Enhanced charging pile metering performance monitoring method and system considering accuracy and cost balance

Abstract

The invention discloses an enhanced charging pile metering performance monitoring method and system considering accuracy and cost balance. The method comprises the following steps: S101, performing preliminary estimation on a charging pile error coefficient by using an ensemble Kalman filter; s102, on the basis of the preliminary estimation result, a standard electric energy module configuration or manual verification scheme is optimized through a multi-objective optimization algorithm; and S103, according to the selected optimization scheme, taking a standard electric energy module output result or a manual verification result as a reference, and realizing enhanced estimation of the error coefficient of the charging pile. According to the invention, by introducing a standard electric energy module or a manual verification result as a reference, the accuracy and stability of an online monitoring algorithm are effectively improved. On the basis, the contradictory relation between accuracy improvement brought by introduction of the reference and extra cost is further considered, and optimal design of the reference configuration scheme is achieved. The method well solves the problem of tradeoff between online monitoring and reference configuration in practical application, and has high practical value.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charging pile calibration technology, and specifically to an enhanced charging pile metering performance monitoring method and system that balances accuracy and cost. Background Technology

[0002] Charging stations are a key infrastructure supporting the development of the electric vehicle industry. Based on differences in charging technology, mainstream charging facilities can currently be divided into two main types: AC charging stations and DC charging stations. AC charging stations have a simple structure but slower charging speeds; DC charging stations, on the other hand, significantly improve the user experience with their high efficiency and fast charging capabilities, but have higher initial investment and maintenance costs. With the continuous improvement of the intelligence and networking level of charging equipment, the coverage and operational efficiency of charging networks are constantly being optimized, laying a solid foundation for the large-scale application of electric vehicles. Therefore, accelerating the optimized layout and intelligent upgrading of charging infrastructure is of significant strategic importance and practical necessity for building a low-carbon, sustainable, and green transportation system.

[0003] During the charging process, various factors can affect the metering accuracy of charging piles, directly impacting the fair settlement between users and operators. According to the "Metrology Law of the People's Republic of China," charging piles, as instruments for electricity trade settlement, fall under the category of mandatory verification and require periodic metrological performance validation. Traditional on-site manual verification methods are costly and inefficient, making them unsuitable for the rapid expansion of charging pile scale. While using high-precision standard energy modules can effectively improve accuracy and reduce measurement errors, their high cost and maintenance burden during large-scale deployment limit their short-term feasibility. Against this backdrop, intelligent verification methods based on big data have emerged. Existing research mainly falls into two categories based on their principles: vehicle-pile interaction and energy conservation methods. The energy conservation method often uses regression methods to solve the established energy conservation model of the charging system, thereby obtaining the metering deviation. However, the optimality theory of this type of regression method strictly relies on the Gaussianity of the variables, which is often difficult to satisfy in practical applications. When the system exhibits significant nonlinear characteristics, or when key variables are non-Gaussian, such as skewed or multimodal distributions, the regression estimator will lose its optimality, leading to a decrease in the accuracy of the model estimation. In severe cases, it may even cause the estimation variance to diverge, rendering the calculation results invalid. Furthermore, the estimation effect of the energy conservation method is highly dependent on the quality of the input data, and its robustness has inherent deficiencies. Its mathematical model is essentially a balance equation, which is extremely sensitive to disturbances in the input data. Once outliers appear in the input data, they will disrupt the energy balance relationship, causing the model to become inaccurate, affecting the final estimator. This manifests as significant fluctuations and instability in the output results, seriously impacting the reliability of the econometric bias assessment.

[0004] For example:

[0005] Chinese invention patent CN119936723A, published on May 6, 2025, discloses a method, apparatus, device, and medium for analyzing the error characteristics of a charging pile group. The invention includes the following steps: constructing an initial set of state variables; calculating a set of estimated state variable values ​​for the charging pile group error coefficients at time k; calculating a set of estimated output variable values ​​at time k; calculating a set of corrected state variable values ​​at time k and their covariance; and outputting the corrected vector estimate of the charging pile group error coefficients at time k and its uncertainty. In practice, this invention relies solely on the energy readings from the main meter and sub-meters, resulting in an insufficiently comprehensive and detailed characterization of the system's internal operating state, thus presenting certain limitations. Especially when faced with outliers in the energy data, the system's state estimation process is easily affected, leading to significant deviations in the estimated error coefficients. This not only reduces the overall estimation accuracy but also affects the long-term reliability of the system and the credibility of the state assessment, demonstrating significant insufficient robustness.

