A distributed power data acquisition method, medium and system

Abstract

This invention provides a distributed power data acquisition method, medium, and system, belonging to the field of distributed measurement technology of electrical variables. The method includes: First, establishing a topology diagram of the power system under test and acquiring real-time data such as voltage, current, frequency, and three-phase imbalance from power data acquisition terminals at each node. Then, establishing a set of data relationship equations between nodes based on the topology diagram and fitting these equations using historical data. Next, constructing a node correlation matrix and a node state estimation matrix, including indicators such as voltage, phase angle, frequency, and three-phase imbalance. Integrating these matrices into a weighted least squares estimation model to optimize the real-time acquired power data, obtaining corrected power data. Finally, updating the state estimation matrix based on the corrected power data and storing the distributed power acquisition results. This invention solves the problem of high uncertainty in the quality of power data acquired by existing technologies.

Description

Technical Field

[0001] This invention belongs to the field of distributed measurement technology of electrical variables, and specifically relates to a distributed power data acquisition method, medium and system. Background Technology

[0002] As a crucial component of modern societal infrastructure, the real-time monitoring and analysis of the power system's operational status is of paramount importance for ensuring power supply quality, improving system efficiency, and reducing losses. Traditional power monitoring technologies primarily rely on sampling data from a few key nodes, making it difficult to obtain real-time status information across the entire network and failing to meet the increasingly complex power system management and control requirements. With the development of smart grid technology, distributed power data acquisition and centralized monitoring have become essential means of perceiving the operational status of the power system.

[0003] Currently, power systems commonly employ a multi-node deployment of data acquisition terminals to collect real-time power data, including physical quantities such as voltage, current, frequency, and three-phase balance. This collected data provides a foundation for real-time monitoring and analysis of the power system's operating status. However, existing data acquisition and analysis methods still suffer from high uncertainty in the quality of the acquired power data: power acquisition terminals are affected by factors such as installation environment and measurement accuracy, resulting in certain systematic and random errors in the acquired data, leading to significant uncertainty in data quality. This directly impacts the reliability of subsequent state estimation and analysis results based on this data.

[0004] In summary, given the aforementioned problems in the current distributed data acquisition and centralized monitoring of power systems, there is an urgent need to research a new method for power data acquisition and state estimation to improve data quality and estimation accuracy, and provide strong support for real-time monitoring and efficient management of power systems. Summary of the Invention

[0005] In view of this, the present invention provides a distributed power data acquisition method, medium and system that can solve the problem of high uncertainty in the quality of power data acquired by existing technologies.

[0006] This invention is implemented as follows:

[0007] The first aspect of the present invention provides a distributed power data acquisition method, comprising: establishing a topology map of the power system under test; acquiring power data collected in real time by multiple power data acquisition terminals; establishing a set of data relationship equations between nodes based on the line topology map; fitting the set of data relationship equations to historical acquisition data; constructing a node correlation matrix; constructing a state estimation matrix for each node; constructing a weighted least squares estimation model for node power data; optimizing the real-time power data using the weighted least squares estimation model to obtain corrected power data; updating the state estimation matrix based on the corrected power data; and generating and storing distributed power acquisition results.

[0008] The nodes of the topology represent distribution transformers, busbars, and load connection points, including node number, node type, node location coordinates, and node voltage level; the edges of the topology represent transmission lines and distribution lines, including edge number, edge type, starting node number of the edge connection, edge impedance value, and edge admittance value.

[0009] The power data includes voltage, current, frequency, and three-phase balance; the multiple power data acquisition terminals are deployed on the feeder device of each node.

[0010] The data relationship equation set includes voltage relationship equation, current relationship equation, frequency relationship equation, and three-phase balance relationship equation.

[0011] In the node association matrix, rows represent starting nodes, columns represent ending nodes, and matrix elements represent data relationship equations between nodes.

[0012] The state estimation matrix includes a node voltage magnitude submatrix, a phase angle submatrix, a frequency submatrix, and a three-phase unbalance submatrix.

[0013] The weighted least squares estimation model for node power data is constructed based on the node association matrix and the state estimation matrix of each node.

[0014] The weighted least squares estimation model optimizes the real-time power data collected by the power data acquisition terminal to obtain corrected power data.

[0015] Specifically, the present invention includes the following steps:

[0016] S10. Establish the topology of the power system under test;

[0017] S20. Obtain real-time power data from multiple power data acquisition terminals deployed on the feeder device at each node, including voltage, current, frequency, and three-phase balance.

[0018] S30. Based on the aforementioned line topology, establish a set of data relationship equations between any two nodes, including voltage relationship equations, current relationship equations, frequency relationship equations, and three-phase balance relationship equations. The voltage relationship equations are based on Kirchhoff's voltage law, considering line impedance drop, capacitance effect, and dynamic variation characteristics. The current relationship equations are based on Kirchhoff's current law, including the shunt effect caused by admittance, ground capacitance effect, and dynamic regulation characteristics. The frequency relationship equations are based on power-frequency characteristics, considering first-order and second-order dynamic responses and static regulation characteristics. The three-phase imbalance relationship equations are based on the statistical characteristics of three-phase current deviations, including dynamic variations and limit constraints.

[0019] S40. Obtain historical data from the plurality of power data acquisition terminals, and use the historical data to fit the data relationship equation set;

[0020] S50. Construct a node association matrix based on the topology graph. The rows of the node association matrix represent the starting nodes, the columns represent the ending nodes, and the matrix elements represent the data relationship equations between the nodes.

