Hybrid oscillation source positioning method and device, computer device and storage medium

Abstract

Embodiments of the present application provide a positioning method and device of a hybrid oscillation source, computer equipment and a storage medium, the method comprising: constructing a data set reflecting the high-dimensional linear space dynamics of a target power grid system; obtaining an equivalent linear model according to the data set, performing modal analysis on the equivalent linear model, and determining a potential oscillation mode; obtaining strongly correlated units according to participation factors of each unit under the potential oscillation mode, and determining dissipation energy flow and forward dissipation energy flow integral of the strongly correlated units according to active power, reactive power, voltage and frequency of each strongly correlated unit; determining whether the dissipation energy flow of a current unit is positive and the forward dissipation energy flow integral of the current unit exceeds a threshold value, if yes, identifying the current unit as an oscillation source, and if no, identifying the current unit as a non-oscillation source, which can quickly and accurately locate a single or multiple hybrid oscillation source causing oscillation of the power grid system, and improve stability and security of the power grid system.

Description

Technical Field

[0001] This application relates to the field of power grid system technology, and more specifically, to a method, apparatus, computer equipment, and storage medium for locating a hybrid oscillation source. Background Technology

[0002] The large-scale integration of renewable energy sources (such as wind and solar power) and the widespread application of various power electronic devices (such as converters, flexible AC transmission systems, high-voltage direct current transmission, and electric vehicle charging facilities) have jointly contributed to the trend of modern power grid systems evolving towards "high-voltage and high-efficiency" characteristics. This profound structural transformation has fundamentally impacted the dynamic behavior of power grid systems, making low-damped oscillations and forced oscillations more prevalent than ever before, thus making the online identification and location of oscillation sources more important and complex than ever before.

[0003] Against the backdrop of high proportions of renewable energy and high proportions of power electronic equipment, the power grid system faces the problem of significantly reduced inertia and damping due to the large-scale replacement of synchronous machines. This leads to more complex system dynamic characteristics, and a significant increase in the frequency and risk of low-damped oscillations, new broadband oscillations (such as subsynchronous and supersynchronous oscillations), and forced oscillations. Simultaneously, a large number of distributed renewable energy generation units with varying parameters and power electronic loads become potential and hidden sources of oscillations. Their rapid dynamics and strong coupling with traditional electromechanical dynamics make the oscillation propagation paths intricate and complex. This situation makes it difficult to quickly and accurately pinpoint the oscillation source using only traditional experience or models. Therefore, developing online oscillation source localization technology based on wide-area measurement data is crucial. It is a key prerequisite and core challenge for rapidly implementing targeted emergency control (such as adjusting controllers and cutting off disturbance sources) to quell oscillations and ensure the safe and stable operation of the "dual-high" system.

[0004] In related technologies, the instantaneous voltage and current values ​​at the ports of each component in the target system are collected and transformed from a three-phase stationary coordinate system to a two-phase rotating coordinate system. Then, the collected data is substituted into the derived energy flow formula to obtain the transient energy flow of each component at different frequencies. By linearly fitting the energy flow versus time curve, the slope of the fit represents the energy flow power, which can be used to determine the direction and magnitude of the energy flow. Furthermore, the dominant oscillation frequency of the system is determined by finding the frequency with the largest absolute value of the energy flow power. Based on this, components with positive energy flow power at the dominant frequency are identified as oscillation sources, while components with negative energy flow power are identified as energy consumers. However, in actual wind farms, the number of wind turbines is large, and the system scale is enormous, leading to a sharp increase in the computational workload of energy flow calculations. There is a lack of efficient calculation methods to improve overall analysis efficiency. In addition, when multiple oscillation sources exist in the system, or when multi-frequency, wide-band oscillations occur, a single energy flow component cannot distinguish the relative contributions of each oscillation source, easily leading to the risk of ambiguous positioning or even misjudgment.

[0005] Traditional methods based on linearized models or local signals are ill-suited to the strong nonlinearity and dynamic coupling characteristics of systems. They also heavily rely on power grid models and equipment parameters that are difficult to obtain precisely, resulting in insufficient accuracy in identifying novel broadband oscillations (such as sub / supersynchronous oscillations) and distributed disturbance sources. Traditional vibration source location methods only provide the location of the oscillation source without distinguishing the specific oscillation modes associated with these sources. Furthermore, existing algorithms cannot effectively decouple propagation paths in multi-oscillation source coupling scenarios, and their high computational complexity makes it difficult to meet the real-time requirements of online location. In particular, their ability to extract periodic disturbance features of forced oscillations is weak, leading to misjudgments or ambiguities in the location results, which seriously restricts the implementation of precise emergency control. Summary of the Invention

[0006] This application provides a method, apparatus, computer device, and storage medium for locating a hybrid oscillation source.

[0007] A first aspect of this application provides a method for locating a hybrid oscillation source, comprising:

[0008] The original nonlinear oscillation signal of the target power grid system is obtained, and a finite number of observation functions are constructed by time delay based on polynomial functions and trigonometric functions. The original nonlinear oscillation signal is then upgraded in dimensionality through the observation functions to construct a dataset that reflects the dynamics of a high-dimensional linear space.

[0009] Singular value decomposition and orthogonal decomposition are performed on the dataset to obtain an equivalent linear model. Modal analysis is then performed on the equivalent linear model to determine the damping ratio and frequency of each mode of the target power grid system, so as to screen out potential oscillation modes that may oscillate.

[0010] Based on the participation factors of each unit under the potential oscillation mode, the strongly correlated units under the potential oscillation mode are obtained, and the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode are determined. Based on the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode, the dissipated energy flow of each strongly correlated unit under each potential oscillation mode is determined, and the integral of the positive dissipated energy flow of the strongly correlated unit under each potential oscillation mode during the oscillation time is determined.

[0011] If the dissipated energy flow of the current unit in the current potential oscillation mode is positive and its positive dissipated energy flow integral exceeds the threshold, the current unit in the current potential oscillation mode is identified as an oscillation source. If the dissipated energy flow of the current unit in the current potential oscillation mode is negative or its positive dissipated energy flow integral exceeds the threshold, the current unit in the current potential oscillation mode is identified as a non-oscillation source.

