New energy power grid voltage stability analysis method based on measurement data

Abstract

The invention discloses a new energy power grid voltage stability analysis method based on measurement data, and belongs to the technical field of power system operation and maintenance, and the method comprises the following steps: constructing a voltage sensitivity estimation frame based on data driving, and collecting the measurement data of a new energy power grid in real time; carrying out abnormal value detection and elimination on the acquired measurement data by adopting an isolated forest, constructing a recursive least square voltage sensitivity calculation model with a forgetting factor by taking the measurement data subjected to abnormal value detection and elimination as input, and analyzing the voltage stability of the new energy power grid; according to the method, the precision of new energy power grid voltage stability evaluation indexes is improved, compared with a traditional method, the method is completely achieved based on measurement data calculation, the problem of curse of dimensionality generated by system modeling and a large matrix is avoided, new energy power grid operation state fluctuation caused by a distributed power supply can be tracked online, and the accuracy of the new energy power grid voltage stability evaluation indexes is improved. Therefore, the method has high practical application value.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system operation and maintenance, and specifically relates to a new energy power grid voltage stability analysis method based on measurement data. Background Art

[0002] With the widespread integration of renewable energy and the rapid development of intelligent power systems, operational issues such as voltage stability in renewable energy power grids are becoming increasingly prominent. Therefore, online assessment of the safe and stable operation of power systems has become a primary goal. Voltage sensitivity is a commonly used electrical indicator that provides a reliable basis for assessing system safety in the face of rapid grid voltage fluctuations. However, traditional voltage sensitivity analysis methods, which rely primarily on system parameter models and simulation tools, struggle to cope with the complexity brought about by the randomness of renewable energy output and load fluctuations. Furthermore, traditional analysis methods typically require precise system parameters and network topology information. When faced with large-scale power networks, they are limited by matrix dimensions, resulting in high computational complexity and difficulty in real-time application in actual operations. Summary of the Invention

[0003] Therefore, there is an urgent need for a voltage stability analysis method based on measurement data that can evaluate the voltage stability of the new energy power grid in real time without relying on the system model.

[0004] The technical solution of the present invention is as follows: a method for analyzing voltage stability of a new energy power grid based on measurement data, comprising the following steps:

[0005] S1 builds a data-driven voltage sensitivity estimation framework;

[0006] S2 collects measurement data of new energy power grid in real time;

[0007] S3 uses isolation forest to detect and remove outliers from the measurement data collected in S2;

[0008] S4 uses the measured data detected and eliminated by outliers in S3 as input to construct a voltage sensitivity calculation model of recursive least squares with a forgetting factor;

[0009] S5 analyzes the voltage stability of the new energy grid.

[0010] Furthermore, step S1 is specifically as follows:

[0011] S101 For an n-node power grid of arbitrary topology, the voltage-power sensitivity matrix is obtained by inverting the Jacobian matrix and continuously correcting it through the correction equation:

[0012]

[0013] Among them, J is the Jacobian matrix, J -1 is the voltage sensitivity matrix, θS P is the phase angle-active power sensitivity matrix, θS Q is the phase angle-reactive power sensitivity matrix, VS P is the active power-voltage sensitivity moment, VS Q is the reactive-voltage sensitivity matrix, Δθ is the phase deviation, ΔU is the voltage deviation, U is the node voltage, ΔP is the active deviation, ΔQ is the reactive deviation,

[0014] S102 performs Gaussian elimination on the S101 formula to obtain the relationship between voltage sensitivity and system voltage change and power change:

[0015] ΔU=VS P ΔP+VS Q ΔQ.

[0016] Furthermore, step S2 is specifically to collect measurement data such as voltage, current, active power, and reactive power in real time through intelligent measurement devices (such as phasor measurement units PMU) installed at each node of the distribution network. The measurement data includes the voltage amplitude, phase angle, active power, and reactive power of each node.

[0017] Furthermore, step S3 is specifically as follows:

[0018] S301 Consider the measurement data set Y={y1,y2,…,y n}, it contains N data points, each data point has Q = {q1,q2,…q d} attribute parameters,

[0019] S302 binary search tree constructs the dataset path length h(y):

[0020]

[0021] Where T is the number of trees, Y i is the i-th test sample, h t is the path length where the sample is isolated.