[0021] S60. Construct a state estimation matrix for each node based on the historical data collected. The state estimation matrix includes a node voltage amplitude sub-matrix, a phase angle sub-matrix, a frequency sub-matrix, and a three-phase unbalance sub-matrix.

[0022] S70. Construct a weighted least squares estimation model for node power data based on the node association matrix and the state estimation matrix of each node;

[0023] S80. The weighted least squares estimation model is used to optimize the real-time power data collected by the power data acquisition terminal to obtain the corrected power data.

[0024] S90. Update the state estimation matrix based on the corrected power data, and generate distributed power acquisition results for storage.

[0025] Based on the above technical solution, the distributed power data acquisition method of the present invention can be further improved as follows:

[0026] The nodes of the topology represent distribution transformers, busbars, and load connection points, including node number, node type, node location coordinates, and node voltage level; the edges of the topology represent transmission lines and distribution lines, including edge number, edge type, starting node number of the edge connection, edge impedance value, and edge admittance value.

[0027] Furthermore, step S10 includes the following specific implementation details:

[0028] The node descriptions of the topology graph use the following data structure:

[0029] N i ={id i ,type i ,(x i ,y i ),v i};

[0030] In the formula, N i For node i, the set of attributes; id i For node number (integer); type i Node type (1 represents distribution transformer, 2 represents busbar, 3 represents load connection point); (x i ,y i ) represents the node's geographic coordinates (in meters); v i The rated voltage level (kV) for the node.

[0031] The edge description of the topological graph uses the following data structure:

[0032] E ij ={id ij ,type ij ,start i end j Z ij ,Y ij};

[0033] In the formula, E ij Let id be the set of attributes of edge ij; ij edge number (integer); type ij For edge types (1 represents transmission lines, 2 represents distribution lines); start i Number the starting node; end j Z is the termination node number; ij Y represents the line impedance (Ω). ij S is the line admittance.

[0034] The topology graph is constructed using an adjacency matrix:

[0035]

[0036] In the formula, T ij These are elements of the topological adjacency matrix.

[0037] Voltage data acquisition in step S20: U i,t =U i,real +ΔU i +η U,i ;

[0038] In the formula, U i,t U represents the measured voltage value (kV) at node i at time t;i,real The actual voltage value (kV); ΔU i The system error is (kV); η U,i For random error (following a normal distribution) ).

[0039] Current data acquisition: I i,t =I i,real +ΔI i +η I,i ;

[0040] In the formula, I i,t Let I be the measured current value (A) at node i at time t; i,real The actual current value (A); ΔI i Let η be the systematic error (A); I,i For random error (following a normal distribution) ).

[0041] Frequency data acquisition: f i,t =f i,real +Δf i +η f,i ;

[0042] In the formula, f i,t f is the measured frequency value (Hz) of node i at time t; i,real The actual frequency value (Hz); Δf i System error (Hz); η f,i For random error (following a normal distribution) ).

[0043] Three-phase balance data acquisition: ∈ i,t =∈ i,real +Δ∈ i +η ∈,i ;

[0044] In the formula, ∈ i,t Let be the measured three-phase unbalance value (%) at node i at time t; ∈ i,real This represents the actual three-phase unbalance value (%); Δ∈ i Systematic error (%); η ∈,i For random error (following a normal distribution) ).

[0045] Furthermore, the voltage relationship equation is as follows:

[0046] Variable meaning: U j U is the voltage value at node j (kV); i Z represents the voltage value at node i (kV); ij The line impedance (Ω) between nodes i and j; Iij X represents the current value (A) between nodes i and j; c For line capacitance reactance (Ω); k U The voltage dynamic adjustment coefficient (dimensionless, range [0.01-0.1]); The voltage change rate (kV / s); δ U is the voltage correction term (kV); j is the imaginary unit.

[0047] Current relationship equation:

[0048] Variable meaning: I ij U represents the current value (A) between nodes i and j; i U j Z represents the voltage values ​​(kV) at nodes i and j; ij Y is the line impedance (Ω) between nodes i and j; ij The line admittance (S) between nodes i and j; k I , where is the current dynamic adjustment coefficient (dimensionless, range [0.01-0.1]); B is the rate of change of current (A / s); c Line susceptance to ground (S); δ I This is the current correction term (A).

[0049] Frequency relationship equation:

[0050] Variable meaning: f j ,f i Here are the frequency values ​​(Hz) of nodes j and i; k f k is the first-order frequency adjustment coefficient (Hz / MW); f2 The second-order frequency adjustment coefficient (Hz·s / MW); The power change rate (MW / s); Acceleration due to power change (MW / s) 2 );α f P is the static frequency adjustment factor (Hz / MW); ij P represents the actual power transfer value (MW) between nodes i and j; ij,nom The rated power transmission value (MW); δ f This is the frequency correction term (Hz).

[0051] Three-phase unbalance equation:

[0052] Variable meaning: ∈ j ,∈ i Here are the three-phase unbalance values ​​(%) at nodes j and i; k ∈I is the imbalance coefficient (dimensionless, range [0.01-0.1]); ij,n Let n be the current value of the nth phase (A); The average value of the three-phase current (A); α ∈ The unbalance dynamic adjustment coefficient (dimensionless, range [0.01-0.1]); The rate of change of unbalance (%); β ∈ ∈ The limit adjustment coefficient (dimensionless, range [0.1-1]); i,lim The unbalance limit (%); δ ∈ This is the imbalance correction term (%).

