A data-driven online estimation method for DC energy meter measurement error

Abstract

The present invention discloses a data-driven online estimation method for DC electric energy meter measurement errors, comprising: collecting a first clock after correction of an electric energy meter in a DC distribution network, and collecting first active energy data and first voltage data in the DC distribution network; performing abnormal value detection on the first active energy data and the first voltage data according to normal distribution models corresponding to the first active energy data and the first voltage data, and correcting the abnormal values to obtain second active energy data and second voltage data; establishing an optimization model for DC electric energy meter measurement error estimation based on the second active energy data and the second voltage data, introducing equivalent variables to perform linear substitution on the optimization target of the optimization model, obtaining a convex optimization model and solving it, and obtaining an online measurement error estimation value, so that operation and maintenance can be performed according to the online measurement error estimation value; the accuracy of the electric energy meter measurement error estimation model in the DC distribution network can be improved, and the model solution efficiency can be improved.

Description

Technical Field

[0001] The present invention relates to the field of error estimation of direct current (DC) ammeters, and in particular to a data-driven online estimation method for measurement errors of DC electric energy meters. Background Art

[0002] Driven by policies such as the "dual carbon" initiative and the construction of a new power system, large-scale distributed renewable energy is being connected to distribution networks. Common distributed renewable energy sources such as photovoltaics, wind power, and energy storage are typically DC power or converted to DC power after rectification, and then connected to the AC distribution network using inverters. Furthermore, with the continuous advancement of power electronics technology in recent years, the overall load-side structure has also undergone significant changes. When connected to the AC distribution system, loads such as electric vehicles, computers, and mobile phones require AC / DC power supply devices. Household inverter devices such as air conditioners, refrigerators, and washing machines, as well as energy storage systems, also require AC / DC / AC devices to achieve frequency conversion and ensure power quality and reliability. This has led to the emergence of DC distribution networks and AC / DC hybrid distribution networks. DC metering is one of the most critical operational safeguards for DC distribution networks, and accurate and reliable DC metering is crucial for their safe and stable operation. However, the operating conditions of DC distribution networks and AC / DC hybrid distribution networks are complex and diverse. Different types of DC loads and power sources, such as DC charging stations and photovoltaic grid-connected systems, have different operating conditions and different impacts. The voltage and current waveforms under different operating conditions will be distorted, and there will be significant differences from the ideal DC load waveform, affecting the accuracy of DC energy meter measurement. Therefore, it is of great significance to carry out measurement error estimation of DC energy meters.

[0003] Currently, most operations and maintenance personnel carry calibration equipment to conduct on-site inspections. However, with the large number of electricity meter locations, manual inspections are time-consuming, labor-intensive, and impractical. Currently, electricity meters are equipped to measure user electricity usage data, which essentially reflects the physical characteristics of grid operation. Mining and analyzing this data can facilitate online estimation of meter error. Existing technologies often employ online estimation models for meter error based on the law of conservation of energy.

[0004] However, current energy meter error estimation models based on the law of conservation of electric energy are mostly applied in AC distribution network scenarios, and research on their application in DC distribution network scenarios is still lacking. Secondly, current energy meter error estimation models for AC distribution networks are significantly affected by line losses. The accuracy of AC energy meter error estimation decreases as the proportion and rate of change of line losses increase. In DC distribution networks, in addition to line losses, losses in links such as AC / DC converters and DC / DC converters are even more significant and account for a larger proportion. Ignoring the calculation of typical loss links in DC distribution networks will lead to increased errors in the original model calculation results, or even failure. Furthermore, current energy meter error estimation models for AC distribution networks are mostly solved using the least squares method, which requires the number of collected data periods to be greater than the total number of energy meters, making its application conditions relatively demanding. Summary of the Invention

[0005] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art and propose a data-driven online estimation method for DC electric energy meter measurement errors, which can improve the accuracy of the electric energy meter measurement error estimation model in the DC distribution network and improve the efficiency of the model solution.

[0006] The present invention provides a data-driven online estimation method for DC electric energy meter measurement error, comprising:

[0007] Collecting a calibrated first clock of an electric energy meter in a DC distribution network, and collecting first active energy data of the same frequency within a preset time from an AC input side total meter and a DC electric energy meter in the DC distribution network under the first clock, as well as first voltage data from the AC input side total meter;

[0008] performing outlier detection on the first active energy data and the first voltage data according to a first normal distribution model of the first active energy data and a second normal distribution model of the first voltage data, and correcting the outliers to obtain second active energy data and second voltage data, respectively;

[0009] An optimization model for estimating the measurement error of a DC electric energy meter is established based on the second active energy data and the second voltage data. Equivalent variables are introduced to linearly replace the optimization objective of the optimization model to obtain a convex optimization model and solve it to obtain an online measurement error estimate, so that operation and maintenance are performed based on the online measurement error estimate.

