, providing a reliable basis for subsequent correlation calculation.

[0162] Exemplarily, the embodiment can be implemented according to the following detailed steps to calculate the correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on the second attention weight coefficient:

[0163] Firstly, the fluctuation single-phase voltage data and the non-fluctuation single-phase voltage data are divided. The intelligent fusion terminal calls the single-phase voltage time series data of the past 24 hours of the fluctuation load target user. The terminal presets the fluctuation judgment threshold as 0.4V, and the proportion of the sampling points meeting the threshold in each hour is counted: if the proportion of a certain hour is greater than or equal to 15%, the data corresponding to the hour is divided into fluctuation single-phase voltage data; otherwise, it is divided into non-fluctuation single-phase voltage data. It is assumed that the division result is that the fluctuation data corresponds to the time period 8:00-18:00, and the non-fluctuation data corresponds to the time period 22:00-6:00 of the next day and 6:00-8:00, 18:00-22:00.

[0164] Secondly, the second attention weight coefficient is weighted. The terminal calls the second attention weight coefficient parameters: the second stationary period weight (non-fluctuation data weight) =2.0, and the second non-stationary period weight (fluctuation data weight) =0.9. According to , the weighting value of each point in the non-fluctuation single-phase voltage data is calculated; according to , the weighting value of each point in the fluctuation single-phase voltage data is calculated; and all the weighted data points are merged in the time stamp order to obtain the weighted single-phase voltage time series data.

[0165] Thirdly, the second operation is performed to construct a multi-dimensional feature vector (taking the B-phase voltage time series data U_B in the three-phase voltage time series data as an example):

[0166] The basic voltage features and derived features are extracted: in the basic voltage features, the voltage amplitude is the instantaneous value of 1440 sampling points of U_B, the voltage mean μ_B, the voltage standard deviation σ_B, and the voltage mutation number , wherein I is an indicator function, which is 1 if the condition is met, and 0 otherwise; in the derived features, the mutation duration ratio R_B and the mutation amplitude concentration C_B.

[0167] The low-order frequency domain features are extracted: the fast Fourier transform is performed on U_B to obtain the frequency domain signal F_B; the phase angle of the 50Hz fundamental wave component is extracted , and the 2nd harmonic ratio H_B2 and the 3rd harmonic ratio H_B3 are calculated.

[0168] The fundamental wave stability features are calculated: a narrowband filter with a center frequency of 50Hz and a bandwidth of 2Hz is used to filter F_B to obtain the filtered fundamental wave signal F_Bfilter; the fundamental wave amplitude sequence of F_Bfilter is extracted, and the fluctuation variance of the fundamental wave amplitude is calculated , total 1-dimensional features.

[0169] Constructing multi-dimensional feature vector: concatenating the basic derived features, low-order frequency domain features and fundamental stability features in order to form a multi-dimensional feature vector V_B of multi-dimensional B-phase voltage time series data; similarly, performing the above operations on the weighted single-phase voltage time series data to construct a multi-dimensional vector V_user.

[0170] Fourth step, calculating the correlation coefficient. Z-score standardization is performed on V_user and V_B; the improved Pearson correlation coefficient algorithm is adopted, and feature weights (low-order frequency domain features + fundamental stability features), (basic voltage features), (derived features) are introduced , and the final correlation coefficient is calculated:

[0171] The second attention weight coefficient is used to differentiate and weight the fluctuation and non-fluctuation single-phase voltage data in this embodiment, so as to strengthen the influence of low-interference non-fluctuation data with a higher second stationary period weight, and weaken the fluctuation data interference with a lower second non-stationary period weight, thereby avoiding dilution of the effective phase correlation signal and accurately adapting to the characteristics of fluctuation load type low fluctuation users. Meanwhile, the second operation extracts features from multiple dimensions of time domain (basic features contain voltage mutation times, derived features focus on mutation distribution), frequency domain (low-order frequency domain features match the influence law of load mutation), and fundamental stability (calculate fluctuation variance after narrowband filtering), constructs a multi-dimensional feature vector that comprehensively describes the voltage characteristics, calculates the correlation coefficient by combining the improved Pearson algorithm and feature weight fusion, significantly improves the coefficient accuracy, provides a reliable basis for phase recognition of fluctuation load type low fluctuation users, and effectively solves the poor adaptability and low recognition accuracy of the prior art for this type of users.