[0053] Furthermore, in step S40, the historical data fitting employs the least squares method:

[0054]

[0055] In the formula, y i,t These are actual measured values; W is the fitted estimate; i,t is the weighting coefficient; T is the historical data time length; N is the total number of nodes.

[0056] In step S40, parameter optimization uses the gradient descent method:

[0057]

[0058] In the formula, θ k Here are the parameter values ​​for the k-th iteration; η is the learning rate (ranging from [0.001, 0.1]). The gradient of the objective function.

[0059] Furthermore, the complete expression of the node association matrix in step S50 is as follows:

[0060]

[0061] In the formula, F U ,F I ,F f ,F ∈ This is a function relating various physical quantities.

[0062] The specific form of the relational function in step S50:

[0063]

[0064]

[0065] Furthermore, the decomposition of the state estimation matrix in step S60:

[0066]

[0067] In the formula, σ i It is a singular value; w i ,v i E are singular vectors on the left and right; k Let r be the error matrix; r is the rank of the matrix.

[0068] The update rules for the submatrix in step S60 are as follows:

[0069] U k =U k-1 +ΔU k +α U (U k-1 -U k-2 );

[0070] θ k =θ k-1 +Δθ k +α θ (θ k-1 -θ k-2 );

[0071] f k =f k-1 +Δf k +α f (f k-1 -f k-2 );

[0072] ∈ k =∈ k-1 +Δ∈ k +α ∈ (∈ k-1 -∈ k-2 );

[0073] In the formula, α U ,α θ ,α f ,α ∈ denoted as the momentum factor for each state variable.

[0074] Furthermore, the objective function of the weighted least squares estimation model is:

[0075] Variable meanings: J(x) is the optimization objective function; w ij The weighting coefficients are dimensionless and range [0-1]; z ij This is the actual measured value; h ij (x) is a nonlinear function of the state variable and the measured value; m and n are the dimensions of the measurement data.

[0076] Iterative formula for state estimation:

[0077] Variable meaning: x k+1 ,x k H represents the state estimates for iterations k+1 and k; k W is a Jacobian matrix; k z is the weight matrix; k h(x) represents the measurement value after k iterations. k ) is a nonlinear function of the state variable and the measured value.

[0078] Furthermore, the complete expression of the data optimization processing in step S80 is as follows:

[0079]

[0080] In the formula, λ k The inertia coefficient (with a value range of [0,1]).

[0081] Update of the measurement error covariance matrix in step S80:

[0082] R k+1 =βR k +(1-β)(z k -h(x k ))(z k -h(x k )) T ;

[0083] In the formula, β is the forgetting factor (with a value range of [0.9,1]).

[0084] The power data optimization processing formula based on the weighted least squares estimation model in step S80 can be expressed as:

[0085] X opt =X mea +K wls (Z mea -HX mea );

[0086] Among them, K wls =(H T R -1 H+Q -1 ) -1 H T R -1 ;

[0087] It also includes the formula for optimizing the objective function:

[0088] The meanings of each variable are as follows:

[0089] X optThe optimized state variable vector includes state quantities such as voltage, current, frequency, and three-phase imbalance; X mea Z is the original state variable vector obtained from the measurement; mea H is the vector of actual measured values; H is the Jacobian matrix of the measurement equation, representing the sensitivity relationship between the measured values ​​and the state variables; K wls R is the weighted least squares gain matrix; R is the measurement noise covariance matrix, with diagonal elements representing the variances of each measurement, ranging from

[10] . -6 10 -2 Q is the prior state estimation covariance matrix, with diagonal elements representing the prior variances of each state variable, ranging from

[10] . -4 10 -1 ]; J obj To optimize the objective function, this is used to evaluate the accuracy of the optimization results.

[0090] Special function description: (·) T This represents the matrix transpose operation; (·) -1 This represents the matrix inversion operation.

[0091] Calculation process explanation:

[0092] 1. First, construct the gain matrix K. wls ;

[0093] 2. Using K wls The optimized state estimate X is obtained by weighting and correcting the measurement residuals. opt ;

[0094] 3. Through the objective function J obj Evaluate the optimization results, when J obj Less than a preset threshold (usually 10) -3 When the optimization is considered to be convergent, it is assumed that the optimization is successful.

[0095] Furthermore, the complete expression for the state matrix update in step S90 is as follows:

[0096]

[0097] In the formula, γ k This is a dynamic adjustment factor (with a value range of [0, 0.1]); The rate of change of the state variable.

[0098] Convergence criterion in step S90:

[0099] ||X k+1 -X k || F ≤∈ tol ;

[0100] In the formula, ||·||F Denotes the Frobenius norm; ∈ tol The convergence threshold (usually taken as 1e) -6 ).

[0101] The parameters and variables in each of the above steps can be calibrated and optimized using actual operating data. Specific calibration methods can include machine learning or statistical analysis. Simultaneously, various error terms and random disturbances need to be adjusted based on actual operating conditions to ensure the stability and reliability of the algorithm.

[0102] A second aspect of the present invention provides a computer-readable storage medium storing program instructions that, when executed in a computer, perform the aforementioned power data acquisition device and data acquisition method.