[0010] The present invention detects abnormal values of DC electric energy meters in a DC distribution network through normal distribution, can obtain abnormal values of active power data and voltage data, and correct the abnormal values, thereby removing the limitation on the number of data collection and obtaining sufficient and complete DC electric energy meter data, thereby improving the accuracy and reliability of the DC power distribution electric energy meter measurement error estimation model in the DC distribution network; and, by introducing equivalent variables to perform linear replacement of the optimization target of the optimization model, the efficiency of the model solution can be improved.

[0011] Further, performing abnormal value detection on the first active energy data and the first voltage data according to a first normal distribution model of the first active energy data and a second normal distribution model of the first voltage data, respectively, includes:

[0012] A first normal distribution model is established based on the first active power data. If the first active power is outside a first preset value range of the first normal distribution model, the first active power is a first abnormal value; wherein the first active power data is obtained based on the difference between the forward active power and the reverse active power.

[0013] Furthermore, the abnormal value is corrected to obtain the second active power data and the second voltage data, respectively, including:

[0014] Obtain the middle value of the data before and after the first abnormal value of the first active power data to replace the first abnormal value, and obtain the middle value of the data before and after the second abnormal value of the first voltage data to replace the second abnormal value, and obtain the second active power data and the second voltage data respectively.

[0015] Furthermore, the introduction of equivalent variables performs linear substitution on the optimization objective of the optimization model to obtain a convex optimization model and solve it to obtain an estimated value of the online metering error, including:

[0016] Obtaining a first difference between the second active power data of the AC input side total meter and the power value of the loss model; wherein the loss model includes: a line loss model, a conversion loss model of the AC / DC converter, and a conversion loss model of the DC / DC converter before the DC power meter;

[0017] The optimization target is linearly replaced according to the first difference, and a linear programming solution is performed on the obtained convex optimization model to obtain an estimated value of the online measurement error.

[0018] The present invention uses equivalent variables to solve an optimization model that includes a line loss model in a DC distribution network, a conversion loss model of an AC / DC converter, and a conversion loss model of a DC / DC converter in front of a DC electric energy meter. Compared with traditional models, the present invention effectively reduces the impact of high loss proportions and large fluctuations on the accuracy of evaluation results by finely quantifying the losses of multiple typical links, thereby improving the accuracy of the electric energy meter measurement error estimation model in a DC distribution network and having higher applicability.

[0019] Furthermore, performing linear replacement on the optimization target according to the first difference includes:

[0020] A first equivalent variable identical to the first difference is introduced into the optimization model, a second difference between the first equivalent variable and the electricity value of the DC electric energy meter measurement error model is obtained, and a second equivalent variable identical to the absolute value of the second difference is introduced, so that the optimization objective is linearly replaced according to the second equivalent variable.

[0021] Furthermore, the collecting of the first clock after correction of the electric energy meter in the DC distribution network includes:

[0022] collecting a second clock corresponding to each electric energy meter in the DC distribution network, and determining in sequence whether a difference between the second clock and a third clock of the master station system is greater than a preset threshold;

[0023] If the difference between the second clock and the third clock of the master station system is greater than a preset threshold, clock correction is performed to obtain a corresponding corrected first clock.

[0024] Furthermore, before the first normal distribution model based on the first active energy data and the second normal distribution model based on the first voltage data, the method further includes:

[0025] Performing missing value detection on the first active energy data and the first voltage data respectively, and if values exist at adjacent moments before and after the missing value, filling the missing value according to a linear difference to obtain first active energy data and first voltage data without missing values;

[0026] If the adjacent moments before and after the missing value do not all have values, the first active power data and the first voltage data at that moment are all discarded.

[0027] Furthermore, the performing abnormal value detection on the first active energy data and the first voltage data according to the first normal distribution model of the first active energy data and the second normal distribution model of the first voltage data, respectively, further includes:

[0028] A second normal distribution model is established based on the first voltage data. If the voltage data is outside a second preset value range of the second normal distribution model, the voltage data is a second abnormal value.

[0029] Preferably, the introduction of equivalent variables to perform linear replacement on the optimization target of the optimization model can be expressed as:

[0030]

[0031] Among them, X t and Y t are the first equivalent variable and the second equivalent variable at time t in the optimization objective respectively; T a is the total number of time periods during which DC energy meter data can be collected; r 0j is the unit resistance of the jth DC trunk line, in Ω / km; L j is the length of the jth DC trunk line; η AC-DCj is the conversion efficiency of the jth AC / DC converter; η DC-DCk is the conversion efficiency of the DC / DC converter before the kth DC energy meter; N AC-DC is the total number of AC / DC converters; N DC-DC N is the total number of DC / DC converters before the DC energy meter; m is the total number of DC energy meters; W 0j,t is the second active energy data of the jth AC input side total meter at time t; U 0j,t is the second voltage data of the j-th AC input side total meter at time t; W i,t is the second active energy data of the i-th DC energy meter at time t; and They are the sum of the total electricity of the AC input side meter, the sum of the electricity values of the line loss model, the sum of the electricity values of the conversion loss model of the AC / DC converter, the sum of the electricity values of the conversion loss model of the DC / DC converter in front of the DC energy meter, and the sum of the electricity values of the DC energy meter measurement error model; ε i is the measurement error rate of DC electric energy meter i.

