Intelligent metering switch with electric energy error monitoring function

Abstract

The present application relates to the technical field of electric energy error monitoring, and particularly relates to an intelligent measurement switch with electric energy error monitoring function. The switch comprises: a data acquisition unit for acquiring a voltage sequence in a current period and a theoretical voltage at each time; a reference weight acquisition unit for acquiring an abnormality degree of each real-time voltage and a reference weight; a stability coefficient acquisition module for determining an initial window, obtaining a stability coefficient corresponding to the window based on real-time voltages in the window; a window sliding module for sliding the initial window at the beginning of the voltage sequence, updating the window before each sliding based on the stability coefficient, and acquiring real-time voltages in the window when each updated window is not sliding; and an error monitoring module for acquiring a real voltage at each time by using the real-time voltages in the window when each updated window is not sliding, the reference weight of the real-time voltage, and a constructed sensor error model. The present application can improve the accuracy of electric energy error monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric energy error monitoring, and in particular to an intelligent measurement switch with electric energy error monitoring function. BACKGROUND

[0002] With the acceleration of smart grid construction, distributed energy, electric vehicle charging piles and other new loads are connected, and the power grid presents the characteristics of "two-way interaction and dynamic balance", which puts forward higher requirements for measurement equipment. Some enterprises have launched intelligent measurement switch products. The intelligent measurement switch is installed at the total incoming line of the meter box, and can realize node voltage, current, electric energy and other information monitoring, has intelligent functions such as high-precision measurement, multi-mode communication and topology identification, and can provide support for transformer area line loss calculation, electric energy meter error analysis and accurate fault positioning. Combined with edge computing of fusion terminal, the overall operation state of the metering box is holographic perceived and monitored, which can improve the intelligent management level of the transformer area and the economic operation level of the power grid.

[0003] In the electric energy error monitoring of the intelligent measurement switch, the voltage acquisition is generally collected by a sensor, so there may be measurement errors (such as zero drift, gain error, and non-linear error, which are generally caused by high temperature in summer or low temperature in winter, electromagnetic interference or mechanical vibration impact) affecting the electric energy error calculation result. The least square method can efficiently calibrate device parameters and quantify errors, but the least square method is a periodic calibration method, the static calibration model cannot track changes in real time, lacks a dynamic updating mechanism, and is prone to overfitting of abnormal values, resulting in parameter estimation deviation and affecting the accuracy and real-time performance of the result. SUMMARY

[0004] In order to solve the above technical problems, the purpose of the present application is to provide an intelligent measurement switch with electric energy error monitoring function, and the technical solution adopted is as follows:

[0005] An embodiment of the present application provides an intelligent measurement switch with electric energy error monitoring function, which comprises:

[0006] A data acquisition unit is configured to collect real-time voltages at each time in a current period of a transformer area power grid by using sensors on the intelligent measurement switch, form a voltage sequence, and obtain theoretical voltages at each time in the period;

[0007] A reference weight acquisition unit is configured to analyze real-time voltages around a real-time voltage in the voltage sequence to obtain an abnormality degree of the real-time voltage, and obtain a reference weight of each real-time voltage according to the abnormality degree of each real-time voltage;

[0008] a stability coefficient acquisition module, configured to determine an average data length according to a phase point distribution of each real-time voltage in a last period of a current period, and obtain an initial window; and obtain a stability coefficient corresponding to the window based on the real-time voltage in the window;

[0009] a window sliding module, configured to start sliding on the voltage sequence with a set step length by using the initial window, update the initial window according to the stability coefficients corresponding to the initial window before and after the sliding to obtain a first update window, perform secondary sliding on the voltage sequence by using the first update window, and update the first update window according to the stability coefficients corresponding to the first update window before and after the sliding to obtain a second update window, and so on until the voltage sequence is traversed, and the real-time voltage in the window when each update window is not slid is obtained;

[0010] an error monitoring module, configured to obtain a real-time voltage at each moment by using the real-time voltage in the window when each update window is not slid, a reference weight of the real-time voltage, a theoretical voltage and a sensor error model constructed in combination with a least square method.