[0103] A third aspect of the present invention provides a distributed power data acquisition system, wherein the system includes the aforementioned computer-readable storage medium.

[0104] Compared with existing technologies, the beneficial effects of the distributed power data acquisition method, medium, and system provided by this invention are:

[0105] 1. Improved data quality. The method of this invention makes full use of the power system topology and physical relationships, establishes a detailed mathematical model, and effectively eliminates systematic and random errors in the collected data through techniques such as weighted least squares estimation, thereby improving the accuracy and reliability of the data.

[0106] 2. Enhanced Data Fusion Capabilities. The method of this invention constructs a node correlation matrix, reflecting the coupling relationships between nodes, and combines it with the state estimation matrix of each node to achieve systematic fusion of distributed acquisition data. This not only fully utilizes the complementarity of multi-source data but also better captures the operating status of the entire power system.

[0107] 3. Improved state estimation accuracy. Based on data fusion, the method of this invention further optimizes the system state using a weighted least squares estimation model, overcoming the problem of sensitivity to initial value selection in traditional methods, and significantly improving the accuracy and robustness of state estimation.

[0108] In summary, this invention solves the problem of high uncertainty in the quality of power data acquired by existing technologies. Attached Figure Description

[0109] Figure 1 A flowchart of the method provided by the present invention. Detailed Implementation

[0110] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0111] like Figure 1 The diagram shown is a flowchart of a distributed power data acquisition method provided by the first aspect of the present invention. This method includes the following steps:

[0112] S10. Establish the topology of the power system under test;

[0113] S20. Obtain real-time power data from multiple power data acquisition terminals deployed on the feeder device at each node, including voltage, current, frequency, and three-phase balance.

[0114] S30. Based on the line topology diagram, establish a set of data relationship equations between any two nodes, including voltage relationship equations, current relationship equations, frequency relationship equations, and three-phase balance relationship equations.

[0115] S40. Acquire historical data from multiple power data acquisition terminals and use the historical data to fit a set of data relationship equations;

[0116] S50. Construct a node association matrix based on the topology graph. The rows of the node association matrix represent the starting nodes, the columns represent the ending nodes, and the matrix elements represent the data relationship equations between the nodes.

[0117] S60. Construct a state estimation matrix for each node based on historical data. The state estimation matrix includes a node voltage amplitude submatrix, a phase angle submatrix, a frequency submatrix, and a three-phase unbalance submatrix.

[0118] S70. Construct a weighted least squares estimation model for node power data based on the node association matrix and the state estimation matrix of each node;

[0119] S80. The real-time power data collected by the power data acquisition terminal is optimized using a weighted least squares estimation model to obtain the corrected power data.

[0120] S90. Update the state estimation matrix based on the corrected power data, generate distributed power acquisition results, and store them.

[0121] The specific implementation methods of the above steps are described in detail below:

[0122] First, the specific implementation of step S10 is to establish a topology diagram of the power system under test. The node attributes in the topology diagram are represented using data structure N. i ={id i ,type i ,(x i,y i ),v i} represents, where id i For node number (integer), type i For node types (1 represents distribution transformer, 2 represents busbar, 3 represents load connection point), (x i ,y i ) represents the node's geographic coordinates (in meters), v i The node's rated voltage level (kV) is used. Edge attributes are represented using the data structure E. ij ={id ij ,type ij ,start i end j Z ij ,Y ij} represents, where id ij edge number (integer), type ij For edge types (1 represents transmission lines, 2 represents distribution lines), start i Number the starting node, end j Z is the termination node number. ij Y is the line impedance (Ω). ij The line admittance is (Siemens). The topology graph is constructed using an adjacency matrix T. ij This indicates that when node i and node j are directly connected, T ij =1, otherwise T ij =0. Establishing a topology diagram of the power system can provide necessary prerequisite information for subsequent data relationship modeling and state estimation.

[0123] Next, the specific implementation of step S20 involves acquiring real-time power data collected by multiple power data acquisition terminals deployed on the feeder device at each node. First, voltage data acquisition can be represented as:

[0124] U i,t =U i,real +ΔU i +η U,i ;

[0125] Among them, U i,t Let U be the measured voltage value (KV) of node i at time t. i,real The actual voltage value (KV), ΔU i Let η be the systematic error (Kv). U,i The error is random and follows a normal distribution. Secondly, current data acquisition can be represented as:

[0126] I i,t =I i,real +ΔIi +η I,i ;

[0127] Among them, I i,t Let I be the measured current value (A) at node i at time t. i,real The actual current value (A), ΔI i Let η be the systematic error (A). I,i The error is random and follows a normal distribution. Furthermore, frequency data acquisition can be represented as:

[0128] f i,t =f i,real +Δf i +η f,i ;

[0129] Where, f i,t Let f be the measured frequency value (Hz) of node i at time t. i,real Δf is the actual frequency value (Hz). i For the system error (Hz), η f,i The error is random and follows a normal distribution. Finally, the three-phase balance data acquisition can be expressed as:

[0130] ∈ i,t =∈ i,real +Δ∈ i +η ∈,i ;

[0131] Where, ∈ i,t Let be the measured three-phase unbalance value (%) at node i at time t, ∈ i,real The actual three-phase unbalance value (%), Δ∈ i For systematic error (%), η ∈,i The error is random and follows a normal distribution. The real-time acquisition of the above-mentioned power data provides the necessary foundation for subsequent data fusion and state estimation.