[0012] In an optional embodiment of this application, determining the dissipated energy flow of each strongly correlated unit under each potential oscillation mode based on the active power, reactive power, voltage, and frequency of each strongly correlated unit under each potential oscillation mode includes:

[0013]

[0014] in, Let h be the dissipated energy flow of strongly correlated unit h in the i-th mode. Let be the reactive power change at the nth sampling point of the strongly correlated unit h under the i-th mode. Let be the change in active power at the nth sampling point of the strongly correlated unit h under the i-th mode. Let n be the frequency change at the nth sampling point of strongly correlated unit h in the i-th mode. Let N be the logarithmic change in the voltage amplitude at the nth sampling point of the strongly correlated unit h under the i-th mode, where N is the total number of sampling points.

[0015] In an optional embodiment of this application, the integral of the positive dissipated energy flow of strongly correlated units under each potential oscillation mode during the oscillation time is determined by the following expression:

[0016]

[0017] Among them, I h,i For the dissipated energy flow integral of strongly correlated unit h under oscillation mode i, For the integral of the positive dissipated energy flow of strongly correlated unit h under oscillation mode i, t0 is the sampling interval, t0 is the starting sampling time, and N is the total number of sampling points.

[0018] In an optional embodiment of this application, the strongly correlated units under the potential oscillation mode are obtained using the following expression, based on the participation factors of each unit under the potential oscillation mode:

[0019]

[0020] in, For the set of strongly correlated units under potential oscillation mode j, The preset participation factor threshold is n, where n is the number of units. Let be the normalized integrated participation factor of equipment h in the j-th mode, d be the strongly correlated unit, and C be the set of strongly correlated units.

[0021] In an optional embodiment of this application, the method further includes:

[0022] The relative contribution of each strongly correlated unit in each potential oscillation mode is determined based on the integral of the positive dissipated energy flow of the strongly correlated units in each potential oscillation mode during the oscillation time.

[0023]

[0024] in, Let m be the proportion of the positive dissipative energy flow of strongly correlated generating units h in the target power grid system under potential oscillation mode i, and m be the number of strongly correlated generating units under mode i. For the positive energy dissipation flow of strongly correlated unit h in the target power grid system under potential oscillation mode i, Let K be the positive dissipation energy flow of strongly correlated unit k in the target power grid system under potential oscillation mode i.

[0025] In an optional embodiment of this application, the step of increasing the dimensionality of the original nonlinear oscillating signal through an observation function to construct a dataset reflecting the dynamics of a high-dimensional linear space includes:

[0026] Discretize the original nonlinear oscillation signal to obtain the measurement data of the original nonlinear oscillation signal:

[0027]

[0028] Constructing a finite number of observation functions based on time delay using polynomial and trigonometric functions:

[0029]

[0030]

[0031] Where τ is the delay step number, m is the embedding dimension, m≥2n+1, x k For the measurement data at time k, k ≥ (m-1)τ+1, and These are finite observation functions, each consisting of a polynomial function and a trigonometric function with time delays. Let n be the active power of the nth device. Let n be the reactive power of the nth device;

[0032] The measurement data is increased in dimensionality using a finite number of observation functions to obtain the increased-dimensional measurement data. This increased-dimensional measurement data, together with the original measurement data, forms the dataset.

[0033]

[0034] in, This is a dataset.

[0035] In an optional embodiment of this application, performing singular value decomposition and orthogonal decomposition on the dataset to obtain an equivalent linear model includes:

[0036] Singular value decomposition is performed on the dataset using the following expression:

[0037]

[0038]

[0039]

[0040] in, and For dataset The two sub-data matrices formed are used to reflect the evolution dynamics of the system in a high-dimensional linear space. This is the conjugate transpose. It is a singular value matrix. U and V are orthogonal characteristic matrices. , 'a' represents the number of measurements, and 'b' represents the number of sampling points.

[0041] Select the first r-order components with larger singular values, and determine the equivalent linear model using the orthogonal decomposition method based on the corresponding singular value matrix and orthogonal eigenma matrix:

[0042]

[0043] Where K is the equivalent linear model, Let be the inverse matrix of the first r-order components with larger singular values ​​in the singular value matrix. This is the conjugate transpose of the first r-order components with larger singular values ​​in the orthogonal eigenma matrix U. These are the first r-order components with larger singular values ​​in the orthogonal eigenma matrix V. For dataset The first r-order components with relatively large singular values.

[0044] A second aspect of this application provides a positioning device for a hybrid oscillation source, comprising:

[0045] The module is used to acquire the original nonlinear oscillation signal of the target power grid system, and construct a finite number of observation functions based on the time delay of polynomial functions and trigonometric functions. The original nonlinear oscillation signal is then upgraded in dimensionality through the observation functions to construct a dataset that reflects the dynamics of a high-dimensional linear space.

[0046] The decomposition module is used to perform singular value decomposition and orthogonal decomposition on the dataset to obtain an equivalent linear model, and to perform modal analysis on the equivalent linear model to determine the damping ratio and frequency of each mode of the target power grid system, so as to screen out potential oscillation modes that may oscillate.

[0047] The determination module is used to determine the strongly correlated units under the potential oscillation mode based on the participation factors of each unit, and to determine the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode. Based on the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode, the module determines the dissipated energy flow of each strongly correlated unit under each potential oscillation mode, and determines the integral of the positive dissipated energy flow of the strongly correlated unit under each potential oscillation mode during the oscillation time.

[0048] The identification module is used to identify the current unit of the current potential oscillation mode as an oscillation source when the dissipated energy flow of the current unit in the current potential oscillation mode is positive and its positive dissipated energy flow integral exceeds a threshold, and to identify the current unit of the current potential oscillation mode as a non-oscillation source when the dissipated energy flow of the current unit in the current potential oscillation mode is negative or its positive dissipated energy flow integral exceeds a threshold.

[0049] A third aspect of this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method for locating a hybrid oscillation source.

[0050] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method for locating a hybrid oscillation source.