[0006] Chinese invention patent CN119667332A, published on March 21, 2025, discloses a method and system for detecting metering errors in DC charging piles. The method includes the following steps: collecting the total AC power supply of multiple charging piles and the charging amount of their corresponding charging piles; collecting the AC power supply of the multiple charging piles and the charging power of the charging piles; obtaining the final conversion efficiency based on the AC power supply and the charging power; obtaining the conversion loss of the charging pile based on the final conversion efficiency and the total power supply; collecting the AC power supply of the charging pile in standby mode for multiple metering cycles the previous day; obtaining the standby loss of the charging pile based on the AC power supply; and obtaining the final metering error based on the total power supply, the charging amount, the conversion loss, and the standby loss, according to the dynamic energy conservation of the charging pile. This method essentially relies on a static, open-loop model framework, which cannot accurately capture the drift phenomenon of charging pile errors, lacks real-time feedback and dynamic correction mechanisms for the charging pile's operating status, and lacks the ability to continuously track and adaptively adjust dynamic changes in errors.

[0007] Chinese invention patent CN119199705A, published on December 27, 2024, discloses a method, system, device, and storage medium for the metering verification of DC charging piles. The method includes the following steps: collecting charging data from different models of electric vehicles on the same charging pile and preprocessing the collected charging data; collecting charging data from all electric vehicle meters and calculating the basic charging error of the charging pile based on the charging data from each electric vehicle's meter and the charging pile's charging data; classifying the charging pile charging data under different charging modes using an improved Gaussian mixture model; calculating the error contribution of each charging mode based on the charging pile charging data under each charging mode, and then calculating the final charging error of the charging pile under each charging mode based on the error contribution of each charging mode and the basic charging error of the charging pile. This method heavily relies on the accuracy of external electric vehicle meters as a "true value" reference. However, in practice, the performance of terminal metering devices varies, and their reliability is difficult to guarantee uniformly, potentially introducing third-party errors into the charging pile's error evaluation system and affecting the source reliability of the verification results. Secondly, this method focuses on pattern classification and error allocation of existing historical data, which is essentially a post-hoc statistical analysis and lacks the ability to estimate and calibrate the operating status of charging piles in real time. Summary of the Invention

[0008] The technical problem this invention aims to solve is to provide an enhanced charging pile metering performance monitoring method and system that balances accuracy and cost. Based on an energy conservation model, it achieves a preliminary estimation of the charging pile error coefficient. Furthermore, by using the output results of a standard power module or manual verification results as a reference, it improves the accuracy and stability of the algorithm. Building upon this, it further considers the contradictory relationship between the accuracy improvement brought by introducing a reference and the additional cost, achieving an optimized design of the reference configuration scheme. This method effectively solves the trade-off between online monitoring and reference configuration in practical applications, demonstrating strong practical value.

[0009] In a first aspect, the present invention provides an enhanced method for monitoring the metering performance of charging piles, taking into account a balance between accuracy and cost, comprising the following steps:

[0010] S101. Use an ensemble Kalman filter to make a preliminary estimate of the charging pile error coefficient;

[0011] S102. Based on the preliminary estimation results, optimize the standard power module configuration or manual verification scheme using a multi-objective optimization algorithm;

[0012] S103. Based on the selected optimization scheme, and with reference to the output results of the standard power module or the results of manual verification, the error coefficient of the charging pile is enhanced and estimated.

[0013] In a second aspect, the present invention provides an enhanced charging pile metering performance monitoring system that balances accuracy and cost, for implementing the method described in the first aspect.

[0014] The present invention has the following technical effects:

[0015] 1. This invention, based on the energy conservation model, estimates the error coefficients of charging piles. Furthermore, by using the output results of standard power modules or manual verification results as references, it improves the accuracy and stability of the algorithm. Building upon this, it further considers the trade-off between the accuracy improvement brought by introducing a reference and the additional cost, thus optimizing the reference configuration scheme. This method effectively solves the trade-off between online monitoring and reference configuration in practical applications, demonstrating strong practical value.

[0016] 2. This method replaces the manual on-site verification of each charging pile with a remote, large-scale verification approach. By installing standard power modules on some piles or conducting manual spot checks to obtain reference information, the accuracy of the algorithm can be calibrated and improved, thereby efficiently completing the metering performance monitoring of a massive number of charging piles.

[0017] 3. By combining software algorithms with a small number of high-precision standard power modules, the contradiction between accuracy and cost is effectively balanced while ensuring monitoring accuracy. While improving accuracy by about 15%, the cost is reduced by about 30% compared to traditional high-precision sensor solutions. This not only significantly reduces the cost of later maintenance, but also reduces the dependence on hardware.