[0132] In step S30, a set of data relationship equations between any two nodes is established based on the line topology diagram. First, the complete expression of the voltage relationship equation is:

[0133]

[0134] Where, X c For line capacitance reactance (Ω), k U This is the voltage dynamic adjustment coefficient. The voltage change rate (Kv / s), δ U This represents the error term in the voltage relationship equation. Secondly, the complete expression of the current relationship equation is:

[0135]

[0136] Where, k I This is the current dynamic adjustment coefficient. B is the rate of change of current (A / s). c For line-to-ground susceptance (Siemens), δ I This represents the error term in the current relationship equation. Furthermore, the complete expression of the frequency relationship equation is:

[0137]

[0138] Where, k f2 This is the second-order frequency adjustment coefficient. Acceleration due to power change (MW / s) 2 ),α f P is the frequency static adjustment coefficient. ij,nom δ is the rated power transmission value (MW). f This represents the error term in the frequency relation equation. Finally, the complete expression of the three-phase unbalance relation equation is:

[0139]

[0140] Where, α ∈ This is the dynamic adjustment coefficient for the imbalance. β is the rate of change of unbalance (%) / s. ∈ The limit adjustment coefficient, ∈ i,lim The unbalance limit (%) is δ ∈ This represents the error term in the three-phase unbalance equations. These data equations provide a theoretical basis for subsequent state estimation.

[0141] The specific implementation of step S40 involves acquiring historical data from the multiple power data acquisition terminals and fitting the data relationship equations using the historical data. First, the historical data fitting employs the least squares method:

[0142]

[0143] Where, y i,t These are actual measured values. To fit the estimated value, W i,t Here, T represents the weighting coefficient, T is the duration of the historical data, and N is the total number of nodes. Secondly, parameter optimization employs gradient descent.

[0144]

[0145] Where, θ kLet be the parameter value for the k-th iteration, and η be the learning rate (ranging from [0.001, 0.1]). This represents the gradient of the objective function. By fitting historical data and optimizing the parameters, reliable mathematical model parameters can be obtained, providing a foundation for subsequent state estimation.

[0146] In step S50, a node association matrix is ​​constructed based on the topology graph. The complete expression of the node association matrix is ​​as follows:

[0147]

[0148] Among them, A ij F represents the elements of the topological adjacency matrix. U ,F I ,F f ,F ∈ These are the relationship functions for voltage, current, frequency, and three-phase unbalance, respectively. Specifically:

[0149]

[0150] The node association matrix constructed in this step provides the necessary input for subsequent weighted least squares state estimation.

[0151] The specific implementation of step S60 involves acquiring the historical data to construct a state estimation matrix for each node. The decomposition of the state estimation matrix can be expressed as:

[0152]

[0153] Where, σ i For singular values, w i ,v i These are the left and right singular vectors, E. k Let r be the error matrix and r be the rank of the matrix. The update rule for the state submatrix is ​​as follows:

[0154] U k =U k-1 +ΔU k +α U (U k-1 -U k-2 );

[0155] θ k =θ k-1 +Δθ k +α θ (θ k-1 -θ k-2 );

[0156] f k =f k-1 +Δf k +αf (f k-1 -f k-2 );

[0157] ∈ k =∈ k-1 +Δ∈ k +α ∈ (∈ k-1 -∈ k-2 );

[0158] Where, α U ,α θ ,α f ,α ∈ The momentum factors for each state variable are defined. This step constructs the state variables for each node, including voltage magnitude, phase angle, frequency, and three-phase imbalance, providing the necessary input for subsequent weighted least squares state estimation.

[0159] In step S70, a weighted least squares estimation model for node power data is constructed based on the node association matrix and the state estimation matrix of each node. The objective function of the weighted least squares estimation model is:

[0160]

[0161] Where, w ij z is the weighting coefficient. ij h is the measured value. ij (x) is a nonlinear function of the state variable and the measured value. The iterative formula for state estimation is:

[0162]

[0163] Among them, W k H is the weight matrix. k This is the Jacobian matrix. Through weighted least squares estimation, the nodal power data can be comprehensively optimized to obtain more accurate and reliable power state parameters.

[0164] The specific implementation of step S80 involves using the weighted least squares estimation model to optimize the real-time power data collected by the power data acquisition terminal, obtaining corrected power data. The complete expression of the data optimization process is as follows:

[0165]

[0166] Where, λ k This represents the inertia coefficient (with values ​​ranging from [0,1]). Meanwhile, the update rule for the measurement error covariance matrix is:

[0167] R k+1 =βR k +(1-β)(zk -h(x k ))(z k -h(x k )) T ;

[0168] Where β is the forgetting factor (with a value range of [0.9, 1]). This step uses the weighted least squares estimation results to correct and optimize the real-time acquired data to obtain more accurate and reliable power state quantities.