[0051] Compared with the prior art, the technical solutions provided in this application have at least some or all of the following advantages:

[0052] The hybrid oscillation source localization method described in this application acquires the original nonlinear oscillation signal of the target power grid system, and constructs a finite number of observation functions based on time delays using polynomial and trigonometric functions. The original nonlinear oscillation signal is then upgraded in dimensionality using these observation functions to construct a dataset reflecting the dynamics of a high-dimensional linear space. Singular value decomposition and orthogonal decomposition are performed on the dataset to obtain an equivalent linear model. Modal analysis is then performed on the equivalent linear model to determine the damping ratio and frequency of each mode in the target power grid system, thereby screening out potential oscillation modes that may cause oscillations. Based on the participation factors of each unit under the potential oscillation mode, the strongly correlated units under that potential oscillation mode are obtained, and the active power, reactive power, voltage, and frequency of each strongly correlated unit under each potential oscillation mode are determined. Finally, based on the participation factors of each strongly correlated unit under each potential oscillation mode, the active power, reactive power, voltage, and frequency of each strongly correlated unit are determined. By analyzing the active power, reactive power, voltage, and frequency of the shut-down units, the dissipated energy flow of each strongly correlated unit under each potential oscillation mode is determined, and the integral of the positive dissipated energy flow of the strongly correlated units under each potential oscillation mode during the oscillation time is determined. If the dissipated energy flow of the current unit in the current potential oscillation mode is positive and its positive dissipated energy flow integral exceeds a threshold, the current unit in the current potential oscillation mode is identified as an oscillation source. If the dissipated energy flow of the current unit in the current potential oscillation mode is negative or its positive dissipated energy flow integral exceeds a threshold, the current unit in the current potential oscillation mode is identified as a non-oscillation source. This method can quickly and accurately locate single or multiple mixed oscillation sources causing power grid system oscillations, improve the stability and security of the power grid system, and provide support for quickly isolating and cutting off oscillation sources. Attached Figure Description

[0053] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0054] Figure 1 A flowchart illustrating a method for locating a hybrid oscillation source according to an embodiment of this application;

[0055] Figure 2 A flowchart illustrating a method for locating a hybrid oscillation source according to another embodiment of this application;

[0056] Figure 3 A power grid system topology diagram provided in one embodiment of this application;

[0057] Figures 4(a) to 4(d) are schematic diagrams of the predicted trajectory and the measured trajectory provided in one embodiment of this application;

[0058] Figures 5(a) to 5(d) are schematic diagrams of the dissipated energy flow of each strongly correlated unit under different critical oscillation modes provided in one embodiment of this application;

[0059] Figure 6 A schematic diagram of the positioning device structure for a hybrid oscillation source provided in one embodiment of this application;

[0060] Figure 7 This is a schematic diagram of a computer device structure provided in one embodiment of this application. Detailed Implementation

[0061] To make the technical solutions and advantages of the embodiments of this application clearer, the exemplary embodiments of this application will be described in further detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0062] Please see Figure 1 and Figure 2 The method for locating a hybrid oscillation source provided in this application includes the following steps S100~S400:

[0063] S100: Obtain the original nonlinear oscillation signal of the target power grid system, and construct a finite number of observation functions based on the time delay of polynomial functions and trigonometric functions. Upgrade the original nonlinear oscillation signal through the observation functions to construct a dataset that reflects the dynamics of a high-dimensional linear space.

[0064] S200, Singular value decomposition and orthogonal decomposition are performed on the dataset to obtain an equivalent linear model, and modal analysis is performed on the equivalent linear model to determine the damping ratio and frequency of each mode of the target power grid system in order to screen out potential oscillation modes that may oscillate.

[0065] S300, based on the participation factors of each unit under the potential oscillation mode, obtains the strongly correlated units under the potential oscillation mode, and determines the active power, reactive power, voltage, and frequency of each strongly correlated unit under each potential oscillation mode. Based on the active power, reactive power, voltage, and frequency of each strongly correlated unit under each potential oscillation mode, it determines the dissipated energy flow of each strongly correlated unit under each potential oscillation mode, and determines the integral of the positive dissipated energy flow of each strongly correlated unit under each potential oscillation mode during the oscillation time, quantifies the impact of each unit on the power system oscillation, and realizes the location of the hybrid oscillation source;

[0066] S400: If the dissipated energy flow of the current unit in the current potential oscillation mode is positive and its positive dissipated energy flow integral exceeds the threshold, the current unit in the current potential oscillation mode is identified as an oscillation source. If the dissipated energy flow of the current unit in the current potential oscillation mode is negative or its positive dissipated energy flow integral exceeds the threshold, the current unit in the current potential oscillation mode is identified as a non-oscillation source.

[0067] In an optional embodiment of this application, in step S200, the equivalent linear model is the Koopman equivalent linear model, and the potential oscillation modes that may oscillate are weakly damped and negatively damped modes. In step S500, if the dissipated energy flow of the current unit in the current potential oscillation mode is negative, it means that the unit under this mode absorbs oscillation energy, or its positive dissipated energy flow integral does not exceed the threshold, and it is identified as a non-oscillation source. If the dissipated energy flow of the current unit in the current potential oscillation mode is positive, it means that the unit under this mode absorbs oscillation energy, and its positive dissipated energy flow integral exceeds the threshold, and it is identified as an oscillation source, and its oscillation frequency is the frequency under this mode.

[0068] The method for locating hybrid oscillation sources in this application, based on the Koopman algorithm, is used to locate and identify hybrid broadband oscillation sources in power grid systems. For the broadband oscillation problem in power grid systems, it first constructs a finite number of observation functions based on time delays using polynomial and trigonometric functions, increasing the dimensionality of the original nonlinear oscillation signal to create a new dataset reflecting the dynamics of a high-dimensional linear space. Next, singular value decomposition and orthogonal decomposition are performed on this dataset to obtain a Koopman equivalent linear model. Then, modal analysis is performed on the obtained equivalent linear model to calculate the damping ratio and frequency of each mode in the system. Finally, weakly damped and negatively damped modes are selected to obtain potential oscillation modes that may cause oscillations. The participation factors of each device under the potential oscillation modes are calculated to identify the strongly correlated devices under these potential oscillation modes. The active power, reactive power, voltage, and frequency of strongly correlated units under each potential oscillation mode are determined. Finally, the dissipated energy flow of these strongly correlated units under these modes is calculated along with the integral of the positive dissipated energy flow of each device during the oscillation time. This quantifies the impact of each device on power system oscillation. If its dissipated energy flow is negative or its positive dissipated energy flow integral does not exceed a threshold, it indicates that the unit under that mode absorbs oscillation energy and is identified as a non-oscillation source. If its dissipated energy flow is positive and its positive dissipated energy flow integral exceeds a threshold, it indicates that the unit under that mode absorbs oscillation energy and is identified as an oscillation source, with its oscillation frequency being the frequency of that mode. By using mode identification technology, only the modes that may oscillate are analyzed, improving the identification efficiency. Compared with the traditional oscillation energy method, it can be used for the identification of oscillation sources in more complex power grid systems.