[0022] S303 and h(Y i ) is further rewritten in average form:

[0023] c(ψ)=2H(ψ-1)-2(ψ-1) / n

[0024] Where c(ψ) represents the normalized path length, n is the size of the dataset, and H(i) is the harmonic number;

[0025] S304 by normalized path length:

[0026]

[0027] Among them, E(h(Y i )) is the expected value of the path length, C(T) is a constant, and according to the set experience, if B(Y i ) is greater than the set value, then Y i will be marked as an outlier.

[0028] Furthermore, step S4 is specifically as follows:

[0029] S401 constructs E(h(Y i ))S's difference equation:

[0030]

[0031] Among them, ΔU i,t m , ΔP j,t m , ΔQ j,t m For measurement data, is the voltage active and reactive sensitivity matrix, m is the number of groups with small changes, n is the number of unknown quantities and m is much larger than n, e(t) is the error vector of the system,

[0032] S402 introduces the residual sum of squares function to obtain the optimal value function of formula S401:

[0033]

[0034] Where φ(t)=[U 1 (t),…,U n (t),X 1 (t),…,X n (t)], U(t) is the voltage variation matrix of the measured data, and X(t) is the active and reactive power variation matrix of the measured data;

[0035] S403 simplifies the S402 formula by introducing the covariance matrix P(t):

[0036]

[0037] in, are the t-th and t-1-th system parameter identification results;

[0038] S404 introduces the forgetting factor λ to improve the system's adaptability to sensitivity changes caused by operating state changes, and obtains the parameter identification function of the recursive least squares with the forgetting factor:

[0039]

[0040] Where k is the number of iterations, * represents the updated function with forgetting factor λ, I is the identity matrix, K k (t) is the gain matrix, P k (t), P k (t-1) is the t-th and t-1-th order covariance matrix, and the value range of λ is usually 0.95~1.

[0041] Furthermore, step S5 specifically includes forgetting and retaining the input measurement data through the forgetting factor, and repeatedly updating the calculated value of voltage sensitivity through the parameter identification function in S4, so as to determine the nodes and control variables that have the greatest impact on voltage stability, providing a basis for installing the compensation device.

[0042] The beneficial effects of the present invention are:

[0043] This paper first proposes a method for calculating the global voltage sensitivity of a new energy grid based on measured data. By introducing a forgetting factor, this method enables online acquisition of voltage sensitivity. Furthermore, by constructing an isolated forest outlier detection and elimination model, this method achieves more accurate voltage sensitivity calculations and improves the precision of voltage stability assessment indicators for new energy grids. Compared with traditional methods, this method is based entirely on measured data, avoiding the curse of dimensionality associated with system modeling and large matrices. Furthermore, it can achieve online tracking of fluctuations in the operating status of new energy grids caused by distributed power sources, thus possessing high practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a flowchart of the present invention.

[0045] Figure 2 This is a topological structure diagram in an example of the present invention.

[0046] Figure 3 Figure 2 is the active voltage sensitivity diagram based on the model.

[0047] Figure 4 Figure 2 is the reactive voltage sensitivity diagram based on the model.

[0048] Figure 5 This is a diagram of the active voltage sensitivity of the present invention.

[0049] Figure 6 This is a diagram of the reactive voltage sensitivity of the present invention.

[0050] Figure 7 This is an effect diagram of the online tracking of state fluctuations of the present invention. DETAILED DESCRIPTION

[0051] A method for analyzing voltage stability of a new energy power grid based on measurement data includes the following steps, which are performed in sequence:

[0052] Step 1: Build a data-driven voltage sensitivity estimation framework.

[0053] ① For an n-node power grid of arbitrary topology, the voltage-power sensitivity matrix is obtained by inverting the Jacobian matrix and continuously correcting the correction equation:

[0054]

[0055] Among them, J is the Jacobian matrix, J -1 is the voltage sensitivity matrix.

[0056] ②After performing Gaussian elimination on the above equation, the relationship between voltage sensitivity and system voltage change and power change is obtained:

[0057] ΔU=VS P ΔP+VS Q ΔQ

[0058] Step 2: Collect measurement data of the new energy power grid in real time.

[0059] Intelligent measurement devices (such as PMUs) installed at each node in the distribution network collect real-time measurement data such as voltage, current, active power, and reactive power. The measurement data includes information such as voltage amplitude, phase angle, active power, and reactive power at each node.

[0060] Step 3: Use isolation forest to detect and remove outliers from the measurement data collected in step 2.