[0011] Preferably, the abnormality degree of one real-time voltage in the voltage sequence is obtained by analyzing the real-time voltage around the real-time voltage, including:

[0012] a data sequence is composed by collecting a set number of real-time voltages on both sides of the real-time voltage as the center of the real-time voltage in the voltage sequence, and different data sequences corresponding to different values of the set number are obtained; a forward slope between the real-time voltage and a previous real-time voltage is calculated and recorded as a forward slope; a backward slope between the real-time voltage and a subsequent real-time voltage is calculated and recorded as a backward slope; an absolute value of a difference between the forward slope and the backward slope is obtained, and multiplied by an average value of standard deviations of different data sequences corresponding to different values of the set number to obtain the abnormality degree of the real-time voltage.

[0013] Preferably, the reference weight of each real-time voltage is obtained according to the abnormality degree of each real-time voltage, including:

[0014] a first coefficient of one real-time voltage in the voltage sequence is recorded as a difference between a first preset value and a normalized value of the abnormality degree of the real-time voltage; and the reference weight of each real-time voltage is obtained by comparing the first coefficient of each real-time voltage with a sum of the first coefficients of all real-time voltages.

[0015] Preferably, the average data length is determined according to a phase point distribution of each real-time voltage in a last period of a current period, and the initial window is obtained, including:

[0016] The CUSUM algorithm is used to process the sequence composed of the real-time voltage data in the previous period of the current period in time sequence, to obtain all phase points, to calculate the average value of the data length between each two phase points and to round up to obtain the average data length, and to take the average data length as the size of the window to obtain the initial window.

[0017] Preferably, the stability coefficient corresponding to the window is obtained based on the real-time voltage in the window, including:

[0018] The product of a real-time voltage in a window and the number of real-time voltages in the window is subtracted from the sum of the real-time voltages in the window and the absolute value is obtained to obtain the second coefficient corresponding to the real-time voltage; the reciprocal of the standard deviation of all real-time voltages in the window is multiplied by the reciprocal of the average value of the second coefficients of all real-time voltages and is normalized to obtain the stability coefficient corresponding to the window.

[0019] Preferably, the initial window is updated to obtain a first updated window according to the stability coefficients corresponding to the initial window before and after sliding, including:

[0020] If the stability coefficient corresponding to the initial window before sliding is greater than the stability coefficient corresponding to the initial window after sliding, the absolute value of the difference between the stability coefficient corresponding to the initial window after sliding and the stability coefficient corresponding to the initial window before sliding is added to the first preset value, and the first updated size is obtained by rounding, and the first updated window is obtained by expanding the initial window after sliding forward by the first updated size; if the stability coefficient corresponding to the initial window before sliding is equal to the stability coefficient corresponding to the initial window after sliding, the initial window is the first updated window; if the stability coefficient corresponding to the initial window before sliding is less than the stability coefficient corresponding to the initial window after sliding, the first updated size is obtained by rounding the first preset value minus the absolute value of the difference between the stability coefficient corresponding to the initial window after sliding and the stability coefficient corresponding to the initial window before sliding, and the first updated window is obtained by reducing the initial window after sliding forward by the first updated size.

[0021] Preferably, the real-time voltage in each update window without sliding, the reference weight of the real-time voltage, the theoretical voltage and the constructed sensor error model are used to obtain the true voltage at each time point by combining the least square method, including:

[0022] The gain coefficient and the zero point offset coefficient of the sensor error model are obtained by fitting based on the real-time voltage in the first updated window without sliding, the reference weight of the real-time voltage and the theoretical voltage by using the least square method, wherein the theoretical voltage is taken as the true voltage during fitting; the last time point real-time voltage in the first updated window is brought into the sensor error model with the determined gain coefficient and zero point offset coefficient to obtain the last time point true voltage; the last time point true voltage in each update window is obtained in the same way;

[0023] The sensor error model is: y=ax+b, wherein y represents the real-time voltage, x represents the real voltage, a represents a gain coefficient, and b represents a zero-point offset coefficient.