[0069] In an optional embodiment of this application, step S100, which involves upscaling the original nonlinear oscillation signal using an observation function to construct a dataset reflecting the dynamics of a high-dimensional linear space, includes:

[0070] Discretize the original nonlinear oscillation signal to obtain the measurement data of the original nonlinear oscillation signal:

[0071]

[0072] Constructing a finite number of observation functions based on time delay using polynomial functions and trigonometric functions:

[0073]

[0074]

[0075] Where τ represents the delay steps, m represents the embedding dimension (satisfying m≥2n+1), and x k The measurement data represents the data at time k (to ensure the integrity of the delay data, k ≥ (m-1)τ+1). , These represent a finite number of observation functions, each consisting of a polynomial function and a trigonometric function with time delays.

[0076] The measurement data is increased in dimensionality using a finite number of observation functions to obtain the increased-dimensional measurement data. This increased-dimensional measurement data, together with the original measurement data, forms the dataset.

[0077]

[0078] in, This is a dataset.

[0079] In an optional embodiment of this application, the Koopman operator is an infinite-dimensional linear operator that can map the evolution trajectory of a finite-dimensional nonlinear dynamic system to an infinite-dimensional linear space for analysis. Within this infinite-dimensional linear space, powerful linear system analysis tools can be used to study the dynamic characteristics of the original nonlinear system. If the nonlinear dynamic model of the original system is:

[0080]

[0081] Where x is the state vector, t is the time variable, з is the system parameter, and f is the nonlinear function of the system dynamics.

[0082] In the nonlinear dynamic model of the original system, the system's time evolution trajectory is usually obtained through discretization, and the measurement data can be expressed as:

[0083]

[0084] Where, x k It is measurement data in the power system.

[0085] Constructing a finite number of observation functions based on time delay using polynomial functions and trigonometric functions:

[0086]

[0087]

[0088] Where τ represents the delay steps, m represents the embedding dimension (satisfying m≥2n+1), and x k The measurement data represents the data at time k (to ensure the integrity of the delay data, k ≥ (m-1)τ+1). , These represent a finite number of observation functions, each consisting of a polynomial function and a trigonometric function with time delays. Let n be the active power of the nth device. Let be the reactive power of the nth device.

[0089] The main idea of ​​the Koopman operator is to use a series of observation functions g(x) k After elevating the measurement data trajectory of the original system to an infinite dimension, a linear model K is found in this infinite-dimensional linear space (i.e., the observation space) to analyze the evolution of the system over time in the observation space, namely:

[0090]

[0091] Let the system's measurement data be There are *a* measurements and *b* sampling points. The measurement data is then subjected to a finite number of observation functions g(x) to increase its dimensionality, and combined with the original measurement data to form a new dataset X. L :

[0092]

[0093] in, This is a dataset.

[0094] Two sub-data matrices are constructed to reflect the evolutionary dynamics of the system in a high-dimensional linear space, namely:

[0095]

[0096]

[0097] in, and For dataset The two sub-data matrices are constructed to reflect the evolution dynamics of the system in a high-dimensional linear space.

[0098] Based on the above expression, we have:

[0099]

[0100] The Koopman equivalent linear model can then be calculated as follows:

[0101]

[0102] Where mp is the Moore-Penrose pseudo-inverse operation.

[0103] Although It is possible to estimate the Koopman equivalent linear model, but the actual measurement data is of a huge type and has a huge number of sampling points, so directly using... The computational cost is very high, so Singular Value Decomposition (SVD) can be used to ignore the corresponding data with smaller singular values, thereby reducing the computational cost.

[0104] In an optional embodiment of this application, step S200, which involves performing singular value decomposition and orthogonal decomposition on the dataset to obtain an equivalent linear model, includes:

[0105] Singular value decomposition is performed on the dataset using the following expression:

[0106]

[0107]

[0108]

[0109] in, and For dataset The two sub-data matrices are constructed to reflect the evolutionary dynamics of the system in a high-dimensional linear space. This is the conjugate transpose. It is a singular value matrix. U and V are orthogonal characteristic matrices. , , where a is the number of measurements and b is the number of sampling points.

[0110] Select the first r-order components with larger singular values, and determine the equivalent linear model using the orthogonal decomposition method based on the corresponding singular value matrix and orthogonal eigenma matrix:

[0111]

[0112] Where K is the equivalent linear model, Let be the inverse matrix of the first r-order components with larger singular values ​​in the singular value matrix. This is the conjugate transpose of the first r-order components with larger singular values ​​in the orthogonal eigenma matrix U. These are the first r-order components with larger singular values ​​in the orthogonal eigenma matrix V. For dataset The first r-order components with relatively large singular values.

[0113] The method for locating hybrid oscillation sources in this application is based on increasing the dimensionality of the original measurement signal using a finite number of observation functions. It combines singular value decomposition (SVD) and orthogonal decomposition methods to obtain a Koopman equivalent linear model that reflects the dynamics of a high-dimensional linear space, while effectively reducing the computational load while ensuring model accuracy.

[0114] In an optional embodiment of this application, the equivalent linear model K is a finite-dimensional approximate Koopman equivalent linear model, according to... The dynamic evolution trajectory of the system can be further predicted, and the dynamic trajectory at time k is...

[0115]

[0116] In an optional embodiment of this application, step S200, which involves performing modal analysis on the equivalent linear model to determine the damping ratio and frequency of each mode of the target power grid system in order to screen out potential oscillation modes that may oscillate, includes:

[0117] The oscillation characteristics of the equivalent linear model are analyzed using the following expression. Since K in the Koopman equivalent linear model is a linear operator, it satisfies the following:

[0118]

[0119] in, For eigenvalues, These are the eigenvectors.

[0120] The following expression converts discrete eigenvalues ​​into continuous-time eigenvalues:

[0121]

[0122] in, The sampling interval is... Represents continuous eigenvalues. These are the discrete eigenvalues ​​of K in the Koopman approximation.

[0123] Based on continuous eigenvalues Determine the damping ratio, and identify oscillation modes with damping ratios less than a preset threshold as potential oscillation modes, i.e., critical oscillations.

[0124]

[0125]

[0126] Where σ is the real part of the eigenvalue (attenuation rate) and ω is the imaginary part (oscillation frequency).

