A remote estimation method for metering error of residential community charging station electric energy meter

Abstract

The application discloses a kind of residential area charging station electric energy metering error remote estimation method, including acquisition electric energy meter side, charging pile side and environmental data;Through improved type damping recursive least square method, electric energy metering error estimation is carried out;Input parameter is obtained, and initial value is determined, initial forgetting factor initial parameter estimation vector is set, and initial covariance matrix is calculated;The latest electric metering data are obtained to calculate t-time covariance matrix and gain matrix, and damping factor initial value is calculated;After each new measurement data and later, damping recursive least square method is used to update and solve t-time covariance matrix, gain matrix, predicted parameter and parameter error, and damping factor and forgetting factor are adjusted and corrected, and the estimated value of electric energy meter error is repeatedly recursively obtained;The problems that the prior art exists in the error evaluation of the electric energy metering of the charging station, such as slow operation speed, easy to be ill-conditioned and poor real-time performance, are solved.

Description

Technical Field

[0001] This invention belongs to the field of electricity metering technology, and in particular relates to a remote estimation method for the metering error of electricity meters in residential community charging stations. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the construction of residential charging infrastructure has entered a stage of rapid growth to meet the daily charging needs of residents. A large number of charging piles have been deployed in residential parking lots, underground garages, and public supporting areas. These residential charging stations are characterized by their large number, scattered distribution, small individual power, and fragmented operating time. In residential charging stations, the electricity meter, as the core metering device for trade settlement and energy consumption statistics, directly affects the fairness of electricity bill settlement, the revenue of the power grid company, and users' trust in charging services. Compared to traditional residential electricity loads, the electric vehicle charging loads at residential charging stations exhibit significant nonlinearity, impulsiveness, and high-order harmonic characteristics. This can easily lead to temperature drift and gain shift in metering chips and transformers. Therefore, a high-precision method for estimating electricity metering errors is needed that requires no power outages, no frequent on-site operations, can be performed remotely, and enables continuous monitoring, early warning, and scientific operation and maintenance of metering devices. However, existing methods for estimating electricity metering errors suffer from problems such as slow computation speed, ill-conditioned calculations, poor real-time performance, and difficulty in covering all meters due to the large volume of measurement data at charging stations and the existence of multicollinearity among the data. Summary of the Invention

[0003] The technical problem this invention aims to solve is to provide a remote estimation method for the metering error of electricity meters in residential charging stations, thereby addressing the shortcomings of existing technologies for error assessment of electricity meters in charging stations, such as slow calculation speed, susceptibility to ill-conditioned problems, poor real-time performance, and difficulty in covering all meters.

[0004] Technical solution of the present invention:

[0005] A remote estimation method for metering errors of electricity meters in residential charging stations, the method comprising:

[0006] Step 1: Collect data from the electricity meter, charging pile, and environment.

[0007] Step 2: Estimate the metering error of the electricity meter using the modified damped recursive least squares method (DRLS); specifically including:

[0008] Step 2.1: Obtain the input parameters, determine the initial values, and set the initial forgetting factor. 0 = 0.95, initial parameter estimation vector , Calculate the initial covariance matrix , , The initial scaling factor for the covariance matrix is ​​I, where I is the identity matrix.

[0009] Step 2.2: Obtain the latest electricity meter measurement data and calculate the covariance matrix at time t. and gain matrix And calculate the initial value of the damping factor. Each time a new set of measurement data is acquired and Then, the damped recursive least squares method is used to calculate the covariance matrix at time t. Gain matrix Prediction parameters and parameter error Perform an updated solution for the damping factor at time t. and forgetting factor The values ​​are adjusted and corrected, and the estimated value of the electricity meter error is obtained by repeated iterations.

[0010] The collected data from the electricity meter side include voltage U(t), current I(t), active power Pm(t), and cumulative electricity Em(t); the collected data from the charging pile side includes charging power Pc(t), charging status, and SOC; environmental data includes temperature and humidity.

[0011] The initial scaling factor of the covariance matrix is ​​between 10⁴ and 10⁶.

[0012] The formula for solving the damping recursive least squares problem at the t-th measurement time is as follows:

[0013] (1.12);

[0014] (1.13);

[0015] In the formula, Let be the covariance matrix at time t. Let be the parameter estimation vector at time t. (t-1) is the parameter estimation vector at time t-1. (t-2) is the parameter estimation vector at time t-2, and y0(t) is the actual value at time t; Forgetting factor, is the damping factor; I is the identity matrix.