[0172] Corresponding to the phase recognition method based on the intelligent fusion terminal of the above embodiments, Figure 2 is a structural block diagram of a phase recognition device based on an intelligent fusion terminal provided by an embodiment of the present application. Only parts related to the embodiments of the present application are shown for ease of illustration. Reference Figure 2 The phase recognition device based on the intelligent fusion terminal 20 includes a data acquisition module 21, a user division module 22, a first phase recognition module 23, a second phase recognition module 24, and a phase analysis module 25.

[0173] The data acquisition module 21 is configured to acquire single-phase voltage time series data from a smart meter of a target user and acquire three-phase voltage time series data from a total meter of an outlet transformer of a power distribution area.

[0174] The user classification module 22 is configured to determine a user load type based on historical power consumption data of the target user.

[0175] The first phase recognition module 23 is configured to, if the user load type is a steady load type, calculate a correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on a first attention weight coefficient. The first attention weight coefficient includes a first steady period weight. The first steady period weight is a weight of steady period data in the single-phase voltage time series data.

[0176] The second phase recognition module 24 is configured to, if the user load type is a fluctuation load type, perform a first operation to obtain the correlation coefficient. The first operation includes:

[0177] extracting a voltage fluctuation frequency and a voltage fluctuation amplitude based on the single-phase voltage time series data;

[0178] if the voltage fluctuation frequency exceeds a first fluctuation frequency threshold and / or the voltage fluctuation amplitude exceeds a first fluctuation amplitude threshold, determining a non-fluctuation period based on the single-phase voltage time series data, extracting target single-phase voltage data corresponding to the non-fluctuation period from the single-phase voltage time series data, extracting target three-phase voltage data corresponding to the non-fluctuation period from the three-phase voltage time series data, and calculating a correlation coefficient of the target single-phase voltage data and the target three-phase voltage data;

[0179] if the voltage fluctuation frequency does not exceed the first fluctuation frequency threshold and the voltage fluctuation amplitude does not exceed the first fluctuation amplitude threshold, calculating the correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on a second attention weight coefficient. The second attention weight coefficient includes a second steady period weight. The second steady period weight is greater than the first steady period weight.

[0180] The phase analysis module 25 is configured to determine a target phase corresponding to the target user based on the correlation coefficient.

[0181] In an embodiment of the present application, the user classification module 22, when determining the user load type based on the historical power consumption data of the target user, is specifically configured to:

[0182] extract a daily average power fluctuation coefficient, a peak period power proportion, and a daily average power maximum value from the historical power consumption data of the target user;

[0183] construct a target feature vector based on the daily average power fluctuation coefficient, the peak period power proportion, and the daily average power maximum value;

[0184] calculate a first matching degree of the target feature vector and a first data set; the first data set includes a plurality of feature vectors corresponding to a plurality of first reference users respectively, and the user load types of the plurality of first reference users are all stable load types;

[0185] calculate a second matching degree of the target feature vector and a second data set; the second data set includes a plurality of feature vectors corresponding to a plurality of second reference users respectively, and the user load types of the plurality of second reference users are all fluctuant load types;

[0186] if the first matching degree is greater than or equal to the second matching degree, determine that the user load type is a stable load type;

[0187] if the first matching degree is less than the second matching degree, determine that the user load type is a fluctuant load type.

[0188] In an embodiment of the present application, the user division module 22, when extracting the voltage fluctuation frequency and the voltage fluctuation amplitude based on the single-phase voltage time series data, is specifically configured to:

[0189] extract a first voltage fluctuation frequency and a first voltage fluctuation amplitude of a power consumption peak period based on the single-phase voltage time series data;

[0190] extract a second voltage fluctuation frequency and a second voltage fluctuation amplitude of a power consumption valley period based on the single-phase voltage time series data; the power consumption peak period and the power consumption valley period are both calculated based on historical single-phase voltage time series data of a plurality of second reference users;

[0191] perform weighted summation on the first voltage fluctuation frequency and the second voltage fluctuation frequency to obtain the voltage fluctuation frequency;

[0192] perform weighted summation on the first voltage fluctuation amplitude and the second voltage fluctuation amplitude to obtain the voltage fluctuation amplitude.