[0169] The formula for optimizing power data processing based on the weighted least squares estimation model can be expressed as:

[0170] X opt =X mea +K wls (Z mea -HX mea );

[0171] Among them, K wls =(H T R -1 H+Q -1 ) -1 H T R -1 ;

[0172] It also includes the formula for optimizing the objective function:

[0173] The meanings of each variable are as follows:

[0174] X opt The optimized state variable vector includes state quantities such as voltage, current, frequency, and three-phase imbalance; X mea Z is the original state variable vector obtained from the measurement; mea H is the vector of actual measured values; H is the Jacobian matrix of the measurement equation, representing the sensitivity relationship between the measured values ​​and the state variables; K wls R is the weighted least squares gain matrix; R is the measurement noise covariance matrix, with diagonal elements representing the variances of each measurement, ranging from

[10] . -6 10 -2 Q is the prior state estimation covariance matrix, with diagonal elements representing the prior variances of each state variable, ranging from

[10] . -4 10 -1 ]; J obj To optimize the objective function, this is used to evaluate the accuracy of the optimization results.

[0175] Special function description: (·) T This represents the matrix transpose operation; (·) -1 This represents the matrix inversion operation.

[0176] Calculation process explanation:

[0177] 1. First, construct the gain matrix K. wls ;

[0178] 2. Using K wls The optimized state estimate X is obtained by weighting and correcting the measurement residuals. opt ;

[0179] 3. Through the objective function J obj Evaluate the optimization results, when J obj Less than a preset threshold (usually 10) -3 When the optimization is considered to be convergent, it is assumed that the optimization is successful.

[0180] Finally, the specific implementation of step S90 involves updating the state estimation matrix based on the corrected power data, generating distributed power acquisition results, and storing them. The complete expression for the state matrix update is:

[0181]

[0182] Where, γ k This is a dynamic adjustment factor (with a value range of [0, 0.1]). Let be the rate of change of the state variable. The convergence criterion is:

[0183] ||X k+1 -X k || F ≤∈ tol ;

[0184] Among them, ||·|| F Denotes the Frobenius norm, ∈ tol The convergence threshold (usually taken as 1e) -6 By updating the state matrix and satisfying the convergence condition, the final distributed power acquisition result can be generated.

[0185] A second aspect of the present invention provides a computer-readable storage medium storing program instructions that, when executed in a computer, perform the aforementioned power data acquisition device and data acquisition method.

[0186] A third aspect of the present invention provides a distributed power data acquisition system, wherein the system includes the aforementioned computer-readable storage medium.

[0187] Specifically, the principle of this invention is to fully utilize the topology and physical relationships of the power system, establish a detailed mathematical model, and ultimately achieve accurate perception and effective control of the power system's operating status through advanced data fusion and state estimation techniques. The specific principle is as follows:

[0188] First, the method of this invention establishes a topology graph of the power system under test, including descriptions of node and edge attributes, providing basic information for subsequent data relationship modeling and state estimation. Node attributes include node number, node type, node coordinates, and node voltage level; edge attributes include edge number, edge type, starting node, ending node, line impedance, and line admittance. The topology graph is constructed using an adjacency matrix, which reflects the direct connections between nodes.

[0189] Secondly, the method of this invention acquires real-time physical quantity data such as voltage, current, frequency, and three-phase balance from power data acquisition terminals deployed at each node, and establishes mathematical relationship models between the terminal nodes. These relationship models include voltage relationship equations, current relationship equations, frequency relationship equations, and three-phase balance relationship equations, covering the main physical characteristics of the power system. These models not only consider the coupling relationship between nodes, but also introduce some dynamic adjustment coefficients and error terms to better describe the complex behavior of the actual system.

[0190] Based on this, the method of this invention acquires historical data and uses the least squares method to fit and optimize the parameters of the mathematical model to ensure the accuracy of the model's description of the actual system. Simultaneously, a node correlation matrix is ​​constructed to reflect the physical quantity relationships between nodes. Combining the state estimation matrix of each node, this invention further employs weighted least squares estimation technology to optimize the estimation of the operating state of the entire power system. This comprehensive estimation method based on the physical relationships of the entire network can effectively overcome the limitations of local data estimation and improve the accuracy and robustness of state estimation.

[0191] Finally, the method of this invention utilizes the optimized state estimation results to further correct and optimize the real-time collected power data, eliminating the influence of systematic and random errors and obtaining more reliable power state quantities. Based on this, the method of this invention periodically updates the state estimation matrix to generate the final distributed power acquisition results, providing support for real-time monitoring and intelligent management of power systems.

[0192] The following is an example of a specific application scenario of the present invention: A power grid company is responsible for monitoring and controlling the operating status of the power system within its jurisdiction. This power grid covers a wide area with uneven load distribution, making it difficult for traditional centralized data acquisition and state estimation methods to fully reflect the real-time operating status of the entire system. Therefore, the company decided to adopt the power data acquisition device and data acquisition method proposed in this invention to achieve distributed monitoring and centralized control of the entire power system.

[0193] First, the company's technical department conducted a detailed survey and modeling of the power system within its coverage area. Based on the survey results, the power grid topology is shown in Table 1, including three 500 kV transmission substations, ten 220 kV distribution substations, and 120 10 kV distribution line access points. i ={id i ,type i ,(x i ,y i ),v i In the context of node ID, ... i Numbered sequentially from 1 to 133, node type i These are 1 (500 kV transmission substation), 2 (220 kV distribution substation), and 3 (10 kV distribution line access point), with node geographic coordinates (x...). i ,y i The actual measured values ​​are used, and the node rated voltage level is v. i They are 500 kV, 220 kV and 10 kV respectively.