[0032] Preferably, the constraint of the second equivalent variable can be expressed as:

[0033]

[0034] Among them, α t is a 0-1 variable introduced at time t; M is a preset positive number. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 11 is a flow chart of a method for online estimation of measurement error of a DC electric energy meter based on data driving provided by an embodiment of the present invention;

[0036] Figure 2 1 is a schematic diagram of the topology of a DC distribution network simulation model provided by an embodiment of the present invention;

[0037] Figure 3 3 is a schematic diagram of a curve showing the average relative error of the online measurement error estimation values of the DC electric energy meters provided by the embodiment of the present invention;

[0038] Figure 4 It is a structural diagram of the data-driven online estimation system for DC electric energy meter measurement errors provided by the present invention. DETAILED DESCRIPTION

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0040] See also Figure 1 , is a flow chart of a data-driven online estimation method for DC electric energy meter measurement error provided by an embodiment of the present invention, including steps S11 to S13, specifically:

[0041] Step S11: collecting a first clock after correction of an electric energy meter in a DC distribution network, and collecting first active energy data of the same frequency within a preset time from an AC input side total meter and a DC electric energy meter in the DC distribution network under the first clock, as well as first voltage data from the AC input side total meter.

[0042] The method of collecting the corrected first clock of the electric energy meter in the DC distribution network includes collecting the second clock corresponding to each electric energy meter in the DC distribution network, and sequentially determining whether a difference between the second clock and a third clock of the master station system is greater than a preset threshold; if the difference between the second clock and the third clock of the master station system is greater than the preset threshold, performing clock correction to obtain a corrected first clock.

[0043] It is worth noting that the remote master station system sends clock acquisition commands to the AC input side master meter and each DC energy meter of the target DC distribution network one by one. After receiving the clock acquisition command, the DC distribution network collects the clock value of each energy meter. The clock value includes: year, month, day, hour, minute and second, a total of 6 dimensions of data. The clock value of the energy meter is recorded as C mi At the same time, record the clock value of the master system each time a command is issued, and record the clock value of the master system as Csi When comparing the clock value of each energy meter with the clock value of the master station system, if the difference between the clock value of the energy meter and the clock value of the master station system is not greater than the preset threshold, a clock correction command is issued to the energy meter so that after receiving the clock correction command, the energy meter changes the clock value of the energy meter to the clock value of the master station system; otherwise, there is no need to operate the clock value of the energy meter. That is: if |C mi ―C si |≤t0, then send a clock correction command to the energy meter i. After receiving the clock correction command, the energy meter i modifies its own clock value to C si Otherwise, no operation is required.

[0044] It is worth noting that the first active power data of the AC input side total meter and the DC power meter in the DC distribution network are collected at the same frequency within the preset time, as well as the first voltage data of the AC input side total meter; wherein the preset time is the time span of data collection. For example, such as: a preset time of 10 days from November 1, 2022 to November 10, 2022. The frequency is the time interval for recording the active power data of each power meter and the voltage data of the AC input side total meter, and the frequency includes: 1 minute, 5 minutes or 15 minutes. According to the frequency of recording power meter data, the number of time periods in a day that the power meter data can be collected can be calculated, and the total number of time periods that the power meter data can be collected within the set time can be calculated; wherein, the total number of time periods for data collection should be greater than the total number of power meters in the target distribution network.

[0045] Preferably, the number of collectable time periods and the total number of time periods can be expressed as:

[0046]

[0047] T a =d×T d ,

[0048] Wherein, Δt is the time interval for recording the active energy of the energy meter and the total voltage of the AC input side, in minutes; d is the time span of data collection, in days.

[0049] Step S12: performing abnormal value detection on the first active power data and the first voltage data according to the first normal distribution model of the first active power data and the second normal distribution model of the first voltage data, and correcting the abnormal values to obtain second active power data and second voltage data, respectively.

[0050] The collected first active energy data and first voltage data need to be preprocessed for missing data and abnormal values. Missing value detection is performed on the collected first active energy data and first voltage data, and the missing values are filled using linear interpolation to obtain first active energy data without missing values and first voltage data without missing values. Abnormal value detection is performed on the first active energy data and first voltage data without missing values, and abnormal values are corrected using linear interpolation to obtain second active energy data and second voltage data.

[0051] Specifically, before respectively forming the first normal distribution model of the first active energy data and the second normal distribution model of the first voltage data, the method further includes: performing missing value detection on the first active energy data and the first voltage data respectively; if the adjacent moments before and after the missing value have values, filling the missing value according to the linear difference to obtain the first active energy data and the first voltage data without missing values; if the adjacent moments before and after the missing value do not all have values, the first active energy data and the first voltage data at that moment are completely discarded.