[0018] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0019] To make the objectives, technical solutions, and advantages of the invention clearer, the invention will now be described in further detail with reference to the accompanying drawings, wherein:

[0020] Figure 1 This is a flowchart illustrating a specific implementation of an enhanced charging pile metering performance monitoring method that balances accuracy and cost, as disclosed in this invention.

[0021] Figure 2 This is a simplified model for establishing the law of conservation of energy in this invention;

[0022] Figure 3 This is a schematic diagram of the binary encoding of the configuration scheme in this invention;

[0023] Figure 4This is a schematic diagram of the system structure in this invention. Detailed Implementation

[0024] This invention provides an enhanced charging pile metering performance monitoring method and system that balances accuracy and cost. Based on an energy conservation model, it estimates the error coefficients of charging piles. Furthermore, by using the output results of a standard power module or manual verification results as a reference, it improves the accuracy and stability of the algorithm. Building upon this, it further considers the trade-off between the accuracy improvement brought by introducing a reference and the additional cost, achieving an optimized design of the reference configuration scheme. This method effectively solves the trade-off between online monitoring and reference configuration in practical applications, demonstrating strong practical value.

[0025] The overall approach of the technical solution in this application is as follows: a multi-level technical solution of "EnKF error coefficient estimation + NSGA-II multi-objective optimization" is adopted. First, EnKF is introduced to incorporate probabilistic characteristics into the state estimation process, enabling continuous and accurate estimation of the charging pile error coefficients. Second, balancing the accuracy and economy of the algorithm, the NSGA-II multi-objective optimization algorithm is used to optimize the reference configuration scheme under limited budget constraints. Finally, based on the selected scheme, the filter is modified to achieve enhanced estimation of the error coefficients, comprehensively improving the overall performance and engineering feasibility of charging pile error monitoring.

[0026] Before conducting metering error detection of charging piles, it is necessary to obtain and preprocess the data of the charging pile group. The data to be obtained includes: sampling time; total meter reading of the charging station; number of charging piles in the charging station and sub-meter readings of each charging pile; and calibration readings or standard power module readings of the charging piles.

[0027] The required preprocessing includes: removing data containing gross errors; calculating the electrical energy per unit time for the master table and each sub-table; interpolating or using models to predict at sampling points with missing data; appropriately correcting sampling points with abnormal values; standardizing the data, including units and types; and aligning various types of data according to time.

[0028] Example 1

[0029] like Figure 1 As shown, this embodiment provides an enhanced charging pile metering performance monitoring method that considers a balance between accuracy and cost, including the following steps:

[0030] S101. Use an ensemble Kalman filter to make a preliminary estimate of the charging pile error coefficient;

[0031] S102. Based on the preliminary estimation results, optimize the standard power module configuration or manual verification scheme using a multi-objective optimization algorithm;

[0032] S103. Based on the selected optimization scheme, and with reference to the output results of the standard power module or the results of manual verification, the error coefficient of the charging pile is enhanced and estimated.

[0033] Step S101 specifically includes the following process:

[0034] S11. For N DC charging piles in the charging station, determine the initial error coefficient based on the calibration results or charging data. Constitute the initial state variables The formula for defining the error coefficient α is as follows:

[0035]

[0036] In the formula, E m E represents the measured value of the output electrical energy, and E represents the actual value of the output electrical energy.

[0037] S12. Determine the covariance matrix P0 of the initial state variables based on prior knowledge, and construct a matrix with mean P0. A normal distribution with variance P0;

[0038] S13. Sample M initial state variables x from a normal distribution. 0.i The initial state variable set X0 = {x} constitutes the error coefficient of the charging pile 0,1 ... x 0,M}, where x 0,i =[α 0,i,1 ... α 0,i,N ] T (i = 1, 2, ..., M), α 0,i,j (j = 1, ..., N; i = 1, ..., M) represents the i-th initial state variable x. 0,i The error coefficient of the j-th charging pile; the sampling number M is the maximum value between twice N and 100. If N is large, such as tens or hundreds, M needs to be gradually increased to 1000 or more.