[0127] In an optional embodiment of this application, oscillation modes with damping ratios less than a preset threshold are determined based on the sampling interval using the following expression, and these oscillation modes with damping ratios less than the preset threshold are identified as potential oscillation modes:

[0128]

[0129] in, It is the set of critical oscillation modes. The sampling interval is... For the damping ratio, This is the damping ratio threshold.

[0130] In an optional embodiment of this application, based on eigenvalues ​​and eigenvectors, the observable function g of the system at time k can be expressed by Koopman modal analysis as follows:

[0131]

[0132] Among them, b i is the activation coefficient of the i-th Koopman mode, determined by the initial conditions of the system; r is the first r orders with larger singular values.

[0133] Due to g a The original measurement value is x, therefore the j-th measurement value is x. j The evolution of time k over time can be expressed as:

[0134]

[0135] Among them, v ji Let be the j-th eigenvalue of the i-th Koopman modality vector.

[0136] The method for locating the hybrid oscillation source in this application involves performing modal analysis on the Koopman equivalent linear model, extracting the damping ratio and frequency of each mode, and setting a damping ratio threshold to screen out the critical oscillation modes with weak damping and negative damping.

[0137] In an optional embodiment of this application, in step S300, only participation factor analysis is performed on these critical oscillation modes to obtain the strongly correlated units under these modes, and then energy flow analysis is performed to improve computational efficiency and positioning accuracy.

[0138] according to Its right eigenvector matrix can be obtained, and its left eigenvector matrix can be obtained according to the following formula. :

[0139]

[0140] in, This is the left eigenvector matrix.

[0141] Its left eigenvector matrix and right eigenvector matrix are orthogonal to each other, that is, they satisfy the following equation.

[0142]

[0143]

[0144]

[0145] The participation factor measures the degree of participation of the i-th state variable in the j-th mode. The coupling degree between mode j and state i can be represented by the right eigenvector components. and left eigenvector components Jointly portrayed:

[0146]

[0147] Among them, PF ij This represents the participation factor of the i-th state variable in the j-th mode.

[0148] To facilitate comparison of the relative contributions of different states to the same mode, the participation factors of each state in mode j are normalized to their maximum values:

[0149]

[0150] in It is the participation factor of the i-th state variable in the j-th mode after normalization.

[0151] The state variables of device h are

[0152]

[0153] in, This is the set of indices for the state variables of device h.

[0154] By synthesizing all the state variables of the device in mode j, the synthesis participation factor of the device in mode j is obtained as follows:

[0155]

[0156] in, Let h be the comprehensive participation factor of device h in mode j.

[0157] To facilitate comparison of the relative contributions of different devices in the same mode, the participation factors of each device in mode j are normalized to their maximum values:

[0158]

[0159] in, It is the normalized integrated participation factor of device h in the j-th mode.

[0160] In an optional embodiment of this application, in step S300, the strongly correlated units under the potential oscillation mode are obtained based on the participation factors of each unit under the potential oscillation mode using the following expression:

[0161]

[0162] in, For the set of strongly correlated units under potential oscillation mode j, The preset participation factor threshold is n, where n is the number of units. Let be the normalized integrated participation factor of equipment h in the j-th mode, d be the strongly correlated unit, and C be the set of strongly correlated units.

[0163] In an optional embodiment of this application, in step S300, the participation factors of the state variables of each unit under the potential oscillation mode are calculated based on the left eigenvector matrix and the right eigenvector matrix. Then, the comprehensive participation factor of each unit under the oscillation mode is obtained based on the participation factors of the state variables of each unit. Units with a comprehensive participation factor exceeding a preset threshold of 0.15 are defined as strongly correlated units under the potential oscillation mode.

[0164] Based on the modal components of each measurement trajectory, these modal components can be used to calculate the dissipated energy flow (MDEF) of strongly correlated units under the potential oscillating modes. In the observable function corresponding to device h, the exponents of P, Q, lnV, and f are given by:

[0165]

[0166] in, The set of state variable indices for device h and satisfying .

[0167] In an optional embodiment of this application, in step S400, the active power, reactive power, voltage, and frequency of strongly correlated units under each potential oscillation mode are determined using the following expressions:

[0168]

[0169] in, Let h be the active power of device h at time k in the i-th mode. Let h be the reactive power of device h at time k in the i-th mode. Let h be the voltage of device h at time k in the i-th mode. Let h be the frequency of device h at time k in the i-th mode. It is the 4(h-1)+1th eigenvalue of the i-th modal vector. It is the 4th (h-1)+2th eigenvalue of the i-th modal vector. It is the 4th (h-1)+3rd eigenvalue of the i-th modal vector. It is the 4th (h-1)+4th eigenvalue of the i-th modal vector.

[0170] In an optional embodiment of this application, step S500, which involves determining the dissipated energy flow of each strongly correlated unit in each potential oscillation mode based on the active power, reactive power, voltage, and frequency of each unit in each potential oscillation mode, includes:

[0171]

[0172] in, Let h be the dissipated energy flow of strongly correlated unit h in the i-th mode. Let be the reactive power change at the nth sampling point of the strongly correlated unit h under the i-th mode. Let be the change in active power at the nth sampling point of the strongly correlated unit h under the i-th mode. Let n be the frequency change at the nth sampling point of strongly correlated unit h in the i-th mode. Let N be the logarithmic change in the voltage amplitude at the nth sampling point of the strongly correlated unit h under the i-th mode, where N is the total number of sampling points.

[0173] Calculate the integral of the positive dissipative energy flow of each strongly correlated unit under the potential oscillation mode during the oscillation time:

[0174]

[0175] In the formula, I h,i For the integral of the dissipated energy flow of strongly correlated unit h under oscillatory mode i; For the integral of the positive dissipated energy flow of strongly correlated unit h under oscillatory mode i; t0 is the sampling interval, t0 is the starting sampling time, and N is the total number of sampling points.

[0176] The relative contribution of each strongly correlated unit under the potential oscillation mode is calculated based on the integral of the positive dissipation energy flow of the equipment:

[0177]

[0178] In the formula, The term represents the proportion of positive dissipative energy flow of strongly correlated units h in the power system under potential oscillation mode i, and m represents the number of strongly correlated units under mode i. For the positive energy dissipation flow of strongly correlated unit h in the target power grid system under potential oscillation mode i, Let K be the positive dissipation energy flow of strongly correlated unit k in the target power grid system under potential oscillation mode i.