[0052] It is worth noting that when the first active power data or the first voltage data at a certain collection moment is missing, if there is data at the adjacent moments before and after this moment, the linear interpolation method is used to fill the data gap; otherwise, the data of all electricity meters at this moment are discarded.

[0053] The method further includes: establishing a first normal distribution model based on the first active energy data and a second normal distribution model based on the first voltage data, performing outlier detection on the first active energy data and the first voltage data, respectively, including: establishing a first normal distribution model based on the first active energy data, and determining that the first active energy is a first outlier if the first active energy is outside a first preset value range of the first normal distribution model; wherein the first active energy data is obtained based on the difference between the forward active energy and the reverse active energy. Furthermore, the method further includes: establishing a second normal distribution model based on the first voltage data, and determining that the voltage data is a second outlier if the voltage is outside a second preset value range of the second normal distribution model.

[0054] Preferably, the 3-sigma criterion is used to perform abnormal value detection on the collected first active power data and the first voltage data respectively.

[0055] For example, when the 3-sigma criterion is used to detect abnormal values of the first active energy data, the normal distribution mean and standard deviation of the first active energy data can be expressed as follows:

[0056]

[0057]

[0058] Among them, W i,t is the first active power data of the energy meter i at time t, T a is the total number of time periods, are the mean and standard deviation of the curve corresponding to the first active energy data of electric energy meter i within the set time.

[0059] Exemplarily, the detection of an abnormal value of the first active energy data can be expressed as:

[0060]

[0061] Among them, W i,t is the first active energy data of the energy meter i at time t, are the mean and standard deviation of the curve corresponding to the first active energy data of electric energy meter i within the set time.

[0062] It is worth noting that, based on the mean and standard deviation of the curve corresponding to the AC input side total meter voltage data, the 3-sigma criterion can also be used to detect abnormal values in the first voltage data. When the first active energy data or the first voltage data is determined to be abnormal, the abnormal value is cleared and filled using a linear interpolation method.

[0063] It is worth noting that the first active energy data of the total meter on the AC input side of the DC distribution network and each DC energy meter at the same frequency within the set time consists of two parts: forward active energy and reverse active energy.

[0064] Preferably, the first active power data can be expressed as:

[0065]

[0066] Among them, W i,t is the active energy value of DC energy meter i during period t; are the forward and reverse active energy values of DC energy meter i in period t respectively.

[0067] Filling the abnormal values to obtain second active power data and second voltage data respectively, including: obtaining the middle value of the data before and after the first abnormal value of the first active power data to replace the first abnormal value, and obtaining the middle value of the data before and after the second abnormal value of the first voltage data to replace the second abnormal value, to obtain second active power data and second voltage data respectively.

[0068] Preferably, the linear interpolation of the first abnormal value of the first active energy and the linear interpolation of the second abnormal value of the first voltage data can be respectively expressed as:

[0069]

[0070]

[0071] Among them, W i,t is the second active power data after linear interpolation of the first abnormal value of the DC input side total meter i at time t, U 0j,t is the second voltage data after linear interpolation corresponding to the first abnormal value of the AC input side total table j at time t.

[0072] Step S13: Establish an optimization model for estimating the DC electric energy meter measurement error based on the second active energy data and the second voltage data, introduce equivalent variables to perform linear substitution on the optimization objective of the optimization model, obtain a convex optimization model, and solve it to obtain an online measurement error estimate, so that operation and maintenance are performed based on the online measurement error estimate.

[0073] The method includes introducing equivalent variables to linearly replace the optimization objective of the optimization model, obtaining a convex optimization model, solving the model, and obtaining an online metering error estimate, which includes: obtaining a first difference between the second active power data on the AC input side and the power value of the loss model; wherein the loss model includes: a line loss model, a conversion loss model of the AC / DC converter, and a conversion loss model of the DC / DC converter in front of the DC power meter; linearly replacing the optimization objective according to the first difference, performing linear programming solution on the obtained convex optimization model, and obtaining an online metering error estimate.

[0074] Preferably, the line loss model of the DC distribution network can be expressed as:

[0075]

[0076] Among them, r 0j is the unit resistance of the jth DC trunk line, in Ω / km; L j is the length of the jth DC trunk line.

[0077] Since the conversion loss of the AC / DC converter is linearly related to the AC side input power, preferably, the AC / DC converter loss model of the DC distribution network can be expressed as:

[0078] ΔW AC-DCj,t =W 0j,t ×(1―η AC-DCj ),

[0079] Among them, η AC-DCj is the conversion efficiency of the jth AC / DC converter.

[0080] DC / DC converters have conversion losses that are linearly proportional to the overall DC input power. In DC distribution networks, DC energy meters are configured differently for different load measurement scenarios, broadly categorized into two scenarios: 1) DC energy meters measure DC / DC converter losses, such as in distributed photovoltaic (PV) and energy storage; 2) DC energy meters do not measure DC / DC converter losses, including those in electric vehicle charging stations.