[0039] S14. Calculate the set of prior estimates of the state variables of the charging pile error coefficient at the current moment. The calculation formula is as follows:

[0040]

[0041] In the formula, X k-1 W is the set of state variable correction values ​​at time k-1; k-1 =[w k-1,1 ... w k-1,M [ ] is the process noise matrix at time k, w k-1,i (i = 1, ..., M) follow an N(0, Q) distribution, where Q is Wk-1 The covariance matrix; where, when k=1, the system performs its first iteration, based on the initial set of state variables X0 of the charging pile error coefficients, and the state variable set at the first moment. A priori estimation is performed, and thereafter, in each iteration, the set of corrected state variable values ​​X from the previous time step is used. k-1 ;

[0042] S15. Calculate the observation matrix at the current time. The calculation formula is as follows:

[0043]

[0044] In the formula, for The estimated energy consumption of the i-th total meter, e k,j (j=1,...,N) represents the measured output electrical energy of the j-th charging pile at time k. The set of prior estimates of state variables The i-th column vector The estimated error coefficient of pile j in the middle, v k,i (i=1,...,M) is the i-th observation noise column vector at time k, which follows an N(0,R) distribution, where R is the covariance matrix of the observation noise;

[0045] S16. Calculate the Kalman filter gain K at the current time step. The calculation formula is as follows:

[0046]

[0047] In the formula,

[0048] S17. Calculate the set of corrected state variable values ​​X at the current moment. k and the final estimate at the current time The calculation formulas are as follows:

[0049]

[0050] In the formula, x represents the final estimated error coefficient of the j-th charging pile at time k; k,i (i = 1, ..., M) is the set of state variable correction values ​​X at time k. k The i-th column vector in E; t,k The output electrical energy obtained from the total meter reading at time k; It is a 1×M all-one matrix.

[0051] Step S102 specifically includes the following process:

[0052] S21. Population initialization, which consists of encoding the standard power module configuration or manual verification scheme and establishing the objective function; firstly, for the N charging piles, the scheme (i.e., individuals in the population) is represented by binary encoding, with each scheme being a binary vector X of length N. q Each element in the vector corresponds to a specific standard power module configuration or manual verification status of a charging pile. The binary vector X... q The definition is as follows:

[0053] X q =[x q,1 ... x q,N ]

[0054] element x in the vector q,j (j=1,...,N) corresponds to the state of pile j under the current scheme. When pile j performs an operation (installation of standard power modules or manual verification), x q,j The value of x is "1"; when stub j does not perform any operation, x q,j The value is "0";

[0055] Next, for multiple objectives, a set of decision objective functions F is established. d ={f d,1 ,f d,2}, where f d,1 Let f be the cost objective function. d,2 The objective function is accuracy.

[0056] Then, set the population size Z. s Maximum number of iterations G n Crossover probability P j With the probability of mutation P y Tournament Scale T s Parameters such as [parameters] are used, and a diversified initialization strategy is adopted to divide the individuals in the initial population into multiple groups, each with a different configuration ratio, i.e., the number of standard power modules installed or the number of manual verifications varies in each scheme; for each individual in each group, in a binary vector X of length N... q In the middle, randomly select multiple elements according to the configuration ratio of the current group, set them to 1, and the rest to 0;

[0057] S22. The final estimated set of charging pile error coefficients at T time points, obtained by combining the Kalman filter estimates before and after correction. Calculate the set of objective functions F for each configuration scheme in the population. d The objective functions in the calculation are as follows:

[0058]

[0059] In the formula, ps The unit price is the price of a standard power module or the price of manual verification, where n is the number of standard power modules or the number of manual verifications in this configuration scheme. Let j be the enhanced estimate of pile j at time k;

[0060] S23. Based on the objective function value f d,1 with f d,2 First, Pareto optimization is used to optimize the population, that is, non-dominated sorting is used to rank the configuration schemes for individuals x in the population. p1 and x p2 The definitions of domination and non-domination are as follows:

[0061] x p1 Dominate x p2 If and only if the following two conditions are met simultaneously,

[0062] (1) For all objective functions f d,i (i = 1, 2, ...), x p1 Not even as good as x p2 Difference, that is:

[0063]

[0064] (2) There exists at least one objective function f d,j x p1 Strictly better than x p2 ,Right now:

[0065]

[0066] Non-dominant: If x p1 Dominate x p2 And x p2 Nor does it control x p1 If they are mutually exclusive, then they are said to be independent of each other; they represent different trade-off preferences and cannot be directly compared in terms of superiority or inferiority.

[0067] Based on the dominance and non-dominance relationship, non-dominant individuals are placed in the front layer and dominant individuals are placed in the back layer. Then, the dominance relationship is determined from the dominant individuals and the dominant individuals are placed in the back layer. This process continues until all individuals in the current layer are mutually non-dominant, thus forming multiple layers {F1 F2...}, where the individuals in F1 are the non-dominant individuals of the current population.