[0179] In an optional embodiment of this application, once the dissipated energy flow E for each strongly correlated unit for the i-th oscillation mode is obtained... hi The proportion of positive dissipated energy flow for each strongly correlated unit under potential oscillation mode i This can be achieved by analyzing E hi The sum of positive and negative values ​​and The oscillation source is identified by whether it exceeds a threshold. Specifically, in mode i, the dissipated energy flow W of each unit... hi If its dissipated energy flow is negative, it means that the device h absorbs oscillating energy in mode i, or its positive dissipated energy flow integral does not exceed the threshold, and it is identified as a non-oscillating source; if it is positive, it means that the device h injects oscillating energy into the system in mode i, and its positive dissipated energy flow integral exceeds the threshold, and it is considered an oscillating source.

[0180] The method for locating hybrid oscillation sources in this application calculates the dissipated energy flow of strongly correlated equipment in different modes and the integral of the positive dissipated energy flow of each strongly correlated unit during the oscillation time based on the active power, reactive power, voltage and frequency components of each mode. It quantifies the impact of each strongly correlated unit on the oscillation of the power system in each mode, and determines whether it is an oscillation source or a non-oscillation source in the corresponding mode by the positive or negative energy flow and the proportion of positive dissipated energy flow, thereby achieving accurate identification and location of multiple frequency bands and multiple oscillation sources.

[0181] In an optional embodiment of this application, based on the Koopman algorithm for locating and identifying hybrid broadband oscillation sources in a power grid system, a four-machine, two-region system connected to a grid-connected converter (GFM) was built on the simulation platform DIgSLENT / PowerFactory. Simulation verification was then performed on this system. The power grid system topology is as follows: Figure 3 As shown. Three oscillation signals of different frequencies were applied: a 5Hz sinusoidal signal on GFM, a 20Hz sinusoidal signal on G1, and a 150Hz sinusoidal signal on G4. A 20s time-domain simulation with a step size of 0.001s was performed to obtain the dynamic response of the test system. To evaluate the accuracy of the Koopman model, the predicted trajectory of the initial values ​​at t=2~5s was compared with the measured values, as shown in Figures 4(a)~4(d). In this example, the modes with a damping ratio below 0.1 (ξ) t The condition < 0.1) was selected as the critical mode, and the equipment's overall participation factor was greater than 0.15. Equipment with a positive energy dissipation flow ratio greater than 0.15 is considered a strongly correlated unit, and a positive energy dissipation flow ratio of 30% is used as a threshold (a positive energy dissipation flow ratio exceeding 30% is considered an oscillation source). The critical oscillation mode of this example is shown in Table 1 below. The comprehensive participation factor of each equipment under this critical oscillation mode is shown in Table 2 below. The integral proportion of positive energy dissipation flow of each equipment under each potential oscillation mode is shown in Table 3 below.

[0182] Table 1

[0183]

[0184] Table 2

[0185]

[0186] Table 3

[0187]

[0188] Four critically damped oscillation modes can be observed, namely 150Hz, 20Hz, 10Hz, and 5Hz, which can be regarded as modes triggered by GFM, G1, and G4 oscillation signals. Then, the comprehensive participation factor of each device is calculated to obtain the strongly correlated units under each oscillation mode. According to Table 2, the strongly correlated units in mode 1 are G3, G4, and GFM; the strongly correlated units in mode 2 are G1 and G2; the strongly correlated units in mode 3 are G2 and GFM; and the strongly correlated units in mode 4 are G1, G2, G3, G4, and GFM. Using the extracted modal components of each strongly correlated unit, P, Q, lnV, and f are obtained, and the dissipated energy flow of each device under different critical modes is calculated as shown in Figures 5(a) to 5(d). The integral proportion of the positive dissipated energy flow of each strongly correlated unit under each potential oscillation mode is shown in Table 2. In Figure 5(a), in mode 1, G4 injects oscillatory energy into the test system and its positive dissipation energy flow integral is greater than the threshold of 30%, while other devices consume or do not inject oscillatory energy (positive dissipation energy flow integral is 0). Similarly, in Figure 5(b), G1 is the oscillator of mode 2, in Figure 5(d), GFM is the oscillator of mode 4, and in Figure 5(c), all devices consume or do not inject oscillatory energy (positive dissipation energy flow integral is 0), therefore there is no oscillator in mode 3.

[0189] Therefore, the proposed method successfully sets GFM to a 5Hz oscillation of mode 4, G1 to a 20Hz oscillation of mode 2, and G4 to a 150Hz oscillation of mode 1, while mode 3 at 10Hz does not oscillate, which is consistent with the position where the oscillation signal is injected.

[0190] The hybrid oscillation source localization method of this application addresses the problem of multiple oscillation sources with different frequencies and broadband oscillation coupling in the system. It can effectively distinguish the contributions of oscillation sources in different frequency bands and avoid the risk of fuzzy or misjudged localization by traditional methods.

[0191] This application presents a method for locating hybrid oscillation sources, providing the location and identification of hybrid broadband oscillation sources in power electronic equipment systems used for new energy access. It utilizes the Koopman algorithm to upscale nonlinear oscillation signals and construct an equivalent linear model. Combined with modal analysis and oscillation energy flow calculation, it achieves precise location of multi-band, multi-source oscillations. This method can quickly distinguish the modal contributions of each source in the presence of multiple oscillation sources and complex propagation paths, improving location accuracy and real-time performance. Based on the relative gain of the input-output signals obtained by the system, it obtains the dynamic interaction analysis matrix of the system, determines the frequency range of strong coupling and the coupling control loop, and provides early warning of situations where negative interactions lead to system instability. Furthermore, this method can be implemented based on wide-area measurement data (WAMS / PMU), does not rely on precise system parameters and models, and has low computational complexity, making it suitable for online oscillation source monitoring and emergency control decision-making in large-scale new energy access systems.

[0192] It should be understood that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order constraint on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the diagram may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0193] Please see Figure 6 One embodiment of this application provides a positioning device 600 for a hybrid oscillation source, comprising:

[0194] The construction module 610 is used to acquire the original nonlinear oscillation signal of the target power grid system, and construct a finite number of observation functions based on the time delay of polynomial functions and trigonometric functions. The original nonlinear oscillation signal is then upgraded in dimensionality through the observation functions to construct a dataset that reflects the dynamics of a high-dimensional linear space.