[0024] Preferably, after obtaining the real voltage at each time point, the method further comprises:

[0025] The power error value at the time point is obtained by subtracting the real-time voltage at the time point from the real voltage at the time point and taking the absolute value.

[0026] The embodiment of the application has at least the following beneficial effects: the application collects the real-time voltage at each time point in the current period of the distribution network power grid by using the sensor on the intelligent measurement switch, forms a voltage sequence, and obtains the theoretical voltage at each time point in the period; then, the abnormality degree of each real-time voltage in the voltage sequence is calculated, and then the reference weight of each real-time voltage is obtained; further, the average data length is determined according to the phase point distribution of each real-time voltage in the previous period of the current period, and the initial window is obtained, and then the stability coefficient corresponding to the window is obtained based on the real-time voltage in the window; finally, the initial window is started to slide on the voltage sequence with a set step size, the initial window is updated to obtain a first update window according to the stability coefficients corresponding to the initial window before and after sliding, the first update window is slid on the voltage sequence for a second time, and the second update window is obtained by updating the first update window according to the stability coefficients corresponding to the first update window before and after sliding, and so on, until the voltage sequence is traversed, the real-time voltage in the window when each update window is not slid is obtained, the calculation of a large amount of voltage data is converted into a sliding window calculation mode according to the change of the real-time voltage in the window, the characteristics of the data around each real-time voltage can be captured in more detail, and then the error of the real-time voltage at each time point is monitored by combining the real-time voltage in the window when each update window is not slid, the reference weight of the real-time voltage, and the constructed sensor error model and combining the least square method, the real voltage at each time point is obtained, and the real-time and accuracy of the least square method for calibrating and quantifying the error of the real-time voltage collected by the sensor are improved. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, and the advantages thereof, a brief introduction will be given to the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can be obtained from these drawings without creative labor.

[0028] Figure 1 A unit block diagram of an intelligent measurement switch with a power error monitoring function is provided for the embodiment of the application. DETAILED DESCRIPTION

[0029] In order to further clarify the technical means and effects taken by the present application to achieve the predetermined inventive purpose, the specific implementation, structure, features and effects of the intelligent measurement switch with electric energy error monitoring function according to the present application are described in detail below in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0031] The specific scheme of the intelligent measurement switch with electric energy error monitoring function provided by the present application is described in detail below in combination with the drawings.

[0032] Embodiment:

[0033] The main application scenario of the present application is to perform electric energy error monitoring of a transformer area through the intelligent measurement switch installed at the total incoming line of a meter box.

[0034] Please refer to Figure 1 which shows a unit block diagram of the intelligent measurement switch with electric energy error monitoring function provided by the embodiment of the present application, which includes the following units:

[0035] The data acquisition unit is used to acquire real-time voltages at each time in the current period of the transformer area power grid by using the sensors on the intelligent measurement switch, to form a voltage sequence, and to obtain the theoretical voltage at each time in the period.

[0036] The main purpose of the present application is to calculate the sensor error model by dynamic weighted least squares method, to improve the real-time performance and accuracy of the least squares method calibration quantization sensor error result.

[0037] Therefore, the acquisition period is set, and the data acquired in each period is archived once. The length of the period can be determined according to the actual situation, for example, one day or one week. The real-time voltage at each time in the current period of the transformer area power grid needs to be acquired by using the sensors on the intelligent measurement switch to form a voltage sequence, and the acquisition interval can be set to 1 second once, which can be adjusted according to the actual situation.

[0038] Further, a known voltage generated by a standard voltage source is acquired, such as a rated voltage of 220V±20% (i.e. 176V~264V) of the intelligent metering switch, a standard voltage source with the known voltage in the interval is selected to access the intelligent metering switch, and a corresponding output voltage is obtained, if the collected real-time voltage is the same as a certain output voltage data, the voltage source voltage corresponding to the output voltage is the theoretical value of the real-time voltage data, that is, the theoretical voltage, and thus the theoretical voltage at each time in the current period can be obtained.