[0068] Secondly, for each individual in each layer, the normalized distances of the individual under each objective function are summed to obtain the overall crowding distance of the individual. The calculation steps are as follows:

[0069] (1) For each objective function f in the set of objective functions d,i Calculate the crowding distance d iFirst, under the current objective function, the individuals are sorted in ascending order. An infinite crowding distance is assigned to the first and last individuals after sorting, which represents the boundary of the objective function dimension. The range f is then calculated. e ,Right now:

[0070] f e =f max,obj -f min,obj

[0071] In the formula f max,obj f is the maximum value of an individual on the objective function. min,obj It is the minimum value;

[0072] (2) Then, for the intermediate individual x pi Calculate the difference f between its predecessor and successor individuals on the objective function. e ′, and calculate the normalized distance f. std :

[0073]

[0074] After calculating the normalized distance for all individuals, the objective function is changed until the normalized distance for all individuals under all objective functions has been calculated.

[0075] (3) Finally, the normalized distances of individuals under each objective function are summed to obtain the overall crowding distance D of the individual. N The calculation formula is:

[0076]

[0077] Finally, based on the aforementioned non-dominated relationships and overall crowding distance, an elite retention strategy is adopted to generate a new generation population. Specifically, for the different layers formed by the non-dominated ranking, starting from the leading layer F1, all individuals in that layer are preferentially included in the new population. If the total number of individuals is less than the size of the new population, the next layer is moved to the next layer and individuals from that layer are included; if the number of individuals is greater than the population size, individuals with larger crowding distances are preferentially included. Steps S22 to S23 are repeated until the maximum number of iterations is reached. The non-dominated individuals of the last generation are output as the optimal configuration set. The individual with the smallest mean normalized distance from the origin at each objective function value is selected as the optimal configuration, where the mean normalized distance is D′. n,i The calculation formula is as follows:

[0078]

[0079] In the formula, D′ n,i F is the mean normalized distance of the i-th solution among the non-dominated solutions. std,i,j F is the normalized distance of the j-th objective function of the i-th solution in the non-dominated solution. i,jLet F be the j-th objective function value of the i-th solution in the non-dominated solution. min,j F is the minimum value of the j-th objective function. std,i,j This represents the maximum value of the j-th objective function.

[0080] Step S103 specifically includes the following process:

[0081] S31. Select a scheme from the set of optimal configuration schemes to achieve enhanced estimation of the charging pile error coefficient; firstly, based on the selected scheme, calculate the actual error coefficient y of the corresponding charging pile using relevant parameters obtained from standard power modules or manual verification. sk,j And constitute the actual observation matrix E ck The formulas are as follows:

[0082]

[0083] In the formula, y sk,j (j=1,...,N) represents the actual error coefficient of pile j at time k, e sk,j e represents the electrical output value obtained at time k from the standard electrical module configured at pile j or through manual verification. k,j The measured output electrical energy value obtained from the sub-meter at pile j at time k; according to the selected scheme, when a pile has not been operated, the corresponding actual error coefficient observation value is 0;

[0084] S32. Next, calculate the current master table observation matrix. and error coefficient observation matrix The calculation formulas are as follows:

[0085]

[0086] In the formula, for The i-th vector, The set of prior estimates of the state variables at time k The i-th column vector in the vector, ⊙ represents the element-wise product (Hadamard product);

[0087] S33. Next, construct the corrected system observation matrix. The covariance matrix R of the measurement noise mtx The formula is as follows:

[0088]

[0089] R mtx =diag(R;R) s1 ⊙X q ;...;R sN ⊙X q )

[0090] In the formula, V k ′ is time k The measurement noise matrix, R sj (j=1,...,N) represents the standard deviation of the measurement noise of pile j;

[0091] S34. Then, the corrected filter gain K′ and the state variable correction value X k The calculation formulas for ′ are as follows:

[0092]

[0093]

[0094] In the formula, The observation matrix at time k The i-th column vector in;

[0095] S35. Finally, calculate the final estimated value of the error coefficient of the charging pile at the current moment. The calculation formula is as follows:

[0096]

[0097] In the formula, x′ represents the final estimated error coefficient of the j-th charging pile at time k. k,i (i = 1, ..., M) represents the state variable correction value X at time k. k The i-th column vector in '.

[0098] Example 2

[0099] Based on the same inventive concept, this application also provides a system corresponding to the method in Embodiment 1, as detailed in Embodiment 2.

[0100] like Figure 4 As shown, this embodiment provides an enhanced charging pile metering performance monitoring system that balances accuracy and cost, including:

[0101] An initial estimation system is used to initially estimate the error coefficients of charging piles. First, a filter model is established, then the filter parameters are initialized, and finally the error coefficients are continuously corrected based on the input. The parameter initialization ensures that the filter can be effectively started, providing stable iterative initial values ​​for subsequent dynamic correction, thereby accelerating the convergence process.