[0081] Preferably, in the case where the DC energy meter does not measure the loss value of the DC / DC converter, the loss model of the DC / DC converter before the DC energy meter can be expressed as:

[0082]

[0083] Among them, η DC-DCk is the conversion efficiency of the DC / DC converter before the kth DC energy meter. It should be noted that due to the existence of multiple voltage levels in the DC distribution network, some distributed power sources or power loads can be directly connected to the grid without the need for a DC / DC converter, so the corresponding η DC-DCi Take it as 1.

[0084] Based on the law of conservation of electric energy, according to the line loss model, the conversion loss model of the AC / DC converter and the conversion loss model of the DC / DC converter in front of the DC power meter, a multivariate linear equation system is established to consider the balance of supplied power, lost power and load power in each time period of the DC distribution network. Based on the multivariate linear equation system, an optimization model for estimating the measurement error of the DC power meter is constructed.

[0085] The present invention transforms the original model into a convex optimization model and adopts a linear programming method to solve it, which can effectively improve the efficiency of model solution and obtain the global optimal solution. At the same time, it removes the limitation of the traditional least squares method on the amount of electricity meter data collection, thereby improving the efficiency of model solution and improving the accuracy of the electricity meter measurement error estimation model in the DC distribution network by changing parameters through equivalent variables.

[0086] Preferably, the multivariate linear equations can be expressed as:

[0087]

[0088] Among them, ε i,t is the measurement error rate of DC energy meter i at time t, N AC-DC is the total number of AC / DC converters, N DC-DC is the total number of DC / DC converters before the DC energy meter; and They are the sum of the total electricity of the AC side input meter, the sum of the electricity values of the line loss model, the sum of the electricity values of the conversion loss model of the AC / DC converter, the sum of the electricity values of the conversion loss model of the DC / DC converter in front of the DC electricity meter, and the sum of the electricity values of the DC electricity meter measurement error model; r 0j is the unit resistance of the jth DC trunk line, in Ω / km; L j is the length of the jth DC trunk line; η AC-DCj is the conversion efficiency of the jth AC / DC converter; η DC-DCk is the conversion efficiency of the DC / DC converter before the kth DC energy meter; N m To obtain the total number of DC energy meters.

[0089] Preferably, the objective function of the optimization model for DC electric energy meter measurement error estimation can be expressed as:

[0090]

[0091] st-∞≤ε i ≤+∞,

[0092] Among them, T a is the total number of time periods during which DC energy meter data can be collected; ε i is the measurement error rate of DC energy meter i at time t, ranging from [-∞, +∞], N AC-DC is the total number of AC / DC converters, N DC-DC is the total number of DC / DC converters before the DC energy meter; and They are the sum of the total electricity of the AC input side meter, the sum of the electricity values of the line loss model, the sum of the electricity values of the conversion loss model of the AC / DC converter, the sum of the electricity values of the conversion loss model of the DC / DC converter in front of the DC energy meter, and the sum of the electricity values of the DC energy meter measurement error model; r 0j is the unit resistance of the jth DC trunk line, in Ω / km; L j is the length of the jth DC trunk line; η AC-DCj is the conversion efficiency of the jth AC / DC converter; η DC-DCk is the conversion efficiency of the DC / DC converter before the kth DC energy meter; N m To obtain the total number of DC energy meters.

[0093] The present invention uses equivalent variables to solve an optimization model that includes a line loss model in a DC distribution network, a conversion loss model of an AC / DC converter, and a conversion loss model of a DC / DC converter in front of a DC electric energy meter. Compared with traditional models, the present invention effectively reduces the impact of high loss proportions and large fluctuations on the accuracy of evaluation results by finely quantifying the losses of multiple typical links, thereby improving the accuracy of the electric energy meter measurement error estimation model in a DC distribution network and having higher applicability.

[0094] The optimization target is linearly replaced according to the first difference, including: introducing a first equivalent variable that is the same as the first difference into the optimization model, obtaining a second difference between the first equivalent variable and the electricity value of the DC electric energy meter measurement error model, and introducing a second equivalent variable that is the same as the absolute value of the second difference, so that the optimization target is linearly replaced according to the second equivalent variable.

[0095] Preferably, the introduction of equivalent variables to perform linear replacement on the optimization objective of the optimization model can be expressed as:

[0096]

[0097] Among them, X t and Y t are respectively the first equivalent variable and the second equivalent variable of the absolute value term at time t in the optimization objective.

[0098] Preferably, the constraint of the second equivalent variable can be expressed as:

[0099]

[0100] Among them, α t is a 0-1 variable introduced at time t; M is a preset positive number.

[0101] The present invention detects abnormal values of DC electric energy meters in a DC distribution network through normal distribution, can obtain abnormal values of active power data and voltage data, and fill in the abnormal values, can remove the limitation of the number of data collection, obtain sufficient and complete DC electric energy meter data, and thus improve the accuracy and reliability of the DC power distribution electric energy meter measurement error estimation model in the DC distribution network; and, introduces equivalent variables to perform linear replacement of the optimization target of the optimization model, which can improve the efficiency of model solution.