[0194] Table 1 Topology of a Power Grid

[0195] Node number Node type Geographic coordinates (x, y) Rated voltage level (kV) 1 1 (12.3,45.6) 500 2 1 (35.8,78.2) 500 3 1 (60.1,23.4) 500 4 2 (18.5,52.7) 220 5 2 (42.6,65.1) 220 … … … … 133 3 (28.4,41.9) 10

[0196] Meanwhile, the company's engineering department deployed multiple intelligent power data acquisition terminals at key nodes of the power grid to collect real-time data on physical quantities such as voltage, current, frequency, and three-phase balance. According to the requirements of the method of the present invention, the acquisition process of these acquisition terminals can be described as follows:

[0197] Voltage data acquisition: U i,t =Ui ,real +ΔU i +η U,i ;

[0198] Among them, U i,t Let U be the measured voltage value of node i at time t. i,real The actual voltage value, ΔU i For systematic error, η U,i To follow a normal distribution N(0, 0.1) 2Random error. Statistical analysis shows that the actual voltage value U at each node of the power grid... i,ral Distributed across three levels: [500±5%] kV, [220±5%] kV, and [10±5%] kV, with a systematic error ΔU i Follows N(0,1) 2 )distributed.

[0199] Current data acquisition: I i,t =I i,real +ΔI i +η I,i

[0200] Among them, I i,t Let I be the measured current value at node i at time t. i,real The actual current value, ΔI i For systematic error, η I,i To follow a normal distribution N(0,1) 2 Random error. Based on historical statistical data, the actual current value I at each node of this power grid... i,real The systematic error ΔI is distributed across three intervals: [1000±10%] Amperes, [500±10%] Amperes, and [100±10%] Amperes. i Obtained by N(0,5) 2 )distributed.

[0201] Frequency data acquisition: f i,t =f i,real +Δf i +η f,i ;

[0202] Where, f i,t f is the measured frequency value of node i at time t. i,real The actual frequency value, Δf i For systematic error, η f,i To follow a normal distribution N(0, 0.05) 2 The random error of the power grid. Through long-term monitoring, the actual frequency value f of the power grid i,real The system frequency remains basically within [50±0.1] Hz, with a systematic error Δf. i It follows the N(0,0.2) rule. 2 )distributed.

[0203] Three-phase balance data acquisition: ∈ i,t =∈ i,real +Δ∈ i +η ∈,i ;

[0204] Where, ∈ i,t Let be the measured three-phase unbalance value at node i at time t, ∈ i,real The actual three-phase unbalance value, Δ∈i For systematic error, η ∈,i To follow a normal distribution N(0,1) 2 The random error of ) is analyzed. The actual three-phase imbalance at each node of the power grid is ∈ i,real The systematic error Δ is basically within the range of [1±0.5]%. i Follows N(0,2) 2 )distributed.

[0205] Secondly, the company's R&D team established a mathematical relationship model between nodes based on the power grid topology and historical data.

[0206] Voltage relationship equation:

[0207] Where, X c The line capacitance is 0.1Ω. δ is the rate of change of voltage. U To conform to N(0,1) 2 Error term in the voltage relationship equation.

[0208] Current relationship equation:

[0209] Among them, B c The line-to-ground susceptance is 0.01 Siemens. δ is the rate of change of current. I To conform to N(0,5) 2 Error term in the current relationship equation.

[0210] Frequency relationship equation:

[0211] Where, P ij,nom The rated power transmission value is 100 megawatts. For the acceleration due to power change, δ f To conform to N(0,0.2) 2 Error term in the frequency relation equation.

[0212] Three-phase unbalance equation:

[0213] in The rate of change of the degree of imbalance, ∈ i,lim The unbalance limit is 3%, δ ∈ To conform to N(0,2) 2 Error term of the three-phase unbalance relation equation.

[0214] Based on this, the company's technical staff used the least squares method to fit and optimize the parameters of these relational models. After repeated iterations, the final model parameters are shown in Table 2.

[0215] Table 2 Parameters of the Mathematical Relationship Model

[0216] parameter optimal value <![CDATA[k U ]]> 0.01 <![CDATA[k I ]]> 0.05 <![CDATA[k f ]]> 0.01 <![CDATA[k f2 ]]> 0.001 <![CDATA[α f ]]> 0.1 <![CDATA[k ∈ ]]> 0.01 <![CDATA[α ∈ ]]> 0.05 <![CDATA[β ∈ ]]> 0.2

[0217] Meanwhile, the company's engineers constructed a node association matrix R based on the topology. ij The relational functions involved are as follows:

[0218]

[0219] In addition, the company's data center constructed a state estimation matrix X for each node based on historical data collection. k This includes the voltage amplitude submatrix, phase angle submatrix, frequency submatrix, and three-phase unbalance submatrix. For the voltage amplitude submatrix U... k Its update rules are as follows:

[0220] U k =U k-1 +ΔU k +0.9(U k-1 -U k-2 );

[0221] Where, ΔU k This represents the change in voltage amplitude during this update. Phase submatrix θ k Frequency submatrix f k and the three-phase imbalance submatrix ∈ k The update rules are similar.

[0222] With the above basic information, the company's control center used a weighted least squares estimation model to comprehensively optimize and estimate the operating state of the entire power system. Specifically, the objective function of the weighted least squares estimation model is:

[0223]

[0224] Where, w ij z is the weighting coefficient determined based on the degree of node association. ij h is the measured value. ij (x) is a nonlinear function of the state variable and the measured value. The iterative formula for state estimation is:

[0225]

[0226] Among them, W k H is a diagonal weight matrix. kThis is the Jacobian matrix. Through multiple iterative calculations, the state estimates of voltage amplitude, phase angle, frequency, and three-phase unbalance at each node of the power system were finally obtained.