[0193] In an embodiment of the present application, the first phase recognition module 23, when calculating the correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on the first attention weight coefficient, is specifically configured to: divide the single-phase voltage time series data into stable single-phase voltage data and non-stable single-phase voltage data; weight the stable single-phase voltage data and the non-stable single-phase voltage data based on the first attention weight coefficient respectively to obtain weighted single-phase voltage time series data; and calculate the correlation coefficient of the weighted single-phase voltage time series data and the three-phase voltage time series data.

[0194] In an embodiment of the present application, the first attention weight coefficient further comprises a first non-stationary period weight, and the first stationary period weight is greater than the first non-stationary period weight; when the first phase recognition module 23 weights the stationary single-phase voltage data and the non-stationary single-phase voltage data based on the first attention weight coefficient respectively to obtain the weighted single-phase voltage time series data, the first phase recognition module 23 is specifically configured to: weight the stationary single-phase voltage data based on the first stationary period weight; weight the non-stationary single-phase voltage data based on the first non-stationary period weight; and obtain the weighted single-phase voltage time series data based on the weighted stationary single-phase voltage data and the weighted non-stationary single-phase voltage data.

[0195] In an embodiment of the present application, when the second phase recognition module 24 determines the non-fluctuation period based on the single-phase voltage time series data, the second phase recognition module 24 is specifically configured to:

[0196] divide the single-phase voltage time series data into a plurality of single-phase voltage data segments based on the first step length;

[0197] for each single-phase voltage data segment, calculate a voltage segment fluctuation frequency and a voltage segment fluctuation amplitude of the single-phase voltage data segment, and if the voltage segment fluctuation frequency exceeds the second fluctuation frequency threshold and / or the voltage segment fluctuation amplitude exceeds the second fluctuation amplitude threshold, determine that the period corresponding to the single-phase voltage data segment is a non-fluctuation period;

[0198] the second fluctuation frequency threshold is greater than the first fluctuation frequency threshold, and the second fluctuation amplitude threshold is greater than the first fluctuation amplitude threshold.

[0199] In an embodiment of the present application, the second attention weight coefficient further comprises a second non-stationary period weight, and the second non-stationary period weight is less than the second stationary period weight; when the second phase recognition module 24 calculates the correlation coefficient of the single-phase voltage time series data and the three-phase voltage time series data based on the second attention weight coefficient, the second phase recognition module 24 is specifically configured to:

[0200] divide the single-phase voltage time series data into fluctuation single-phase voltage data and non-fluctuation single-phase voltage data;

[0201] weight the fluctuation single-phase voltage data and the non-fluctuation single-phase voltage data based on the second attention weight coefficient respectively to obtain the weighted single-phase voltage time series data;

[0202] for any voltage time series data in the weighted single-phase voltage time series data and the three-phase voltage time series data, perform a second operation to obtain a multi-dimensional feature vector; and calculate the correlation coefficient of the weighted single-phase voltage time series data and the three-phase voltage time series data based on the multi-dimensional feature vectors corresponding to the weighted single-phase voltage time series data and the three-phase voltage time series data respectively;

[0203] wherein the second operation comprises:

[0204] Basic voltage features and derived features are extracted from voltage time series data. Basic voltage features include voltage amplitude, voltage mean, voltage standard deviation, and number of voltage abrupt changes. Derived features include the percentage of duration of abrupt changes and the concentration of abrupt change amplitude.

[0205] A fast Fourier transform is performed on the voltage time series data to obtain the frequency domain signal, and low-order frequency domain features are extracted from the frequency domain signal; the low-order frequency domain features include the fundamental phase angle and the proportion of harmonics.

[0206] The frequency domain signal is filtered by a narrowband filter to obtain the filtered fundamental wave signal data. The fluctuation variance of the fundamental wave amplitude is calculated based on the fundamental wave signal data as the fundamental wave stability feature.