[0102] The system is optimized to improve the configuration scheme of the reference information. Based on the preliminary estimation results, the standard power module configuration or manual verification scheme is optimized through a multi-objective optimization algorithm to determine the optimal configuration scheme, providing a basis for correction for the subsequent enhanced estimation of error coefficients.

[0103] An enhanced estimation system is used to achieve enhanced estimation of error coefficients. By using the selected scheme, additional reference information for some charging piles is obtained, which is used to correct the equations for observation, filter gain and state variables, thereby achieving enhanced estimation of error coefficients.

[0104] Specifically, the initial estimation system is used to perform the following process:

[0105] S11. For N DC charging piles in the charging station, determine the initial error coefficient based on the calibration results or charging data. Constitute the initial state variables The formula for defining the error coefficient α is as follows:

[0106]

[0107] In the formula, E m E represents the measured value of the output electrical energy, and E represents the actual value of the output electrical energy.

[0108] S12. Determine the covariance matrix P0 of the initial state variables based on prior knowledge, and construct a matrix with mean P0. A normal distribution with variance P0;

[0109] S13. Sample M initial state variables x from a normal distribution. 0.i The initial state variable set X0 = {x} constitutes the error coefficient of the charging pile 0,1 ... x 0,M}, where x 0,i =[α 0,i,1 ... α 0,i,N ] T (i = 1, 2, ..., M), α 0,i,j (j = 1, ..., N; i = 1, ..., M) represents the i-th initial state variable x. 0,i The error coefficient of the j-th charging pile;

[0110] S14. Calculate the set of prior estimates of the state variables of the charging pile error coefficient at the current moment. The calculation formula is as follows:

[0111]

[0112] In the formula, X k-1 W is the set of state variable correction values ​​at time k-1; k-1 =[w k-1,1 ... w k-1,M [ ] is the process noise matrix at time k, w k-1,i (i = 1, ..., M) follow an N(0, Q) distribution, where Q is W k-1 The covariance matrix;

[0113] S15. Calculate the observation matrix at the current time. The calculation formula is as follows:

[0114]

[0115] In the formula, for The estimated energy consumption of the i-th total meter, e k,j (j=1,...,N) represents the measured output electrical energy of the j-th charging pile at time k. The set of prior estimates of state variables The i-th column vector The estimated error coefficient of pile j in the middle, v k,i (i=1,...,M) is the i-th observation noise column vector at time k, which follows an N(0,R) distribution, where R is the covariance matrix of the observation noise;

[0116] S16. Calculate the Kalman filter gain K at the current time step. The calculation formula is as follows:

[0117]

[0118] In the formula,

[0119] S17. Calculate the set of corrected state variable values ​​X at the current moment. k and the final estimate at the current time The calculation formulas are as follows:

[0120]

[0121] In the formula, x represents the final estimated error coefficient of the j-th charging pile at time k; k,i (i = 1, ..., M) is the set of state variable correction values ​​X at time k. k The i-th column vector in E; t,k The output electrical energy obtained from the total meter reading at time k; It is a 1×M all-one matrix.

[0122] The optimization system is specifically used to perform the following processes:

[0123] S21. Population initialization, which consists of encoding the standard power module configuration or manual verification scheme and establishing the objective function; firstly, for the N charging piles, the scheme (i.e., individuals in the population) is represented by binary encoding, with each scheme being a binary vector X of length N. qEach element in the vector corresponds to a specific standard power module configuration or manual verification status of a charging pile. The binary vector X... q The definition is as follows:

[0124] X q =[x q,1 ... x q,N ]

[0125] element x in the vector q,j (j = 1, ..., N) corresponds to the state of pile j in the current scheme;

[0126] Next, for multiple objectives, a set of decision objective functions F is established. d ={f d,1 ,f d,2}, where f d,1 Let f be the cost objective function. d,2 The objective function is accuracy.

[0127] S22. The final estimated set of charging pile error coefficients at T time points, obtained by combining the Kalman filter estimates before and after correction. Calculate the set of objective functions F for each configuration scheme in the population. d The objective functions in the calculation are as follows:

[0128] f d,1 =p s ×n

[0129]

[0130] In the formula, p s The unit price is the price of a standard power module or the price of manual verification, where n is the number of standard power modules or the number of manual verifications in this configuration scheme. Let j be the enhanced estimate of pile j at time k;

[0131] S23. Based on the objective function value f d,1 with f d,2 A multi-objective optimization algorithm is used to generate a set of configuration schemes and determine the optimal configuration scheme.