[0016] Forgetting factor at time t The method for adjustment and correction is as follows:

[0017] (1.14);

[0018] In the formula, and These are the upper and lower limits of the forgetting factor, respectively; The activation function adjusts the value of the forgetting factor. This refers to the error between the meter's estimated value and the actual measured value. M is the control factor; M is the data window.

[0019] The upper limit of the forgetting factor is set at 0.97, and the lower limit is set at 0.9.

[0020] Damping factor at time t The adjustment and modification methods include:

[0021] First, obtain the latest measurement data for each sub-table. ,Depend on Determine the initial value of the damping factor. :

[0022] (1.15);

[0023] mean is the average of all elements in a matrix;

[0024] By minimizing the function The damping factor is adjusted according to the degree of linearization during the iteration process, that is:

[0025] (1.16).

[0026] when Increase ;when , Unchanged; when ,reduce Damping factor The increase or decrease of the value is achieved through the coefficient. Adjustments were made to increase the size. When γ decreases, / η;

[0027] (1.17);

[0028] The value is between 2 and 10.

[0029] The beneficial effects of this invention are:

[0030] This invention selects the damped recursive least squares method and improves and solves the damping factor and forgetting factor, thereby reducing the amount of computation, solving the technical problem of unstable solution, and effectively overcoming the problems of data saturation and collinearity. This invention can complete the online estimation of metering error without power outages or changes to existing electrical wiring, ensuring the continuity of residential charging services.

[0031] This technology solves the problems of slow calculation speed, ill-conditioned calculation, poor real-time performance, and difficulty in covering all meters in the error assessment of electricity meters in charging stations. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0033] A remote estimation method for metering errors of electricity meters in residential charging stations includes:

[0034] Step 1, Data Collection:

[0035] Charging station scale: Includes 20 7kW AC charging piles, each connected to 20 single-phase smart energy meters.

[0036] Data Acquisition: Both the electricity meter and the charging pile are connected to the concentrator via RS-485 bus and uploaded to the cloud metering management platform via 4G network, without the need for power outages or changes to existing electrical wiring.

[0037] Sampling period: The measurement data sampling period is 5 minutes, and it runs continuously for 30 days.

[0038] The collected data includes:

[0039] (1) Electricity meter side: voltage U(t), current I(t), active power Pm(t) and cumulative electricity Em(t).

[0040] (2) Charging pile side: charging power Pc(t), charging status and SOC.

[0041] (3) Environmental data: temperature, humidity (for auxiliary analysis).

[0042] The voltage and current at the electricity meter side are used to construct the measurement matrix. It reflects the real-time operating condition of the load, with active power and cumulative energy consumption forming a measurement vector. The charging pile side is the direct observation object for error estimation; the charging power, charging status, and SOC are used to judge the degree of load change and play an auxiliary role. Ambient temperature and humidity affect the temperature drift characteristics and insulation performance of the metering chip, shunt, and current transformer inside the electricity meter. By inputting temperature and humidity as auxiliary variables into the error model, additional metering deviations caused by environmental factors can be corrected, improving the long-term stability and generalization ability of remote estimation.

[0043] Step 2: Estimating measurement errors using the improved damped recursive least squares method (DRLS); including:

[0044] Step 2.1: Obtain the input parameters and determine the initial values. Set the initial forgetting factor weights. 0 = 0.95, initial identification parameter estimation vector , Calculate the initial covariance matrix , , Let be the initial scaling factor for the covariance matrix, and I be the identity matrix. This scaling factor is used to ensure the algorithm solves the problem without errors. A value between 104 and 106 is needed;

[0045] Step 2.2, Parameter Update Iteration. Obtain the latest meter measurement data and calculate the covariance matrix at time t. and gain matrix And calculate the initial value of the damping factor. As the amount of measurement data increases, each new set of measurement data... and Then, the damped recursive least squares method can be used to calculate the covariance matrix at time t. Gain matrix Prediction parameters Parameter error The solution is updated, and the damping factor at time t is updated in each solution. and the forgetting factor at time t The magnitude is adjusted and corrected. By repeating this process, an estimate that closely approximates the actual error of the electricity meter can be obtained.