[0207] A multidimensional feature vector is constructed based on low-order frequency domain features, fundamental frequency stability features, fundamental voltage features, and derived features.

[0208] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the modules in the aforementioned device embodiments, for example... Figure 2 The functions of the data acquisition module 21, user segmentation module 22, first phase recognition module 23, second phase recognition module 24, and phase analysis module 25 are shown.

[0209] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), but it may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0210] The input device 302 can include a touchpad, a fingerprint collection sensor (for collecting fingerprint information and direction information of a fingerprint of a user), a microphone, etc., and the output device 303 can include a display (LCD, etc.), a speaker, etc.

[0211] The memory 304 can include a read-only memory and a random access memory, and provide instructions and data to the processor 301. A portion of the memory 304 can also include a non-volatile random access memory. For example, the memory 304 can also store information of voltage data.

[0212] In specific implementations, the processor 301, the input device 302, and the output device 303 described in the embodiments of the present application can perform the implementation manners described in the embodiments of the phase recognition method based on the intelligent fusion terminal provided by the embodiments of the present application, and can also perform the implementation manners of the electronic device 300 described in the embodiments of the present application, which will not be described here.

[0213] In another embodiment of the present application, a computer readable storage medium is provided, and the computer readable storage medium stores a computer program. The computer program includes program instructions, and the program instructions are executed by a processor to implement all or part of the processes in the above-mentioned embodiment methods. The computer program can also be used to instruct related hardware to complete the computer program. The computer program can be stored in a computer readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer readable medium can include any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0214] The computer readable storage medium can be an internal storage unit of the electronic device of any of the preceding embodiments, such as a hard disk or a memory of the electronic device. The computer readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, or the like equipped on the electronic device. Further, the computer readable storage medium can include both the internal storage unit and the external storage device of the electronic device. The computer readable storage medium is used to store a computer program and other programs and data required by the electronic device. The computer readable storage medium can also be used to temporarily store data that has been output or will be output.

[0215] Those skilled in the art can understand that the modules / units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been described in general terms in the foregoing description. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0216] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the electronic device and the units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0217] In several embodiments provided in the present application, it should be understood that the disclosed electronic device and method can be implemented in other ways. For example, the apparatus embodiments described above are only schematic, for example, the division of the modules / units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of modules, units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the shown or discussed each other can be indirect coupling or communication connection through some interfaces or modules / units, and can also be electrical, mechanical or other form of connection.

[0218] The modules / units described as separate components may or may not be physically separate, and the components displayed as modules / units may or may not be physical modules / units, i.e., may be located in one place, or may be distributed to multiple network modules / units. Part or all of the modules / units may be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0219] In addition, each functional module / unit in each embodiment of the present application can be integrated in one processing module / unit, or each module / unit can exist physically alone, or two or more modules / units can be integrated in one module / unit. The integrated module / unit can be realized in the form of hardware or in the form of a software functional module / unit.

[0220] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A phase recognition method based on an intelligent fusion terminal, characterized in that, include: Single-phase voltage timing data is obtained from the smart meters of the target users, and three-phase voltage timing data is obtained from the output meter of the transformer substation. Determine the user load type based on the target user's historical power consumption data; If the user load type is a stable load, then the correlation coefficient between the single-phase voltage time series data and the three-phase voltage time series data is calculated based on the first attention weight coefficient. The first attention weight coefficient includes the weight of the first stable period; The weight of the first stable period is the weight of the stable period data in the single-phase voltage time series data; If the user load type is a fluctuating load type, then perform the first operation to obtain the correlation coefficient; The target phase corresponding to the target user is determined based on the correlation coefficient. The first operation includes: Based on the single-phase voltage time-series data, the voltage fluctuation frequency and voltage fluctuation amplitude are extracted; If the voltage fluctuation frequency exceeds a first fluctuation frequency threshold and / or the voltage fluctuation amplitude exceeds a first fluctuation amplitude threshold, then a non-fluctuation period is determined based on the single-phase voltage time series data; target single-phase voltage data corresponding to the non-fluctuation period is extracted from the single-phase voltage time series data, and target three-phase voltage data corresponding to the non-fluctuation period is extracted from the three-phase voltage time series data; the correlation coefficient between the target single-phase voltage data and the target three-phase voltage data is calculated. If the voltage fluctuation frequency does not exceed the first fluctuation frequency threshold and the voltage fluctuation amplitude does not exceed the first fluctuation amplitude threshold, then the correlation coefficient between the single-phase voltage time series data and the three-phase voltage time series data is calculated based on the second attention weight coefficient; the second attention weight coefficient includes the weight of the second stable period; the weight of the second stable period is greater than the weight of the first stable period.