[0061] ① Consider the measurement data set Y={y1,y2,…,y n}, it contains N data points. Each data point has Q={q1,q2,…q d} attribute parameters.

[0062] ② Binary search tree constructs the dataset path length h(y):

[0063]

[0064] Where T is the number of trees.

[0065] ③ and h(Y i ) is further rewritten in average form:

[0066] c(ψ)=2H(ψ-1)-2(ψ-1) / n

[0067] Where c(ψ) represents the normalized path length, n is the size of the dataset, and H(i) is the harmonic number.

[0068] ④ By standardized path length:

[0069]

[0070] Among them, E(h(Y i )) is the expected value of the path length, and C(T) is a constant. According to the set experience, if B(Y i ) is greater than the set value, then Y i will be marked as an outlier.

[0071] Step 4: Using the measured data detected and eliminated as outliers in step 3 as input, a voltage sensitivity calculation model of the Forgetting Factor Recursive Least Squares (FFRLS) is constructed.

[0072] ①Construct the difference equation of the sensitivity matrix S:

[0073]

[0074] Among them, ΔU i,t m , ΔP j,t m , ΔQ j,t m For measurement data, is the voltage active and reactive sensitivity matrix, m is the number of groups with small changes, n is the number of unknown quantities and m is much larger than n, and e(t) is the error vector of the system.

[0075] ②Introduce the residual square sum function to obtain the optimal value function of the above formula:

[0076]

[0077] Where φ(t)=[U 1 (t),…,U n (t),X 1 (t),…,X n (t)], U(t) is the voltage variation matrix of the measured data, and X(t) is the active and reactive power variation matrix of the measured data.

[0078] ③Simplify the above formula by introducing the covariance matrix P(t):

[0079]

[0080] in, are the t-th and t-1-th system parameter identification results.

[0081] ④ By introducing the forgetting factor λ, the system's adaptability to sensitivity changes caused by operating state changes is improved, and the parameter identification function of the recursive least squares with the forgetting factor is obtained:

[0082]

[0083] Where k is the number of iterations, * represents the updated function with forgetting factor λ, I is the identity matrix, K k (t) is the gain matrix, P k (t), P k (t-1) is the t-th and t-1-th order covariance matrix, and the value range of λ is usually 0.95~1.

[0084] The introduction of the forgetting factor λ enables the parameter identification results to update the parameter estimation at each time step without storing the entire historical data, thereby achieving efficient online parameter estimation that changes with the dynamic changes of the system, effectively improving the tracking ability and algorithm accuracy of the proposed method.

[0085] Step 5: Analyze the voltage stability of the new energy grid.

[0086] The input measurement data is forgotten and retained through the forgetting factor, and the calculated value of the voltage sensitivity is repeatedly updated through the parameter identification function in step 4, so as to determine the nodes and control variables that have the greatest impact on voltage stability, providing a basis for installing compensation devices.

[0087] The present invention constructs a recursive least squares calculation method for the global voltage sensitivity of a new energy power grid with a forgetting factor. By introducing the forgetting factor, the voltage sensitivity can be obtained online without storing the entire historical data. The method is completely based on the measurement data, avoiding the dimensionality curse problem caused by system modeling and large matrices, and realizing data-driven online calculation of voltage sensitivity.

[0088] Below in conjunction with embodiment, the present invention is described in further detail:

[0089] like Figure 2 The improved IEEE 33-node distribution network shown in the figure has node 1 as a balancing node and all other nodes are loaded. Household photovoltaic power sources are connected to nodes 5, 10, 15, 22, 25, and 30 to simulate the operational fluctuations caused by the addition of distributed power sources. Power fluctuations at different historical measurement times relative to the current measurement time are simulated. System voltage response data is obtained through the PMU, and voltage sensitivity, an indicator of voltage stability, is calculated. The calculation results and comparison are shown in [1]. Figure 3-6 ,The voltage sensitivity calculation errors before and after isolation forest processing are shown in Table 1.

[0090] Table 1 Voltage sensitivity perception error of the method of the present invention

[0091]

[0092] The error results further verified the effectiveness and feasibility of the present invention in power system voltage stability assessment.

[0093] When the system operating conditions fluctuate, the system voltage sensitivity will also change. Figure 7 From the online tracking of voltage sensitivity before and after system fluctuations, it can be seen that the method proposed in the present invention can accurately perceive the current distribution of active-voltage and reactive-voltage sensitivities.