[0039] The reference weight acquisition unit is configured to analyze the real-time voltages around a real-time voltage in the voltage sequence to obtain an abnormality degree of the real-time voltage, and obtain a reference weight of each real-time voltage according to the abnormality degree of each real-time voltage.

[0040] The intelligent metering switch converts the actual voltage into a measurable electrical signal through a voltage sensor (such as a voltage transformer or a resistance divider), but the sensor may have the following errors: zero drift (output is not zero when there is no input), gain error (output and input are not strictly proportional), and non-linear error (output and input are in a curve relationship (such as saturation effect)). Among them, high temperature in summer or low temperature in winter will cause zero drift and gain error, switching operation in high-voltage line, arc discharge will produce high-frequency transient pulse, which will be conducted to the sensor through the power line or signal line, forming a sharp peak noise, adjacent high-power motor, frequency converter and other equipment will emit electromagnetic waves, which will induce interference voltage in the sensor circuit, and the sensor is not firmly installed or long-term vibration will cause internal components (such as resistance, capacitance) to contact poorly, and the change of contact resistance will introduce random error, mechanical stress will act on the elastic element (such as strain gauge) of the sensor, if the stress isolation structure is not designed, the stress will change the resistance value through material deformation, and additional error will be generated.

[0041] Therefore, it is necessary to calculate the error model of the sensor. The least square method is simple and fast in convergence, but if there are abnormal data in the measurement data, such as abnormal voltage data caused by transient pulse caused by device switching operation, it will affect the analysis, so it is necessary to reduce the influence of abnormal values in the data and suppress the data quality dependence of the least square method. The main characteristics of abnormal voltage data are suddenness, outlyingness, low repeatability, and isolated distribution. Therefore, the abnormal degree of each voltage data can be calculated based on such characteristics to obtain the reference weight of the data, and the higher the abnormal degree, the lower the reference weight.

[0042] The abnormality degree of the real-time voltage is obtained by analyzing the real-time voltage around the real-time voltage in the voltage sequence. Specifically, since there may be a performance deviation between local data and overall data, for example, a small fluctuation may show abnormality in the range where it is located, but when the data length becomes longer, the abnormality may be weakened. Therefore, in order to ensure the authenticity of the voltage sequence abnormality, the data sequence containing the real-time voltage to be analyzed is obtained by randomly sampling the data with a random data length around the real-time voltage to be analyzed and the surrounding data in the voltage sequence, and then the analysis is continued.

[0043] A set number of real-time voltages are collected on both sides of a real-time voltage in the voltage sequence as the center of the real-time voltage to form a data sequence, and different data sequences corresponding to different values of the set number are obtained; the slope between the real-time voltage and the previous real-time voltage is calculated, denoted as the forward slope; the slope between the real-time voltage and the next real-time voltage is calculated, denoted as the backward slope; the absolute value of the difference between the forward slope and the backward slope is obtained, and multiplied by the average of the standard deviations of the different data sequences corresponding to different values of the set number to obtain the abnormality degree of the real-time voltage.

[0044] The calculation model of the abnormality degree is specifically:

[0045] ,

[0046] Wherein, represents the abnormality degree of the real-time voltage le in the voltage sequence, represents the slope between the real-time voltage le and the previous real-time voltage (forward slope), represents the slope between the real-time voltage le and the next real-time voltage (backward slope), and if The greater the value, the more likely the forward slope and the backward slope are opposite trend slopes, and the more likely the real-time voltage le is highlighted.

[0047] represents the standard deviation of the i-th data sequence in the different data sequences corresponding to different values of the set number, and n represents the number of times of collecting real-time voltages on both sides of the real-time voltage le, it should be noted that the collection is random collection, that is, the set number collected each time is uncertain when collecting data on both sides, and at the same time, in order to facilitate the calculation of the standard deviation, the length of the data sequence obtained each time should be greater than or equal to 3, and at the same time, in order to prevent the real-time voltages with large time difference from affecting the calculation result, the data length of the data sequence should be at most equal to 10 times the minimum data length, that is, 30, and the implementer can adjust it according to the actual situation; n is the number of times of collection, which can also represent the number of data sequences, n is in the range of greater than 3; if A smaller value indicates that the local characteristics of the real-time voltage le are relatively stable, resulting in lower outlier activity, a higher probability of repeatability, and less obvious isolation in the distribution. Therefore, if the selected voltage data has significant salientity and a large local mean standard deviation, the anomaly of the real-time voltage le is relatively large.