[0227] Taking node 101 (a 220 kV distribution substation) as an example, its state estimates are shown in Table 3. It can be seen that the voltage amplitude of this node is approximately 219.5 kV, the phase angle is approximately -5.2 degrees, the frequency is approximately 50.02 Hz, and the three-phase imbalance is approximately 1.2%. The deviations between these state variables and the actual measured values ​​are small, indicating that the weighted least squares estimation method has achieved good results.

[0228] Table 3. State estimates of a certain node

[0229] state variables estimated value Voltage amplitude (kV) 219.5 Phase angle (degrees) -5.2 Frequency (Hz) 50.02 Three-phase imbalance (%) 1.2

[0230] Finally, the company's operations and maintenance personnel used the optimized state estimation results to further correct and optimize the real-time collected data. Taking node 101 as an example, its voltage data optimization process is as follows:

[0231]

[0232] Among them, G 101 and H 101 Let z be the gain matrix and Jacobian matrix for estimating the state of this node, R be the measurement error covariance matrix, and z be the value of z. 101,t The voltage value is collected in real time, h(x) 101 The voltage prediction value corresponds to the state estimate. Through this correction process, the influence of systematic and random errors can be further eliminated, resulting in a more reliable voltage state quantity.

[0233] Similarly, the company also optimized the current, frequency, and three-phase balance data. Based on this, the control center regularly updates the state estimation matrix of each node, generating distributed power acquisition results, providing strong support for the real-time monitoring and intelligent management of the power system.

[0234] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A distributed power data collection method, characterized by, The method comprises: establishing a topology graph of a power system to be measured; acquiring power data collected in real time by a plurality of power data collection terminals; establishing a data relationship equation set between any two nodes based on the topology graph, including a voltage relationship equation, a current relationship equation, a frequency relationship equation, and a three-phase unbalance degree relationship equation; the voltage relationship equation is based on Kirchhoff's voltage law and takes into account line impedance voltage drop, capacitance effect, and dynamic change characteristics; the current relationship equation is based on Kirchhoff's current law and contains shunt effect caused by admittance, ground capacitance effect, and dynamic adjustment characteristics; the frequency relationship equation is based on power-frequency characteristics and takes into account first-order and second-order dynamic responses and static adjustment characteristics; and the three-phase unbalance degree relationship equation is based on statistical characteristics of three-phase current deviation and contains dynamic changes and limit constraints; acquiring historical collection data of the plurality of power data collection terminals and fitting the data relationship equation set using the historical collection data; constructing a node correlation matrix based on the topology graph, wherein a row of the node correlation matrix represents a starting node, a column represents a terminal node, and a matrix element represents a data relationship equation between nodes; constructing a state estimation matrix of each node based on the historical collection data, wherein the state estimation matrix includes a node voltage amplitude sub-matrix, a phase angle sub-matrix, a frequency sub-matrix, and a three-phase unbalance degree sub-matrix; constructing a node power data weighted least squares estimation model based on the node correlation matrix and the state estimation matrix of each node; performing optimization processing on real-time power data using the weighted least squares estimation model to obtain corrected power data; updating the state estimation matrix based on the corrected power data and generating distributed power collection results for storage. wherein the voltage relationship equation is: ; wherein, is the voltage value of the j node; is the voltage value of the i node; is the line impedance between the i and j nodes; is the current value between the i and j nodes; is the line capacitance impedance; is the voltage dynamic adjustment coefficient; is the voltage change rate; is the voltage correction term; is the imaginary unit; Current relationship equation: ; wherein, is the current value between i, j nodes; is the voltage value of i, j nodes; is the line impedance between i, j nodes; is the line admittance between i, j nodes; is the current dynamic adjustment coefficient; is the current change rate; is the line-to-ground admittance; is the current correction term; Frequency relationship equation: ; wherein, is the j, i node frequency value; is the frequency first order regulation coefficient; is the frequency second order regulation coefficient; is the power rate of change; is the power rate of change acceleration; is the frequency static regulation coefficient; is the actual power transmission value between the i, j nodes; is the rated power transmission value; is the frequency correction term; Equation of relationship of three-phase unbalance degree: ; wherein, is the j, i node three-phase unbalance degree value; is the unbalance degree coefficient; is the n-th phase current value; is the three-phase current average value; is the unbalance degree dynamic adjustment coefficient; is the unbalance degree change rate; is the limit value adjustment coefficient; is the unbalance degree limit value; is the unbalance degree correction term; The association matrix is represented as: ; In the formula, is a topological adjacency matrix element, ; ; ; .

2. The distributed power data collection method of claim 1, wherein, The nodes of the topology graph represent distribution transformers, busbars, and load access points and include node numbers, node types, node position coordinates, and node voltage levels; and the edges of the topology graph represent transmission lines and distribution lines and include edge numbers, edge types, starting node numbers connected by edges, edge impedance values, and edge admittance values.

3. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions, which, when executed in a computer, are used to perform the power data collector device and data collection method of any one of claims 1-2.

4. A distributed power data collection system, characterized by, The computer-readable storage medium of claim 3 is included. The computer-readable storage medium of claim 3 is included.

Images