[0132] The enhanced estimation system is specifically used to perform the following process:

[0133] S31. Select a scheme from the set of optimal configuration schemes to achieve enhanced estimation of the charging pile error coefficient; firstly, based on the selected scheme, calculate the actual error coefficient y of the corresponding charging pile using relevant parameters obtained from standard power modules or manual verification. sk,j And constitute the actual observation matrix E ck The formulas are as follows:

[0134]

[0135] E ck =[E t,k y sk,1 ... y sk,N ] T

[0136] In the formula, y sk,j (j=1,...,N) represents the actual error coefficient of pile j at time k, e sk,j e represents the electrical output value obtained at time k from the standard electrical module configured at pile j or through manual verification. k,j The measured output electrical energy value obtained from the sub-meter at pile j at time k; according to the selected scheme, when a pile has not been operated, the corresponding actual error coefficient observation value is 0;

[0137] S32. Next, calculate the current master table observation matrix. and error coefficient observation matrix The calculation formulas are as follows:

[0138]

[0139] In the formula, for The i-th vector, The set of prior estimates of the state variables at time k The i-th column vector in the vector, ⊙ represents the element-wise product (Hadamard product);

[0140] S33. Next, construct the corrected system observation matrix. The covariance matrix R of the measurement noise mtx The formula is as follows:

[0141]

[0142] R mtx =diag(R;R) s1 ⊙X q ;...;R sN ⊙X q )

[0143] In the formula, V k ′ is time k The measurement noise matrix, R sj (j=1,...,N) represents the standard deviation of the measurement noise of pile j;

[0144] S34. Then, the corrected filter gain K′ and the state variable correction value X k The calculation formulas for ′ are as follows:

[0145]

[0146] In the formula, The observation matrix at time k The i-th column vector in;

[0147] S35. Finally, calculate the final estimated value of the error coefficient of the charging pile at the current moment. The calculation formula is as follows:

[0148]

[0149] In the formula x′ represents the final estimated error coefficient of the j-th charging pile at time k. k,i (i = 1, ..., M) represents the state variable correction value X at time k. k The i-th column vector in '.

[0150] Since the system described in Embodiment 2 of this invention is a system used to implement the method of Embodiment 1 of this invention, those skilled in the art can understand the specific structure and variations of this system based on the method described in Embodiment 1 of this invention, and therefore will not be repeated here. All systems used in the method of Embodiment 1 of this invention fall within the scope of protection of this invention.

[0151] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described with reference to preferred embodiments, those skilled in the art should understand that various changes in form and detail can be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. An enhanced method for monitoring the metering performance of charging piles, considering a balance between accuracy and cost, characterized in that, This method mainly includes the following steps: S101. Use an ensemble Kalman filter to make a preliminary estimate of the charging pile error coefficient; S102. Based on the preliminary estimation results, optimize the standard power module configuration or manual verification scheme using a multi-objective optimization algorithm. S103. Based on the selected optimization scheme, and with reference to the output results of the standard power module or the results of manual verification, the error coefficient of the charging pile is enhanced and estimated.

2. The enhanced charging pile metering performance monitoring method considering the balance between accuracy and cost as described in claim 1, characterized in that, Step S101 specifically includes the following process: S11. For N DC charging piles in the charging station, determine the initial error coefficient based on the calibration results or charging data. Constitute the initial state variables The formula for defining the error coefficient α is as follows: In the formula, E m E represents the measured value of the output electrical energy, and E represents the actual value of the output electrical energy. S12. Determine the covariance matrix P0 of the initial state variables based on prior knowledge, and construct a matrix with mean P0. A normal distribution with variance P0; S13. Sample M initial state variables x from a normal distribution. 0.i The initial state variable set X0 = {x} constitutes the error coefficient of the charging pile 0,1 ... x 0,M }, where x 0,i =α 0,i,1 ... α 0,i,N ] T (i = 1, 2, ..., M), α 0,i,j (j = 1, ..., N; i = 1, ..., M) represents the i-th initial state variable x. 0,i The error coefficient of the j-th charging pile; S14. Calculate the set of prior estimates of the state variables of the charging pile error coefficient at the current moment. The calculation formula is as follows: In the formula, X k-1 W is the set of state variable correction values ​​at time k-1; k-1 =w k-1,1 ... w k-1,M [ ] is the process noise matrix at time k, w k-1,i (i = 1, ..., M) follow an N(0, Q) distribution, where Q is W k-1 The covariance matrix; S15. Calculate the observation matrix at the current time. The calculation formula is as follows: In the formula, for The estimated energy consumption of the i-th total meter, e k,j (j=1,...,N) represents the measured output electrical energy of the j-th charging pile at time k. The set of prior estimates of state variables The i-th column vector The estimated error coefficient of pile j in the middle, v k,i (i=1,...,M) is the i-th observation noise column vector at time k, which follows an N(0,R) distribution, where R is the covariance matrix of the observation noise; S16. Calculate the Kalman filter gain K at the current time step. The calculation formula is as follows: In the formula, S17. Calculate the set of corrected state variable values ​​X at the current moment. k and the final estimate at the current time The calculation formulas are as follows: In the formula, x represents the final estimated error coefficient of the j-th charging pile at time k; k,i (i = 1, ..., M) is the set of state variable correction values ​​X at time k. k The i-th column vector in E; t,k The output electrical energy obtained from the total meter reading at time k; It is a 1×M all-one matrix.