[0048] This allows us to determine the degree of anomaly for each real-time voltage in the voltage sequence. Furthermore, a reference weight is assigned to each real-time voltage based on its degree of anomaly. Specifically, the difference between a first preset value and the normalized value of the degree of anomaly for a real-time voltage in the voltage sequence is recorded as the first coefficient of that real-time voltage. The reference weight for each real-time voltage is obtained by comparing its first coefficient with the sum of the first coefficients of all real-time voltages. The greater the degree of anomaly of a real-time voltage, the lower its reference weight. The first preset value is set to 1.

[0049] The stability coefficient acquisition module is used to determine the average data length based on the distribution of stage points of each real-time voltage in the previous cycle of the current cycle, and to obtain an initial window; and to obtain the stability coefficient corresponding to the window based on the real-time voltage within the window.

[0050] Since real-time voltage is related to user electricity consumption behavior, meaning voltage changes may occur in stages, the least squares method, being a periodic calibration method, cannot track changes in real time when new variations occur, such as voltage sampling shifts due to aging electrolytic capacitors or sudden load changes (e.g., motor startup). Therefore, a sliding window calibration can be used, where a sliding window is set, and the model is refitted using only the data within the window. However, because real-time voltage data is acquired in real time, there is a delay between data acquisition and calculation. If the voltage data changes frequently during a certain stage, high real-time performance is required, necessitating minimizing parameter update delays; in this case, a smaller sliding window is used. If the changes are relatively stable, the real-time requirement can be appropriately reduced, and a larger window can be selected to obtain smoother calibration results.

[0051] Furthermore, the size of the initial window is determined by analyzing the real-time voltage data from the previous cycle of the current cycle and sliding it over the voltage sequence.

[0052] Specifically, the CUSUM algorithm (cumulative sum algorithm) is used to process the real-time voltage data in the previous cycle of the current cycle according to the time sequence to obtain all stage points. After each stage point is found, the benchmark is modified to the average value of the data after the stage point. In this way, all stage points in the real-time voltage data in the previous cycle are found according to the time sequence. Furthermore, the average value of the data length between every two stage points is calculated and rounded up to obtain the average data length. The average data length is used as the window size to obtain the initial window.

[0053] The window needs to be updated according to the trend of the data in the window when the window slides, so that the stability coefficient can be obtained according to the data in the window. Specifically, the product of a real-time voltage in a window and the number of real-time voltages in the window is subtracted from the sum of the real-time voltages in the window, and the absolute value is obtained, to obtain the second coefficient corresponding to the real-time voltage; the reciprocal of the standard deviation of all real-time voltages in the window is multiplied by the reciprocal of the average of the second coefficients corresponding to all real-time voltages and normalized to obtain the stability coefficient corresponding to the window.

[0054] The calculation model of the stability coefficient is specifically:

[0055]

[0056] Among them, represents the stability coefficient of the real-time voltage in the window k, represents the standard deviation of the real-time voltage in the window k, and the smaller the standard deviation, the smaller the difference between the data in the window, and the greater the data stability. represents the value of the hth real-time voltage in the window, and H represents the number of real-time voltages in the window, represents the sum of the real-time voltages in the window k, represents the second coefficient corresponding to the hth real-time voltage in the window, The average of the second coefficients corresponding to all real-time voltages is also the average difference between the value of each real-time voltage multiplied by the number of real-time voltages and the sum of real-time voltages. If the difference is large, it indicates that there are obvious changes in the values of the real-time voltages in the window, and the number may be large, so there may be frequent changes in the window, and the stability is low. Norm represents the normalization operation.

[0057] Therefore, the stability coefficient corresponding to each window can be calculated when the window slides on the voltage sequence.