[0102] For example, to verify the effectiveness of the method of the present invention, a DC distribution network simulation model was built. Figure 2Figure 2 is a schematic diagram of the topology of a DC distribution network simulation model provided by an embodiment of the present invention. The DC distribution network comprises 20 load branches, 10 distributed photovoltaic branches, 10 electric vehicle charging load branches, and 5 energy storage device branches. A total of one AC input-side master meter and 45 DC energy meters are configured. For distributed photovoltaic and energy storage device branches, the DC energy meter measures the losses of the DC / DC converter; for low-power DC loads and electric vehicle charging loads, the DC energy meter does not measure the losses of the DC / DC converter. The DC trunk line resistance per unit length, r0, is 0.0754 Ω / km, and the length, L, is 200 m. A DC distribution network power flow simulation calculation is performed using the Matlab platform to obtain the injected active power on the AC input side, which is used as the active power curve for the AC input-side master meter. The active power reference curve is obtained by multiplying the active power of the AC input-side master meter and each DC energy meter by the time interval.

[0103] Preferably, the conversion efficiency η of the AC / DC converter is AC-DC is 0.95, the conversion efficiency η of the DC / DC converter before the DC energy meter DC-DC It is 0.92.

[0104] Among the 45 DC energy meters, the active energy baseline curves of DC energy meters 5, 25, and 35 were superimposed with a normally distributed random error with a mean of 0 and a standard deviation of 5% of the active energy baseline curve. The active energy baseline curves of the remaining DC energy meters were superimposed with a normally distributed random error with a mean of 0 and a standard deviation of 1% of the active energy baseline curve to form new active energy curves for each DC energy meter. Based on this curve, an optimization model for DC energy meter measurement error estimation was established, and the online measurement error estimates for each meter were obtained. The accuracy of the calculation results of the optimization model for DC energy meter measurement error estimation was evaluated based on the average relative error between the online measurement error estimates of the DC energy meters and the superimposed random error values.

[0105] Preferably, the average relative error can be expressed as:

[0106]

[0107] Among them, ε i , κ i are the estimated measurement error and superimposed random error value of the i-th DC electric energy meter respectively.

[0108] See also Figure 3, is a schematic diagram of a curve of the average relative error of the online metering error estimation values of each DC electric energy meter provided by an embodiment of the present invention. In the figure, the average relative error of 45 DC electric energy meters is in the range of [0,4%]. It can be seen that the data-driven online estimation method for DC electric energy meter metering error according to the present invention can improve the accuracy of the electric energy meter metering error estimation model in the DC distribution network.

[0109] See also Figure 4 , is a structural diagram of a data-driven online estimation system for DC electric energy meter measurement errors provided by the present invention, comprising: a data acquisition module 41, a preprocessing module 42 and an online estimation module 43.

[0110] It is worth noting that the data acquisition module 41 is mainly used to collect the first active power data and the first voltage data of each electric energy meter, and transmit the collected data to the preprocessing module 42; after obtaining the collected data, the preprocessing module 42 preprocesses the missing values and abnormal values to obtain the second active power data and the second voltage data, and transmits the second active power data and the second voltage data to the online estimation module 43; after receiving the second active power data and the second voltage data, the online estimation module 43 establishes an optimization model for the measurement error estimation of the DC electric energy meter, and solves the optimization model to obtain the online measurement error estimation value, so that operation and maintenance can be performed according to the online measurement error estimation value.

[0111] The data acquisition module 41 is used to collect the first clock of the calibrated electric energy meter in the DC distribution network, and under the first clock, collect the first active energy data of the AC input side total meter and the DC electric energy meter in the DC distribution network at the same frequency within a preset time, as well as the first voltage data of the AC input side total meter.

[0112] The method of collecting the corrected first clock of the electric energy meter in the DC distribution network includes collecting the second clock corresponding to each electric energy meter in the DC distribution network, and sequentially determining whether a difference between the second clock and a third clock of the master station system is greater than a preset threshold; if the difference between the second clock and the third clock of the master station system is greater than the preset threshold, performing clock correction to obtain a corrected first clock.

[0113] The preprocessing module 42 is used to perform outlier detection on the first active power data and the first voltage data according to the first normal distribution model of the first active power data and the second normal distribution model of the first voltage data, and correct the outliers to obtain second active power data and second voltage data, respectively.

[0114] The method further includes: establishing a first normal distribution model based on the first active energy data and a second normal distribution model based on the first voltage data, performing outlier detection on the first active energy data and the first voltage data, respectively, including: establishing a first normal distribution model based on the first active energy data, and determining that the first active energy is a first outlier if the first active energy is outside a first preset value range of the first normal distribution model; wherein the first active energy data is obtained based on the difference between the forward active energy and the reverse active energy. Furthermore, the method further includes: establishing a second normal distribution model based on the first voltage data, and determining that the voltage data is a second outlier if the voltage is outside a second preset value range of the second normal distribution model.