[0046] The implementation steps of the improved damped recursive least squares method include:

[0047] Principle of Damped Recursive Least Squares Method:

[0048] Least squares is one of the most commonly used methods for parameter estimation. However, due to the large volume of measurement data from charging stations and the presence of multicollinearity, traditional least squares methods are slow, prone to ill-conditioned problems, and cannot provide satisfactory solutions. Recursive least squares can identify parameters online and reduce computation, but it is sensitive to data and prone to "data saturation," leading to data overload and unstable solutions as older data accumulates. Therefore, this invention selects damped recursive least squares and improves upon it by modifying the damping and forgetting factors. The principle of least squares is to minimize the sum of squared errors between observed and true values; therefore, a minimization function can be defined as follows:

[0049] (1.1)

[0050] In the formula, Forgetting factor, The damping factor, J(t) is the objective function at time t. Y(t) is the measurement vector at time t. A(t) is the measurement matrix at time t. Let be the estimated value of the electricity meter error at time t. This is the estimated error value of the electricity meter at time t-1.

[0051] Take the partial derivative of equation (1.1) to minimize its objective function:

[0052] (1.2)

[0053] Simplified,

[0054] (1.3)

[0055] In the formula, Forgetting factor, Let A(t) be the damping factor, I be the identity matrix, and A(t) be the damping factor. T This is the transpose of the measurement matrix at time t.

[0056] because It is reversible and has a unique solution.

[0057] (1.4)

[0058] make,

[0059] (1.5)

[0060] In the formula, P -1 (t-1) is the inverse correlation matrix at time t-1. This is the transpose of the measurement matrix at time t-1.

[0061] Then there is,

[0062] (1.6)

[0063] have to:

[0064] (1.7)

[0065] Simplifying the expression, we get:

[0066] (1.8)

[0067] After adjustment, the result is:

[0068] (1.9)

[0069] In the formula, Let be the inverse correlation matrix at time t. This is the transpose of the measurement matrix at time t-1. The measurement matrix at time t-1 Let a(t) be the transpose of the newly added measurement vector at time t, where a(t) is the newly added measurement vector at time t.

[0070] Rearrange equation (1.9) and add to both sides simultaneously. :

[0071] (1.10)

[0072] In the formula, The inverse correlation matrix at time t-1 It is the reciprocal of the forgetting factor.

[0073] Summarized as follows:

[0074] (1.11)

[0075] Simplifying equations (1.6)-(1.11), we obtain the following formula for solving the damping recursive least squares problem at the t-th measurement time:

[0076] (1.12)

[0077] (1.13)

[0078] In the formula, Let be the covariance matrix at time t. Let be the parameter estimation vector at time t. This is the parameter estimation vector at time t-1. This is the parameter estimation vector at time t-2. This is the actual value at time t.

[0079] Equations (1.12) and (1.13) are the damped recursive least squares formulas. It can be seen that each recursion only requires obtaining a single measurement data point from the meter at a specific moment, eliminating the need for matrix inversion and avoiding the influence of ill-conditioned matrices during the solution process. The forgetting factor and damping factor are fixed values. If these cannot be adjusted promptly when data changes significantly, it may lead to slow algorithm convergence, poor robustness, and suboptimal solutions. This invention improves the algorithm's recognition accuracy, tracking speed, and robustness by constructing suitable functions to guide the changes in the forgetting factor and damping factor.

[0080] Selection of variable forgetting factor:

[0081] In the recursive calculation process of DRLS, when a new set of measurement data is input, it is multiplied by a forgetting factor to avoid the covariance matrix. Rapid decay avoids data saturation and parameter explosion, improving the dynamic tracking capability of time-varying systems. Forgetting factor The smaller the value, the greater the impact of historical data on new data, focusing on overall model optimization, but resulting in slower real-time performance. A larger forgetting factor results in greater weight for new data, emphasizing the dynamic characteristics of the current model and improving convergence speed, but it is sensitive to parameter changes. When parameters change slowly, choosing a smaller forgetting factor can improve identification accuracy; when parameters change abruptly, choosing a larger forgetting factor can improve identification sensitivity.

[0082] In remote estimation of electricity meter errors, the tracking speed and identification accuracy of the recursive solution algorithm are key factors affecting the accuracy of the estimation results. Changes in the forgetting factor directly affect the accuracy of the model parameters; therefore, the appropriate selection of the forgetting factor is crucial to the solution performance of DRLS. This invention measures the accuracy of the parameters by calculating the error between the estimated values ​​and the measured values ​​obtained from the model parameters.