[0058] The window sliding module is configured to start sliding on the voltage sequence with a set step length using the initial window, update the initial window to obtain a first updated window according to the stability coefficients corresponding to the initial window before and after sliding, slide on the voltage sequence using the first updated window, and update the first updated window to obtain a second updated window according to the stability coefficients corresponding to the first updated window before and after sliding. In this way, the real-time voltages in the window when each updated window is not sliding are obtained by iterating the voltage sequence.

[0059] The above calculation method of the initial window and the stability coefficient corresponding to the window is obtained, and then the initial window is used as the starting window to slide and analyze on the voltage sequence.

[0060] ​The initial window is started to slide on the voltage sequence with a set step size, an initial update window is obtained by updating the initial window according to the stability coefficients corresponding to the initial window before and after sliding, the initial update window is slid on the voltage sequence, and a secondary update window is obtained by updating the initial update window according to the stability coefficients corresponding to the initial update window before and after sliding, and the like, until the voltage sequence is traversed, and real-time voltages in the windows of each update window when not sliding are obtained.

[0061] The set step size is 1, an initial update window is obtained by updating the initial window according to the stability coefficients corresponding to the initial window before and after sliding, specifically, if the stability coefficient corresponding to the initial window before sliding is greater than the stability coefficient corresponding to the initial window after sliding, then the absolute value of the difference between the stability coefficient corresponding to the initial window after sliding and the stability coefficient corresponding to the initial window before sliding is added to the first preset value, and the result is rounded to obtain an update size, and the initial window after sliding is enlarged by the update size to obtain the initial update window; if the stability coefficient corresponding to the initial window before sliding is equal to the stability coefficient corresponding to the initial window after sliding, then the initial window is the initial update window; if the stability coefficient corresponding to the initial window before sliding is less than the stability coefficient corresponding to the initial window after sliding, then the first preset value is subtracted from the absolute value of the difference between the stability coefficient corresponding to the initial window after sliding and the stability coefficient corresponding to the initial window before sliding, and the result is rounded to obtain the update size, and the initial window after sliding is reduced by the update size to obtain the initial update window.

[0062] Taking the initial window as an example, the calculation model of the update size is as follows:

[0063]

[0064] wherein, represents the stability coefficient corresponding to the initial window before sliding, represents the stability coefficient corresponding to the initial window after sliding; L represents the update size, represents the size of the initial window, and round represents the rounding operation.

[0065] Similarly, the same operation is performed on the initial update window after sliding once, and the stability coefficients before and after sliding are used to determine whether the size of the initial update window needs to be enlarged, reduced, or not updated.

[0066] The real-time voltages in the windows of each update window when not sliding refer to the data in the window when not sliding after the window is updated, for example, the initial update window is obtained by updating the initial window after sliding, and the initial update window does not slide, at this time, the real-time voltages in the initial update window are obtained, and for the initial window, the data in the window when not sliding is obtained.

[0067] ​For example, the size of the initial window is 4, the voltages are A, B, C, D, E, the data in the window before sliding is A, B, C, D, and the data in the window after sliding is B, C, D, E. At this time, the stability coefficients corresponding to A, B, C, D in the window before sliding and the stability coefficients corresponding to B, C, D, E in the window after sliding are calculated. If the two stability coefficients are equal, the size of the initial window is unchanged and directly used as the size of the first update window. At this time, the data in the first update window is B, C, D, E. If the stability coefficient corresponding to the initial window before sliding is greater than the stability coefficient corresponding to the initial window after sliding, the size of the first update window is 5. At this time, the size of the initial window after sliding is expanded by 1, and the size of the first update window is 5. The data in the first update window is A, B, C, D, E. The data in the window of the first update window before sliding is A, B, C, D, E. If the stability coefficient corresponding to the initial window before sliding is less than the stability coefficient corresponding to the initial window after sliding, the size of the first update window is 3. At this time, the size of the initial window after sliding is reduced by 1, and the size of the first update window is 3. The data in the first update window is C, D, E. The data in the window of the first update window before sliding is C, D, E.

[0068] Thus, the real-time voltages in the window of each update window before sliding can be obtained. Then, the real-time voltages in a window are used as an analysis unit for subsequent error monitoring analysis.