[0094] The foregoing description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by any person skilled in the art within the technical scope disclosed herein and within the spirit and principles of the present invention shall be covered by the scope of protection of the present invention. Furthermore, any matters not described in detail in this specification constitute prior art known to those skilled in the art.

[0095] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

Claims

1. A method for analyzing voltage stability of a new energy power grid based on measurement data, characterized in that: The following steps are involved: S1 builds a data-driven voltage sensitivity estimation framework; S2 collects measurement data of new energy power grid in real time; S3 uses isolation forest to detect and remove outliers from the measurement data collected in S2; S4 uses the measured data detected and eliminated by outliers in S3 as input to construct a voltage sensitivity calculation model of recursive least squares with a forgetting factor; S5 analyzes the voltage stability of the new energy grid.

2. A method for analyzing voltage stability of a new energy power grid based on measurement data according to claim 1, characterized in that: Step S1 is specifically as follows: S101 For an n-node power grid of arbitrary topology, the voltage-power sensitivity matrix is obtained by inverting the Jacobian matrix: Among them, is the Jacobian matrix, is the voltage sensitivity matrix, θSP is the phase angle-active sensitivity matrix, θSQ is the phase angle-reactive sensitivity matrix, VSP is the active-voltage sensitivity moment, VSQ is the reactive-voltage sensitivity matrix, Δθ is the phase deviation, ΔU is the voltage deviation, U is the node voltage, ΔP is the active deviation, ΔQ is the reactive deviation, S102 performs Gaussian elimination on the S101 formula to obtain the relationship between voltage sensitivity and system voltage change and power change: ΔU=VS P ΔP+VS Q ΔQ。 3. The method for analyzing voltage stability of a new energy power grid based on measurement data according to claim 2 is characterized in that: Step S2 specifically involves collecting measurement data such as voltage, current, active power, and reactive power in real time through intelligent measurement devices installed at each node of the distribution network. The measurement data includes the voltage amplitude, phase angle, active power, and reactive power of each node.

4. The method for analyzing voltage stability of a new energy power grid based on measurement data according to claim 3 is characterized in that: Furthermore, step S3 is specifically as follows: S301 Consider the measurement data set Y={y1,y2,…,y n }, it contains N data points, each data point has Q = {q1,q2,…q d } attribute parameters, S302 binary search tree constructs the dataset path length h(y): Where T is the number of trees, Yi is the i-th test sample, and ht is the length of the isolated path of the sample. S303 and h(Y i ) is further rewritten in average form: c(ψ)=2H(ψ-1)-2(ψ-1) / n Where c(ψ) represents the normalized path length, n is the size of the dataset, and H(i) is the harmonic number; S304 by normalized path length: Among them, E(h(Y i )) is the expected value of the path length, and C(T) is a constant.

5. The method for analyzing voltage stability of a new energy power grid based on measurement data according to claim 4 is characterized in that: Step S4 is specifically as follows: S401 constructs the difference equation of the sensitivity matrix S: Among them, ΔU i,t m , ΔP j,t m , ΔQ j,t m For measurement data, is the voltage active and reactive sensitivity matrix, m is the number of groups with small changes, n is the number of unknown quantities and m>n, e(t) is the error vector of the system, S402 introduces the residual sum of squares function to obtain the optimal value function of formula S401: Where φ(t)=[U 1 (t),…,U n (t),X 1 (t),…,X n (t)], U(t) is the voltage variation matrix of the measured data, and X(t) is the active and reactive power variation matrix of the measured data; S403 simplifies the S402 formula by introducing the covariance matrix P(t): in, are the t-th and t-1-th system parameter identification results; S404 introduces the forgetting factor λ to improve the system's adaptability to sensitivity changes caused by operating state changes, and obtains the parameter identification function of the recursive least squares with the forgetting factor: Where k is the number of iterations, * represents the updated function with forgetting factor λ, I is the identity matrix, K k (t) is the gain matrix, P k (t), P k (t-1) is the t-th and t-1-th order covariance matrix, and the value range of λ is usually 0.95~1.

6. A method for analyzing voltage stability of a new energy power grid based on measurement data according to claim 5, characterized in that: Specifically, step S5 is to forget and retain the input measurement data through the forgetting factor, and repeatedly update the calculated value of voltage sensitivity through the parameter identification function in S4, so as to determine the nodes and control variables that have the greatest impact on voltage stability, providing a basis for installing the compensation device.

Images