[0083] This invention dynamically adjusts the magnitude of the forgetting factor based on the model identification results, and dynamically adjusts the forgetting factor... Defined as:

[0084] (1.14)

[0085] In the formula, , These are the upper and lower limits of the forgetting factor, respectively; The activation function adjusts the size of the forgetting factor; This refers to the error between the meter's estimated value and the actual measured value. The control factor is used because the error obtained may vary under different charging stations and operating conditions. The forgetting factor is adjusted to adapt to different charging stations and operating conditions. M is the data window, which eliminates the influence of error distribution over a given time period. For meter error calculation, the forgetting factor is generally between 0.90 and 0.97, so its upper limit is set to 0.97 and its lower limit to 0.9.

[0086] Selection of variable damping factor:

[0087] Damping factor It is a positive number on the main diagonal of the measurement data matrix, and its magnitude indicates the relationship between the model input and output with respect to the minimization function. The importance of this lies in its ability to prevent divergence. If the system exhibits collinearity, a smaller [structure / mechanism] is crucial. It has a very good corrective effect.

[0088] First, obtain the latest measurement data for each sub-table. ,Depend on Determine the initial value of the damping factor. .

[0089] (1.15)

[0090] In the formula, is the initial value of the damping factor, which is generally taken as 0.01; mean is the mean of all elements in the matrix.

[0091] By minimizing the function The damping factor is adjusted according to the degree of linearization during the iteration process. That is:

[0092] (1.16)

[0093] when Increase ;when , Unchanged; when ,reduce .when When, it represents the objective function. The nonlinearity is relatively small in two adjacent iterations, and the damping factor is low. Near the optimal value, no adjustment is needed. The other two cases indicate a significant increase in the iteration error, requiring adjustment of the damping factor. To reduce errors. Damping factor The increase or decrease of the value can be achieved through the coefficient. Adjustments are generally made. Take a value between 2 and 10; as the value increases, When it decreases, .

[0094] (1.17).

[0095] Equation (1.17) quantifies the linearity of the algorithm's iterative process by using the ratio of the current change in the objective function to the predicted change. The magnitude of η reflects the curvature characteristics of the error surface at that point: increasing η when the curvature is large enhances stability, while decreasing η when the curvature is small accelerates convergence. This design enables the damping factor to adapt to different charging load conditions, avoiding oscillations and divergences.

[0096] Taking a charging station in an underground garage of a residential community in a certain city as an application scenario, the metering error of the electricity meter in the charging station is remotely estimated and monitored online.

[0097] 1. Implementation Scenarios and Environment

[0098] Charging station scale: Includes 20 7kW AC charging piles, each connected to 20 single-phase smart energy meters.

[0099] Data acquisition system: Both the electricity meter and the charging pile are connected to the concentrator via RS-485 bus and uploaded to the cloud metering management platform via 4G network.

[0100] Sampling period: The measurement data sampling period is 5 minutes, and it runs continuously for 30 days.

[0101] The collected data includes:

[0102] (1) Electricity meter side: voltage U(t), current I(t), active power Pm(t) and cumulative electricity Em(t)

[0103] (2) Charging pile side: charging power Pc(t), charging status, SOC

[0104] (3) Environmental data: temperature, humidity (for auxiliary analysis)

[0105] 2. Algorithm Initialization

[0106] The improved damped recursive least squares (DRLS) method proposed in this invention is used for measurement error estimation, with the following initialization parameters:

[0107]

[0108] 3. Error estimation process

[0109] (1) Construction of measurement data

[0110] Construct the measurement equation at each sampling time t:

[0111] Y(t) = A(t)·x(t) + e(t)

[0112] (2) Adjustment of variable forgetting factor

[0113] The forgetting factor is dynamically updated according to equation (1.14):

[0114] λ(t)=λ_max-(λ_max-λ_min)×tanh(ε×∑ⱼ₌0ᴹ||e(tj)||²)γ(t)

[0115] In the formula, λ(t) is the forgetting factor at time t, λmax is the maximum value of the forgetting factor, λmin is the minimum value of the forgetting factor, tanh(...) is the hyperbolic tangent function, and e(tj) is the residual at time tj.

[0116] (3) Adjustment of variable damping factor

[0117] Calculate the rate of change δ of the objective function according to equation (1.16):

[0118] δ<0.25: Increase the damping factor to enhance stability;

[0119] 0.25≤δ≤0.75: Maintain the current damping factor;

[0120] δ>0.75: Reduce the damping factor and improve the convergence speed.

[0121] (4) Parameter recursive update

[0122] The error estimate x(t) can be updated online using equations (1.12) and (1.13) without matrix inversion, making it suitable for large-scale measurement data environments.