3. The enhanced charging pile metering performance monitoring method considering the balance between accuracy and cost as described in claim 1, characterized in that, Step S102 specifically includes the following process: S21. Population initialization, which consists of encoding the standard power module configuration or manual verification scheme and establishing the objective function; firstly, for the N charging piles, the scheme (i.e., individuals in the population) is represented by binary encoding, with each scheme being a binary vector X of length N. q Each element in the vector corresponds to a specific standard power module configuration or manual verification status of a charging pile. The binary vector X... q The definition is as follows: X q =[x q,1 ... x q,N ] element x in the vector q,j (j = 1, ..., N) corresponds to the state of pile j in the current scheme; Next, for multiple objectives, a set of decision objective functions F is established. d ={f d,1 ,f d,2 }, where f d,1 Let f be the cost objective function. d,2 The objective function is accuracy. S22. The final estimated set of charging pile error coefficients at T time points, obtained by combining the Kalman filter estimates before and after correction. Calculate the set of objective functions F for each configuration scheme in the population. d The objective functions in the calculation are as follows: f d,1 =p s ×n In the formula, p s The unit price is the price of a standard power module or the price of manual verification, where n is the number of standard power modules or the number of manual verifications in this configuration scheme. Let j be the enhanced estimate of pile j at time k; S23. Based on the objective function value f d,1 with f d,2 A set of configuration schemes is generated using a multi-objective optimization algorithm, and the optimal configuration scheme is determined.

4. The enhanced charging pile metering performance monitoring method considering the balance between accuracy and cost as described in claim 1, characterized in that, Step S103 specifically includes the following process: S31. Select a scheme from the set of optimal configuration schemes to achieve enhanced estimation of the charging pile error coefficient; firstly, based on the selected scheme, calculate the actual error coefficient y of the corresponding charging pile using relevant parameters obtained from standard power modules or manual verification. sk,j And constitute the actual observation matrix E ck The formulas are as follows: AND ck =[E t,k and sk,1 ... and sk,N ] T In the formula, y sk,j (j=1,...,N) represents the actual error coefficient of pile j at time k, e sk,j e represents the electrical output value obtained at time k from the standard electrical module configured at pile j or through manual verification. k,j The measured output electrical energy value obtained from the sub-meter at pile j at time k; According to the selected scheme, when a certain pile is not operated on, the corresponding actual error coefficient observation value is 0; S32. Next, calculate the current master table observation matrix. and error coefficient observation matrix The calculation formulas are as follows: In the formula, for The i-th vector, The set of prior estimates of the state variables at time k The i-th column vector in the vector, ⊙ represents the element-wise product (Hadamard product); S33. Next, construct the corrected system observation matrix. The covariance matrix R of the measurement noise mtx The formula is as follows: R mtx =diag(R;R s1 ⊙X q ;...;R sN ⊙X q ) In the formula, V′ k For time k The measurement noise matrix, R sj (j=1,...,N) represents the standard deviation of the measurement noise of pile j; S34. Then, the corrected filter gain K′ and the state variable correction value X′ k The calculation formulas are as follows: In the formula, The observation matrix at time k The i-th column vector in; S35. Finally, calculate the final estimated value of the error coefficient of the charging pile at the current moment. The calculation formula is as follows: In the formula, x′ represents the final estimated error coefficient of the j-th charging pile at time k. k,i (i = 1, ..., M) represents the state variable correction value X′ at time k. k The i-th column vector in.

5. An enhanced charging pile metering performance monitoring system that balances accuracy and cost, characterized in that: Used to implement the method as described in any one of claims 1 to 4.

Images