2. The phase recognition method based on an intelligent fusion terminal as described in claim 1, characterized in that, The process of determining the user load type based on the target user's historical power consumption data includes: Extract the daily average power fluctuation coefficient, peak hour power ratio, and daily average maximum power from the target user's historical power consumption data; A target feature vector is constructed based on the daily average power fluctuation coefficient, the peak period power ratio, and the daily average power maximum value. Calculate the first matching degree between the target feature vector and the first dataset; the first dataset includes feature vectors corresponding to multiple first reference users, and the user load type of the multiple first reference users is the stable load class; Calculate the second matching degree between the target feature vector and the second dataset; the second dataset includes feature vectors corresponding to multiple second reference users, and the user load type of the multiple second reference users is the fluctuating load class; If the first matching degree is greater than or equal to the second matching degree, then the user load type is determined to be a stable load type; If the first matching degree is less than the second matching degree, then the user load type is determined to be a fluctuating load type.

3. The phase recognition method based on an intelligent fusion terminal as described in claim 2, characterized in that, The extraction of voltage fluctuation frequency and voltage fluctuation amplitude based on the single-phase voltage time series data includes: Based on the single-phase voltage time-series data, the first voltage fluctuation frequency and the first voltage fluctuation amplitude during peak electricity consumption periods are extracted; The second voltage fluctuation frequency and the second voltage fluctuation amplitude during the off-peak electricity consumption period are extracted based on the single-phase voltage time series data; both the peak electricity consumption period and the off-peak electricity consumption period are calculated based on the historical single-phase voltage time series data of multiple second reference users; The voltage fluctuation frequency is obtained by weighted summation of the first voltage fluctuation frequency and the second voltage fluctuation frequency. The voltage fluctuation amplitude is obtained by weighted summation of the first voltage fluctuation amplitude and the second voltage fluctuation amplitude.

4. The phase recognition method based on an intelligent fusion terminal as described in claim 1, characterized in that, The calculation of the correlation coefficient between the single-phase voltage time series data and the three-phase voltage time series data based on the first attention weighting coefficient includes: The single-phase voltage time series data is divided into stationary single-phase voltage data and non-stationary single-phase voltage data; Based on the first attention weighting coefficient, the stationary single-phase voltage data and the non-stationary single-phase voltage data are weighted respectively to obtain the weighted single-phase voltage time series data. Calculate the correlation coefficient between the weighted single-phase voltage time series data and the three-phase voltage time series data.

5. The phase recognition method based on an intelligent fusion terminal as described in claim 4, characterized in that, The first attention weight coefficient also includes a first non-stationary period weight, wherein the first stationary period weight is greater than the first non-stationary period weight; The step of assigning weights to the stationary single-phase voltage data and the non-stationary single-phase voltage data based on the first attention weight coefficient to obtain weighted single-phase voltage time-series data includes: The stable single-phase voltage data are weighted based on the weight of the first stable time period; The non-stationary single-phase voltage data is weighted based on the weight of the first non-stationary time period. The weighted single-phase voltage time series data is obtained based on the weighted stationary single-phase voltage data and the weighted non-stationary single-phase voltage data.