[0195] The decomposition module 620 is used to perform singular value decomposition and orthogonal decomposition on the dataset to obtain an equivalent linear model, and to perform modal analysis on the equivalent linear model to determine the damping ratio and frequency of each mode of the target power grid system, so as to screen out potential oscillation modes that may oscillate.

[0196] The determination module 630 is used to determine the strongly correlated units under the potential oscillation mode based on the participation factors of each unit under the potential oscillation mode, and to determine the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode. Based on the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode, the dissipated energy flow of each strongly correlated unit under each potential oscillation mode is determined, and the integral of the positive dissipated energy flow of the strongly correlated unit under each potential oscillation mode is determined during the oscillation time.

[0197] The identification module 640 is used to identify the current unit of the current potential oscillation mode as an oscillation source when the dissipated energy flow of the current unit in the current potential oscillation mode is positive and its positive dissipated energy flow integral exceeds a threshold, and to identify the current unit of the current potential oscillation mode as a non-oscillation source when the dissipated energy flow of the current unit in the current potential oscillation mode is negative or its positive dissipated energy flow integral exceeds a threshold.

[0198] The hybrid oscillation source localization device of this application converts the original nonlinear signal into an equivalent linear model through the Koopman algorithm, and then performs modal analysis on the equivalent linear model. For each potential oscillation mode that may oscillate, dissipative energy flow analysis is performed to identify and locate single or multiple oscillation sources with mixed oscillations. This solves the problem that the traditional oscillation energy flow method cannot accurately identify and locate oscillation sources and their oscillation frequencies in broadband oscillations of power grid systems.

[0199] For specific limitations regarding the aforementioned device 600, please refer to the above description of the positioning method for the hybrid oscillation source of this application, which will not be repeated here. Each module in the aforementioned device 600 can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in hardware or independently of the processor in a computer device, or stored in software in the memory of a computer device, so that the processor can call and execute the operations corresponding to each module.

[0200] In one embodiment, a computer device is provided, the internal structure of which can be as follows: Figure 7As shown. The computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores data. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements the hybrid oscillation source localization method of this application as described above. It includes: a memory and a processor; the memory stores a computer program; and the processor executes the computer program to implement any step of the hybrid oscillation source localization method of this application as described above.

[0201] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, can implement any step in the method for locating a hybrid oscillation source as described above in this application.

[0202] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0203] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0204] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0205] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0206] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0207] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for locating a hybrid oscillation source, characterized in that, include: The original nonlinear oscillation signal of the target power grid system is obtained, and a finite number of observation functions are constructed by time delay based on polynomial functions and trigonometric functions. The original nonlinear oscillation signal is then upgraded in dimensionality through the observation functions to construct a dataset that reflects the dynamics of a high-dimensional linear space. Singular value decomposition and orthogonal decomposition are performed on the dataset to obtain an equivalent linear model. Modal analysis is then performed on the equivalent linear model to determine the damping ratio and frequency of each mode of the target power grid system, so as to screen out potential oscillation modes that may oscillate. Based on the participation factors of each unit under the potential oscillation mode, the strongly correlated units under the potential oscillation mode are obtained, and the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode are determined. Based on the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode, the dissipated energy flow of each strongly correlated unit under each potential oscillation mode is determined, and the integral of the positive dissipated energy flow of the strongly correlated unit under each potential oscillation mode during the oscillation time is determined. If the dissipated energy flow of the current unit in the current potential oscillation mode is positive and its positive dissipated energy flow integral exceeds a threshold, the current unit in the current potential oscillation mode is identified as an oscillation source. If the dissipated energy flow of the current unit in the current potential oscillation mode is negative or its positive dissipated energy flow integral does not exceed a threshold, the current unit in the current potential oscillation mode is identified as a non-oscillation source. The method of increasing the dimensionality of the original nonlinear oscillating signal through an observation function to construct a dataset reflecting the dynamics of a high-dimensional linear space includes: Discretize the original nonlinear oscillation signal to obtain the measurement data of the original nonlinear oscillation signal: ； Constructing a finite number of observation functions based on time delay using polynomial and trigonometric functions: ； ； Where τ is the delay step number, m is the embedding dimension, m≥2n+1, x k For the measurement data at time k, k ≥ (m-1)τ+1, and These are finite numbers of observation functions, each consisting of a time delay of a trigonometric function and a polynomial function. Let n be the active power of device n. Let n be the reactive power of device n. The measurement data is increased in dimensionality using a finite number of observation functions to obtain the increased-dimensional measurement data. This increased-dimensional measurement data, together with the original measurement data, forms the dataset. ； in, For the dataset, according to Its right eigenvector matrix can be obtained. The transpose of its left eigenvector matrix can be obtained according to the following formula. : ； in, It is the transpose of the left eigenvector matrix. Its left eigenvector matrix and right eigenvector matrix are orthogonal to each other, that is, they satisfy the following equation. ； ； ； The participation factor measures the degree of participation of the i-th state variable in the j-th mode. The coupling degree between mode j and state i can be represented by the right eigenvector components. and left eigenvector components Jointly portrayed: ； Among them, PF ij This represents the participation factor of the i-th state variable in the j-th mode. To facilitate comparison of the relative contributions of different states to the same mode, the participation factors of each state in mode j are normalized to their maximum values: ； in It is the participation factor of the i-th state variable in the j-th mode after normalization. The state variables of device h are ； in, For the set of state variable indices of device h, By synthesizing all the state variables of the device in mode j, the synthesis participation factor of the device in mode j is obtained as follows: ； in, The comprehensive participation factor of device h in mode j. To facilitate comparison of the relative contributions of different devices in the same mode, the participation factors of each device in mode j are normalized to their maximum values: ； in, It is the normalized integrated participation factor of device h in the j-th mode.

2. The method according to claim 1, characterized in that, The determination of the dissipated energy flow of each strongly correlated unit under each potential oscillation mode, based on the active power, reactive power, voltage, and frequency of each unit under each potential oscillation mode, includes: ； in, Let h be the dissipated energy flow of strongly correlated unit h in the i-th mode. Let be the reactive power change at the nth sampling point of the strongly correlated unit h under the i-th mode. Let be the change in active power at the nth sampling point of the strongly correlated unit h under the i-th mode. Let n be the frequency change at the nth sampling point of strongly correlated unit h in the i-th mode. Let N be the logarithmic change in the voltage amplitude at the nth sampling point of the strongly correlated unit h under the i-th mode, where N is the total number of sampling points.