[0115] Filling the abnormal values to obtain second active power data and second voltage data respectively, including: obtaining the middle value of the data before and after the first abnormal value of the first active power data to replace the first abnormal value, and obtaining the middle value of the data before and after the second abnormal value of the first voltage data to replace the second abnormal value, to obtain second active power data and second voltage data respectively.

[0116] It is worth noting that before the first normal distribution model of the first active energy data and the second normal distribution model of the first voltage data are respectively adopted, it also includes: performing missing value detection on the first active energy data and the first voltage data respectively, and if the adjacent moments before and after the missing value have values, filling the missing value according to the linear difference to obtain the first active energy data and the first voltage data without missing values; if the adjacent moments before and after the missing value do not all have values, the first active energy data and the first voltage data at that moment are all discarded.

[0117] The online estimation module 43 is used to establish an optimization model for estimating the measurement error of the DC electric energy meter based on the second active energy data and the second voltage data, introduce equivalent variables to linearly replace the optimization objective of the optimization model, obtain a convex optimization model, and solve it to obtain an online measurement error estimate, so that operation and maintenance can be performed based on the online measurement error estimate.

[0118] The method includes introducing an equivalent variable to linearly replace the optimization objective of the optimization model, obtaining a convex optimization model, solving the model, and obtaining an online metering error estimate, which includes: taking a first difference between the second active power data of the AC input side main meter and the power value of the loss model; wherein the loss model includes a line loss model, a conversion loss model of the AC / DC converter, and a conversion loss model of the DC / DC converter in front of the DC power meter; linearly replacing the optimization objective according to the first difference, performing linear programming solution on the obtained convex optimization model, and obtaining an online metering error estimate.

[0119] The optimization target is linearly replaced according to the first difference, including: introducing a first equivalent variable that is the same as the first difference into the optimization model, obtaining a second difference between the first equivalent variable and the electricity value of the DC electric energy meter measurement error model, and introducing a second equivalent variable that is the same as the absolute value of the second difference, so that the optimization target is linearly replaced according to the second equivalent variable.

[0120] Preferably, equivalent variables are introduced to perform linear replacement on the optimization objective of the optimization model, which can be expressed as:

[0121]

[0122] Among them, X t and Y t are the first equivalent variable and the second equivalent variable at time t in the optimization objective respectively; T a is the total number of time periods during which DC energy meter data can be collected; r 0j is the unit resistance of the jth DC trunk line, in Ω / km; L j is the length of the jth DC trunk line; η AC-DCj is the conversion efficiency of the jth AC / DC converter; η DC-DCk is the conversion efficiency of the DC / DC converter before the kth DC energy meter; N AC-DC is the total number of AC / DC converters; N DC-DC N is the total number of DC / DC converters in the DC energy meter; m is the total number of DC energy meters; W 0j,t is the second active energy data of the jth AC input side total meter at time t; U 0j,t is the second voltage data of the j-th AC input side total meter at time t; W i,t is the second active energy data of the i-th DC energy meter at time t; and They are the sum of the total electricity of the AC input side meter, the sum of the electricity values of the line loss model, the sum of the electricity values of the conversion loss model of the AC / DC converter, the sum of the electricity values of the conversion loss model of the DC / DC converter in front of the DC energy meter, and the sum of the electricity values of the DC energy meter measurement error model; ε i is the measurement error rate of DC electric energy meter i.

[0123] Preferably, the constraint of the second equivalent variable can be expressed as:

[0124]

[0125] Among them, α t is a 0-1 variable introduced at time t; M is a preset positive number.

[0126] This paper proposes a DC power meter error estimation model for DC distribution networks that considers losses in various typical links. Compared to traditional models, this model effectively reduces the impact of high loss proportions and large fluctuations on the accuracy of estimation results by precisely quantifying the losses in each link. Furthermore, by introducing equivalent variables and transforming the original model into a convex optimization model, the linear programming method is employed to solve the model. This effectively improves the efficiency of the model solution, resulting in a globally optimal solution. This also removes the limitations imposed by traditional least squares methods on the amount of meter data collected.

[0127] The present invention also provides a computer terminal device, comprising: one or more processors; a memory coupled to the processor, for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the data-driven online estimation method for DC electric energy meter measurement errors.

[0128] The present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the data-driven online estimation method for DC electric energy meter measurement errors is implemented.