[0123] 4. Implementation Results and Effects

[0124] After 30 consecutive days of remote estimation, the following results were obtained:

[0125]

[0126] #12 Electricity Meter: The error gradually increased from the initial -0.10% to +1.42%, exceeding the ±1.0% limit specified in GB / T17215;

[0127] The system triggered an alert when the error reached +1.2% and automatically generated a maintenance work order;

[0128] After the maintenance personnel replaced the electricity meter on-site, the error was restored to within ±0.2%, verifying the effectiveness of this method.

[0129] As can be seen from this embodiment, the method of the present invention has the following technical features:

[0130] (1) No power outage or meter removal is required to achieve online remote estimation of measurement error;

[0131] (2) By utilizing the variable forgetting factor and the variable damping factor, the problems of data saturation and collinearity can be effectively overcome;

[0132] (3) It can maintain good tracking performance under both sudden load changes and stable operating conditions;

[0133] (4) Realize the transformation from "periodic manual verification" to "real-time intelligent early warning", significantly reducing operation and maintenance costs.

Claims

1. A method for remotely estimating the metering error of electricity meters in residential charging stations, characterized in that: The method includes: Step 1: Collect data from the electricity meter, charging pile, and environment. Step 2: Estimate the metering error of the electricity meter using the modified damped recursive least squares method (DRLS); specifically including: Step 2.1: Obtain the input parameters, determine the initial values, and set the initial forgetting factor. 0 = 0.95, initial parameter estimation vector , Calculate the initial covariance matrix , , The initial scaling factor for the covariance matrix is ​​I, where I is the identity matrix. Step 2.2: Obtain the latest electricity meter measurement data and calculate the covariance matrix at time t. and gain matrix And calculate the initial value of the damping factor. Each time a new set of measurement data is acquired and Then, the damped recursive least squares method is used to calculate the covariance matrix at time t. Gain matrix Prediction parameters and parameter error Perform an updated solution for the damping factor at time t. and forgetting factor The values ​​are adjusted and corrected, and the estimated value of the electricity meter error is obtained by repeated iterations.

2. The method for remote estimation of metering error of electricity meter in a residential community charging station according to claim 1, characterized in that: The collected data from the electricity meter side include voltage U(t), current I(t), active power Pm(t), and cumulative electricity Em(t); the collected data from the charging pile side includes charging power Pc(t), charging status, and SOC; environmental data includes temperature and humidity.

3. The method for remote estimation of metering error of electricity meter in a residential community charging station according to claim 1, characterized in that: The initial scaling factor of the covariance matrix is ​​between 10⁴ and 10⁶.

4. The method for remote estimation of metering error of electricity meter in a residential community charging station according to claim 1, characterized in that: The formula for solving the damping recursive least squares problem at the t-th measurement time is as follows: (1.12)； (1.13)； In the formula, Let be the covariance matrix at time t. Let be the parameter estimation vector at time t. (t-1) is the parameter estimation vector at time t-1. (t-2) is the parameter estimation vector at time t-2, and y0(t) is the actual value at time t; Forgetting factor, It is the damping factor; I is the identity matrix.

5. The method for remote estimation of metering error of electricity meter in a residential community charging station according to claim 1, characterized in that: Forgetting factor at time t The method for adjustment and correction is as follows: (1.14)； In the formula, and These are the upper and lower limits of the forgetting factor, respectively; The activation function adjusts the value of the forgetting factor. This refers to the error between the meter's estimated value and the actual measured value. M is the control factor; M is the data window.

6. The method for remote estimation of metering error of electricity meter in a residential community charging station according to claim 1, characterized in that: The upper limit of the forgetting factor is set at 0.97, and the lower limit is set at 0.

9.

7. A method for remote estimation of metering error of electricity meters in residential community charging stations according to claim 1, characterized in that: Damping factor at time t The adjustment and modification methods include: First, obtain the latest measurement data for each sub-table. ,Depend on Determine the initial value of the damping factor. : (1.15)； mean is the average of all elements in a matrix; By minimizing the function The damping factor is adjusted according to the degree of linearization during the iteration process, that is: (1.16)。 8. A method for remotely estimating the metering error of an energy meter in a residential community charging station according to claim 7, characterized in that: when Increase ;when , Unchanged; when ,reduce ; Damping factor The increase or decrease of the value is achieved through the coefficient. Adjustments were made to increase the size. When γ decreases, / η; (1.17)； The value is between 2 and 10.

Images