[0069] The error monitoring module is configured to obtain the true voltage at each time point by using the real-time voltages in the window of each update window before sliding, the reference weights of the real-time voltages, the theoretical voltages, and the constructed sensor error model in combination with the least square method.

[0070] The real-time voltages in the window of the initial window and each update window before sliding are obtained, and thus the sensor error model is constructed. The sensor error model is y=ax+b, where y represents the voltage measured by the sensor, that is, the real-time voltage, x represents the true voltage, a represents the gain coefficient, and b represents the zero-point offset coefficient.

[0071] Further, the true voltage at each time point is obtained by using the real-time voltages in the window of each update window before sliding, the reference weights of the real-time voltages, the theoretical voltages, and the constructed sensor error model in combination with the least square method. Specifically, the gain coefficient and the zero-point offset coefficient of the sensor error model are obtained by fitting based on the real-time voltages in the window of the first update window before sliding, the reference weights of the real-time voltages, and the theoretical voltages by using the least square method. During the fitting, the theoretical voltages are used as the true voltages. The last time point real-time voltage in the first update window is brought into the sensor error model with the determined gain coefficient and zero-point offset coefficient to obtain the last time point true voltage. Thus, the true voltage at each time point can be obtained.

[0072] In essence, the gain coefficient and the zero point offset coefficient of the sensor error model corresponding to each update window are determined by fitting the corresponding data in the non-sliding update window based on the least square method, the real voltage at the last time in the non-sliding update window is calculated based on the sensor error model corresponding to each update window.

[0073] For the initial window, the real-time voltage, the reference weight of the real-time voltage and the theoretical voltage in the initial window are used to perform the least square weighted fitting operation to determine the gain coefficient and the zero point offset coefficient of the sensor error model when the initial window is not sliding, then the real-time voltage in the initial window is brought into the sensor error model to obtain the real voltage at each time in the initial window. Finally, the power error value at each time is obtained according to the real-time voltage and the real voltage at each time, that is, the power error value at each time is obtained by subtracting the real-time voltage at each time from the real voltage at each time and taking the absolute value.

[0074] Thus, the real voltage at each time can be obtained, and the power error value at each time can be obtained.

[0075] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The above-mentioned specific embodiments of the present application are described. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0076] Each embodiment in the specification is described in a progressive manner, and the same and similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