3. The method according to claim 1, characterized in that, The integral of the positive dissipated energy flow of strongly correlated units under each potential oscillation mode during the oscillation time is determined by the following expression: ； Among them, I h,i For the dissipated energy flow integral of strongly correlated unit h under oscillation mode i, For the integral of the positive dissipated energy flow of strongly correlated unit h under oscillation mode i, Here, t0 is the sampling interval, t0 is the starting sampling time, and N is the total number of sampling points. and Let be the energy dissipation flow of strongly correlated unit h under oscillation mode i at time t, and the energy dissipation of strongly correlated unit h at the kth sampling point under oscillation mode i, respectively.

4. The method according to claim 1, characterized in that, The following expression is used to obtain the strongly correlated units under the potential oscillation mode, based on the participation factors of each unit: ； in, For the set of strongly correlated units under potential oscillation mode j, The preset participation factor threshold is n, where n is the number of units. Let be the normalized integrated participation factor of equipment h in the j-th mode, d be the strongly correlated unit, and C be the set of strongly correlated units.

5. The method according to claim 1, characterized in that, The method further includes: The relative contribution of each strongly correlated unit in each potential oscillation mode is determined based on the integral of the positive dissipated energy flow of the strongly correlated units in each potential oscillation mode during the oscillation time. ； in, Let m be the proportion of the positive dissipative energy flow of strongly correlated generating units h in the target power grid system under potential oscillation mode i, and m be the number of strongly correlated generating units under mode i. For the positive energy dissipation flow of strongly correlated unit h in the target power grid system under potential oscillation mode i, Let K be the positive dissipation energy flow of strongly correlated unit k in the target power grid system under potential oscillation mode i.

6. The method according to claim 1, characterized in that, The step of performing singular value decomposition and orthogonal decomposition on the dataset to obtain an equivalent linear model includes: Singular value decomposition is performed on the dataset using the following expression: ； ； ； in, and For dataset The two sub-data matrices formed are used to reflect the evolution dynamics of the system in a high-dimensional linear space. This is the conjugate transpose. It is a singular value matrix. U and V are orthogonal characteristic matrices. , 'a' represents the number of measurements, and 'b' represents the number of sampling points. Select the first r-order components with larger singular values, and determine the equivalent linear model using the orthogonal decomposition method based on the corresponding singular value matrix and orthogonal eigenma matrix: ； Where K is the equivalent linear model, Let be the inverse matrix of the first r-order components with larger singular values ​​in the singular value matrix. This is the conjugate transpose of the first r-order components with larger singular values ​​in the orthogonal eigenma matrix U. These are the first r-order components with larger singular values ​​in the orthogonal eigenma matrix V. For dataset The first r-order components with relatively large singular values.

7. A positioning device for a hybrid oscillation source, characterized in that, include: The module is used to acquire the original nonlinear oscillation signal of the target power grid system, and construct a finite number of observation functions based on the time delay of polynomial functions and trigonometric functions. The original nonlinear oscillation signal is then upgraded in dimensionality through the observation functions to construct a dataset that reflects the dynamics of a high-dimensional linear space. The decomposition module is used to perform singular value decomposition and orthogonal decomposition on the dataset to obtain an equivalent linear model, and to perform modal analysis on the equivalent linear model to determine the damping ratio and frequency of each mode of the target power grid system, so as to screen out potential oscillation modes that may oscillate. The determination module is used to determine the strongly correlated units under the potential oscillation mode based on the participation factors of each unit, and to determine the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode. Based on the active power, reactive power, voltage and frequency of each strongly correlated unit under each potential oscillation mode, the module determines the dissipated energy flow of each strongly correlated unit under each potential oscillation mode, and determines the integral of the positive dissipated energy flow of the strongly correlated unit under each potential oscillation mode during the oscillation time. The identification module is used to identify the current unit in the current potential oscillation mode as an oscillation source when the dissipated energy flow of the current unit in the current potential oscillation mode is positive and its positive dissipated energy flow integral exceeds a threshold; and to identify the current unit in the current potential oscillation mode as a non-oscillation source when the dissipated energy flow of the current unit in the current potential oscillation mode is negative or its positive dissipated energy flow integral does not exceed the threshold. The method of increasing the dimensionality of the original nonlinear oscillating signal through an observation function to construct a dataset reflecting the dynamics of a high-dimensional linear space includes: Discretize the original nonlinear oscillation signal to obtain the measurement data of the original nonlinear oscillation signal: ； Constructing a finite number of observation functions based on time delay using polynomial and trigonometric functions: ； ； Where τ is the delay step number, m is the embedding dimension, m≥2n+1, x k For the measurement data at time k, k ≥ (m-1)τ+1, and These are finite numbers of observation functions, each consisting of a time delay of a trigonometric function and a polynomial function. Let n be the active power of device n. Let n be the reactive power of device n. The measurement data is increased in dimensionality using a finite number of observation functions to obtain the increased-dimensional measurement data. This increased-dimensional measurement data, together with the original measurement data, forms the dataset. ； in, For the dataset, according to Its right eigenvector matrix can be obtained. The transpose of its left eigenvector matrix can be obtained according to the following formula. : ； in, It is the transpose of the left eigenvector matrix. Its left eigenvector matrix and right eigenvector matrix are orthogonal to each other, that is, they satisfy the following equation. ； ； ； The participation factor measures the degree of participation of the i-th state variable in the j-th mode. The coupling degree between mode j and state i can be represented by the right eigenvector components. and left eigenvector components Jointly portrayed: ； Among them, PF ij This represents the participation factor of the i-th state variable in the j-th mode. To facilitate comparison of the relative contributions of different states to the same mode, the participation factors of each state in mode j are normalized to their maximum values: ； in It is the participation factor of the i-th state variable in the j-th mode after normalization. The state variables of device h are ； in, For the set of state variable indices of device h, By synthesizing all the state variables of the device in mode j, the synthesis participation factor of the device in mode j is obtained as follows: ； in, The comprehensive participation factor of device h in mode j. To facilitate comparison of the relative contributions of different devices in the same mode, the participation factors of each device in mode j are normalized to their maximum values: ； in, It is the normalized integrated participation factor of device h in the j-th mode.

8. A computer device, comprising: A memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method for locating the hybrid oscillation source according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for locating the hybrid oscillation source according to any one of claims 1 to 6.

Images