6. The phase recognition method based on an intelligent fusion terminal as described in claim 1, characterized in that, The determination of non-fluctuation periods based on the single-phase voltage time series data includes: Based on the first step, the single-phase voltage timing data is divided into multiple single-phase voltage data segments; For each single-phase voltage data segment, calculate the voltage segment fluctuation frequency and voltage segment fluctuation amplitude of the single-phase voltage data segment. If the voltage segment fluctuation frequency exceeds the second fluctuation frequency threshold and / or the voltage segment fluctuation amplitude exceeds the second fluctuation amplitude threshold, then determine that the time period corresponding to the single-phase voltage data segment is a non-fluctuation time period. The second fluctuation frequency threshold is greater than the first fluctuation frequency threshold, and the second fluctuation amplitude threshold is greater than the first fluctuation amplitude threshold.

7. The phase recognition method based on an intelligent fusion terminal as described in claim 1, characterized in that, The second attention weight coefficient also includes a second non-stationary period weight, which is less than the second stationary period weight; The calculation of the correlation coefficient between the single-phase voltage time series data and the three-phase voltage time series data based on the second attention weighting coefficient includes: The single-phase voltage time series data is divided into fluctuating single-phase voltage data and non-fluctuating single-phase voltage data; The fluctuating single-phase voltage data and the non-fluctuating single-phase voltage data are weighted based on the second attention weight coefficient to obtain the weighted single-phase voltage time series data. For any one of the weighted single-phase voltage time series data and the three-phase voltage time series data, perform the second operation to obtain a multi-dimensional feature vector; calculate the correlation coefficient between the weighted single-phase voltage time series data and the three-phase voltage time series data based on the multi-dimensional feature vectors corresponding to the weighted single-phase voltage time series data and the three-phase voltage time series data. The second operation includes: Basic voltage features and derived features are extracted from voltage time series data. The basic voltage features include voltage amplitude, voltage mean, voltage standard deviation, and number of voltage abrupt changes. The derived features include the percentage of duration of abrupt changes and the concentration of abrupt change amplitude. A fast Fourier transform is performed on the voltage time series data to obtain a frequency domain signal, and low-order frequency domain features are extracted from the frequency domain signal; the low-order frequency domain features include the fundamental phase angle and the proportion of harmonics. The frequency domain signal is filtered by a narrowband filter to obtain the filtered fundamental wave signal data. The fluctuation variance of the fundamental wave amplitude is calculated based on the fundamental wave signal data as the fundamental wave stability feature. A multidimensional feature vector is constructed based on the low-frequency domain features, the fundamental frequency stability features, the fundamental voltage features, and the derived features.

8. A phase recognition device based on an intelligent fusion terminal, characterized in that, include: The data acquisition module is used to acquire single-phase voltage time-series data through the smart meters of the target users and to acquire three-phase voltage time-series data through the output master meter of the transformer substation. The user segmentation module is used to determine the user load type based on the target user's historical power consumption data. The first phase identification module is used to calculate the correlation coefficient between the single-phase voltage time series data and the three-phase voltage time series data based on the first attention weight coefficient if the user load type is a stable load type. The first attention weight coefficient includes the weight of the first stable period; The weight of the first stable period is the weight of the stable period data in the single-phase voltage time series data; The second phase identification module is configured to perform a first operation to obtain a correlation coefficient if the user load type is a fluctuating load type; wherein the first operation includes: Based on the single-phase voltage time-series data, the voltage fluctuation frequency and voltage fluctuation amplitude are extracted; If the voltage fluctuation frequency exceeds a first fluctuation frequency threshold and / or the voltage fluctuation amplitude exceeds a first fluctuation amplitude threshold, then a non-fluctuation period is determined based on the single-phase voltage time series data; target single-phase voltage data corresponding to the non-fluctuation period is extracted from the single-phase voltage time series data, and target three-phase voltage data corresponding to the non-fluctuation period is extracted from the three-phase voltage time series data; the correlation coefficient between the target single-phase voltage data and the target three-phase voltage data is calculated. If the voltage fluctuation frequency does not exceed the first fluctuation frequency threshold and the voltage fluctuation amplitude does not exceed the first fluctuation amplitude threshold, then the correlation coefficient between the single-phase voltage time series data and the three-phase voltage time series data is calculated based on the second attention weight coefficient; the second attention weight coefficient includes the weight of the second stable period; the weight of the second stable period is greater than the weight of the first stable period. The phase analysis module is used to determine the target phase corresponding to the target user based on the correlation coefficient.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images