[0077] The above-mentioned is only the preferred embodiment of the present application, and does not limit the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. An intelligent measuring switch with power error monitoring function, characterized in that, The switch includes: The data acquisition unit is used to collect the real-time voltage of the power grid in the current period of the distribution area using the sensors on the smart measuring switch, form a voltage sequence, and obtain the theoretical voltage at each moment in the period. The reference weight acquisition unit is used to analyze the real-time voltages around a real-time voltage in the voltage sequence to obtain the degree of anomaly of that real-time voltage; and to obtain the reference weight of each real-time voltage based on the degree of anomaly of each real-time voltage. The stability coefficient acquisition module is used to determine the average data length based on the distribution of stage points of each real-time voltage in the previous cycle of the current cycle, and to obtain an initial window; and to obtain the stability coefficient corresponding to the window based on the real-time voltage within the window. The window sliding module is used to slide an initial window on the voltage sequence with a set step size. The initial window is updated according to the stability coefficient before and after the initial window slide to obtain a first update window. The first update window is used to slide a second time on the voltage sequence, and the first update window is updated according to the stability coefficient before and after the first update window slide to obtain a second update window. This process is repeated until the voltage sequence is traversed, and the real-time voltage in the window when each update window is not sliding is obtained. The error monitoring module is used to obtain the real voltage at each time step by utilizing the real-time voltage within the window when the update window is not sliding, the reference weight of the real-time voltage, the theoretical voltage, and the constructed sensor error model, combined with the least squares method. The analysis of the real-time voltages surrounding a given real-time voltage within a voltage sequence to determine the degree of anomaly of that real-time voltage includes: A set number of real-time voltages are collected from both sides of a given real-time voltage within a voltage sequence to form a data sequence. Different data sequences are obtained when the set number of voltages takes different values. The slope between the real-time voltage and its previous real-time voltage is calculated and denoted as the forward slope. The slope between the real-time voltage and its next real-time voltage is calculated and denoted as the backward slope. The absolute value of the difference between the forward slope and the backward slope is obtained and multiplied by the average standard deviation of the different data sequences corresponding to different values ​​of the set number of voltages to obtain the degree of anomaly of the real-time voltage. The step of obtaining the reference weight for each real-time voltage based on the degree of anomaly of each real-time voltage includes: The difference between the first preset value and the normalized value of the anomaly degree of a real-time voltage in the voltage sequence is recorded as the first coefficient of the real-time voltage; the first coefficient of each real-time voltage in the voltage sequence is compared with the sum of the first coefficients of all real-time voltages to obtain the reference weight of each real-time voltage, wherein the first preset value is 1; The process of obtaining the stability coefficient corresponding to a window based on the real-time voltage within that window includes: The second coefficient corresponding to a real-time voltage is obtained by subtracting the product of a real-time voltage and the number of real-time voltages in a window from the sum of real-time voltages in the window and taking the absolute value. The stability coefficient corresponding to the window is obtained by multiplying the reciprocal of the standard deviation of all real-time voltages in the window with the reciprocal of the average of the second coefficients corresponding to all real-time voltages and normalizing the result. The step of updating the initial window based on the stability coefficients corresponding to the initial window before and after sliding to obtain an updated window includes: If the stability coefficient corresponding to the initial window before sliding is greater than the stability coefficient corresponding to the initial window after sliding, then the absolute value of the difference between the stability coefficient corresponding to the initial window after sliding and the stability coefficient corresponding to the initial window before sliding is added to a first preset value and rounded to obtain an update size. The initial window after sliding is expanded forward using this update size to obtain an update window. If the stability coefficient corresponding to the initial window before sliding is equal to the stability coefficient corresponding to the initial window after sliding, then the initial window is an update window. If the stability coefficient corresponding to the initial window before sliding is less than the stability coefficient corresponding to the initial window after sliding, then the absolute value of the difference between the stability coefficient corresponding to the initial window after sliding and the stability coefficient corresponding to the initial window before sliding is subtracted from the first preset value and rounded to obtain an update size. The initial window after sliding is shrunk forward using this update size to obtain an update window.

2. The intelligent measuring switch with power error monitoring function according to claim 1, characterized in that, The step of determining the average data length and obtaining the initial window based on the distribution of real-time voltage stages in the previous cycle of the current cycle includes: The CUSUM algorithm is used to process the real-time voltage data from the previous cycle of the current cycle according to the time sequence, obtain all stage points, calculate the average data length between every two stage points and round it up to obtain the average data length, and use the average data length as the window size to obtain the initial window.

3. The intelligent measuring switch with power error monitoring function according to claim 1, characterized in that, The process of obtaining the true voltage at each moment by utilizing the real-time voltage within the window when it is not sliding, the reference weight of the real-time voltage, the theoretical voltage, and the constructed sensor error model, combined with the least squares method, includes: Based on the real-time voltage, reference weight of the real-time voltage, and theoretical voltage within the window when the update window is not sliding, the least squares method is used to fit and obtain the gain coefficient and zero-point offset coefficient of the sensor error model, where the theoretical voltage is used as the true voltage during fitting. The real-time voltage at the last moment within the first update window is substituted into the sensor error model with the gain coefficient and zero-point offset coefficient determined to obtain the true voltage at the last moment. Similarly, the true voltage at the last moment within each of the other update windows is obtained. The sensor error model is: y=ax+b, where y represents the real-time voltage, x represents the true voltage, a represents the gain coefficient, and b represents the zero-point offset coefficient.

4. The intelligent measuring switch with power error monitoring function according to claim 1, characterized in that, After obtaining the actual voltage at each moment, the following is also included: The energy error value at that moment is obtained by subtracting the real-time voltage from the actual voltage and taking the absolute value.

Images