[0129] Those skilled in the art will appreciate that the embodiments of the present application may also provide computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0130] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0131] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0133] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A data-driven online estimation method for DC electric energy meter measurement error, characterized in that: include: Collecting a calibrated first clock of an electric energy meter in a DC distribution network, and collecting first active energy data of the same frequency within a preset time from an AC input side total meter and a DC electric energy meter in the DC distribution network under the first clock, as well as first voltage data from the AC input side total meter; performing outlier detection on the first active energy data and the first voltage data according to a first normal distribution model of the first active energy data and a second normal distribution model of the first voltage data, and correcting the outliers to obtain second active energy data and second voltage data, respectively; Establishing an optimization model for estimating a DC electric energy meter's measurement error based on the second active energy data and the second voltage data, introducing equivalent variables to perform linear substitution on an optimization objective of the optimization model to obtain a convex optimization model and solving the model to obtain an estimated online measurement error value, so that operation and maintenance are performed based on the estimated online measurement error value; The introduction of equivalent variables to perform linear replacement on the optimization objective of the optimization model can be expressed as: ， in, and The optimization objectives are The first equivalent variable and the second equivalent variable at the moment; The total number of time periods during which DC energy meter data can be collected; For the The unit resistance of the DC trunk line is Ω / km; For the The length of the DC main line; For the The conversion efficiency of each AC / DC converter; For the Conversion efficiency of the DC / DC converter before each DC energy meter; is the total number of AC / DC converters; is the total number of DC / DC converters before the DC energy meter; is the total number of DC energy meters; For the AC input side total meter The second active power data at the moment; For the AC input side total meter Second voltage data at the moment; For the DC energy meters The second active power data at the moment; 、 、 、 and They are the sum of the total electricity of the AC input side meter, the sum of the electricity values of the line loss model, the sum of the electricity values of the conversion loss model of the AC / DC converter, the sum of the electricity values of the conversion loss model of the DC / DC converter in front of the DC energy meter, and the sum of the electricity values of the DC energy meter measurement error model; DC energy meter The measurement error rate; The constraint of the second equivalent variable can be expressed as: ， in, for 0-1 variables introduced at the moment; A preset positive number.

2. The data-driven online estimation method for DC electric energy meter measurement error according to claim 1, characterized in that: The performing abnormal value detection on the first active energy data and the first voltage data according to a first normal distribution model of the first active energy data and a second normal distribution model of the first voltage data, respectively, includes: A first normal distribution model is established based on the first active power data. If the first active power is outside a first preset value range of the first normal distribution model, the first active power is a first abnormal value; wherein the first active power data is obtained based on the difference between the forward active power and the reverse active power.

3. The data-driven online estimation method for DC electric energy meter measurement error according to claim 1, characterized in that: The abnormal value is corrected to obtain the second active power data and the second voltage data, respectively, including: Obtain the middle value of the data before and after the first abnormal value of the first active power data to replace the first abnormal value, and obtain the middle value of the data before and after the second abnormal value of the first voltage data to replace the second abnormal value, and obtain the second active power data and the second voltage data respectively.

4. The data-driven online estimation method for DC electric energy meter measurement error according to claim 1, characterized in that: The introduction of equivalent variables to linearly replace the optimization target of the optimization model to obtain a convex optimization model and solve it to obtain an online measurement error estimate, including: Obtaining a first difference between the second active power data of the AC input side total meter and the power value of the loss model; wherein the loss model includes: a line loss model, a conversion loss model of the AC / DC converter, and a conversion loss model of the DC / DC converter before the DC power meter; The optimization target is linearly replaced according to the first difference, and a linear programming solution is performed on the obtained convex optimization model to obtain an estimated value of the online measurement error.

5. The data-driven online estimation method for DC electric energy meter measurement error according to claim 4, characterized in that: The linearly replacing the optimization target according to the first difference includes: A first equivalent variable identical to the first difference is introduced into the optimization model, a second difference between the first equivalent variable and the electricity value of the DC electric energy meter measurement error model is obtained, and a second equivalent variable identical to the absolute value of the second difference is introduced, so that the optimization objective is linearly replaced according to the second equivalent variable.

6. The data-driven online estimation method for DC electric energy meter measurement error according to claim 1, characterized in that: The collecting of the first clock after correction of the electric energy meter in the DC distribution network includes: collecting a second clock corresponding to each electric energy meter in the DC distribution network, and determining in sequence whether a difference between the second clock and a third clock of the master station system is greater than a preset threshold; If the difference between the second clock and the third clock of the master station system is greater than a preset threshold, clock correction is performed to obtain a corresponding corrected first clock.

7. The data-driven online estimation method for DC electric energy meter measurement error according to claim 1, characterized in that: Before the first normal distribution model based on the first active energy data and the second normal distribution model based on the first voltage data, the method further includes: Performing missing value detection on the first active energy data and the first voltage data respectively, and if values exist at adjacent moments before and after the missing value, filling the missing value according to a linear difference to obtain first active energy data and first voltage data without missing values; If the adjacent moments before and after the missing value do not all have values, the first active power data and the first voltage data at that moment are all discarded.

8. The data-driven online estimation method for DC electric energy meter measurement error according to claim 2, characterized in that: The method further includes performing abnormal value detection on the first active energy data and the first voltage data according to a first normal distribution model of the first active energy data and a second normal distribution model of the first voltage data, respectively: A second normal distribution model is established based on the first voltage data. If the voltage data is outside a second preset value range of the second normal distribution model, the voltage data is a second abnormal value.

Images