Key flow field feature extraction method for pump turbine

Abstract

The invention relates to the field of pump turbine design methods, in particular to a key flow field feature extraction method for a pump turbine. The feature extraction method comprises the following steps: acquiring flow field data, and constructing a matrix set; transforming and segmenting the matrix set by adopting a discrete Fourier transform method to obtain covariance matrixes of a plurality of time periods; decomposing the covariance matrix of each time period by adopting a spectral eigenorthogonal decomposition method, and drawing a spectral eigenorthogonal decomposition modal shape; and analyzing the modal shape, and identifying key flow field characteristics which have obvious influence on the performance of the pump turbine. According to the key flow field feature extraction method for the pump turbine, information in flow field data can be efficiently extracted on the basis of a spectral intrinsic orthogonal decomposition method, an energy peak value under a specific frequency and a corresponding key feature shape are directly given, the dependence on artificial experience is reduced, and the recognition accuracy of key features is remarkably improved.

Description

Technical Field

[0001] The invention relates to the field of water pump turbine design methods, and in particular to a key flow field feature extraction method for water pump turbines. Background Art

[0002] As a reversible hydraulic machine, the pump-turbine plays a vital role in pumped-storage power stations. It can be used as a pump when the runner rotates forward and as a turbine when it rotates reversely, realizing the storage and release of electric energy. It is of great significance for peak-shaving and valley-filling of power systems and improving the stability and economy of power grid operation. Accurately extracting the key features of the internal flow field of the pump-turbine, such as secondary vortices, energy dissipation groups, pressure pulsations, etc., is crucial for the optimization design of the pump-turbine blade shape and real-time monitoring of stability during operation.

[0003] At present, there are two main methods for obtaining the key characteristics of the internal flow field of a pump-turbine: experimental and numerical research. Among them, the experimental method usually involves installing the pump-turbine on a special test device and evaluating its performance by measuring parameters such as flow, head, and power under different working conditions. The numerical method is based on computational fluid dynamics technology, which discretizes the continuous flow field into small units and solves the set of equations in turbulent flow in each unit to simulate the flow field characteristics and performance parameters of the pump-turbine under different working conditions. However, the internal flow of the pump-turbine is complex and highly turbulent. Identifying and accurately capturing nonlinear key flow field characteristics such as vortices and backflows in the flow field requires a large amount of data support, which is costly and time-consuming. In addition, it is necessary to rely on the existing experience of engineers for manual screening and judgment, and the recognition accuracy of strong unsteady and nonlinear flows is insufficient, and the efficiency is low.

[0004] Therefore, this field needs a new technical solution to solve the above problems. Summary of the invention

[0005] In order to improve or solve the technical problem of insufficient accuracy of flow field feature extraction in the prior art, the present invention provides a key flow field feature extraction method for a pump-turbine. The feature extraction method comprises: acquiring flow field data of the pump-turbine under different working conditions, constructing a matrix set for spectral intrinsic orthogonal decomposition; transforming and segmenting the matrix set using a discrete Fourier transform method to obtain a covariance matrix for multiple time periods; decomposing the covariance matrix of each time period using a spectral intrinsic orthogonal decomposition method, and drawing the spectral intrinsic orthogonal decomposition modal shape; analyzing the modal shape to identify key flow field features that have a significant impact on the performance of the pump-turbine.

[0006] Furthermore, the step of constructing a matrix set for spectral eigenorthogonal decomposition includes:

[0007] intercepting a cross section of the flow field data covering a main operating area of ​​the pump-turbine to obtain a data snapshot in which velocity parameters are arranged in chronological order;

[0008] According to the order of acquisition time, the data snapshots are arranged into a matrix set:

[0009]

[0010] in, is the speed snapshot at the Kth time point; N q is the number of rows in the matrix set, N t is the number of data snapshots.

[0011] Furthermore, the data snapshot at each time point includes the x, y, z coordinates of the internal section of the pump turbine and the u, v, w transient velocity components; u is the instantaneous velocity component in the x direction, v is the instantaneous velocity component in the y direction, and w is the instantaneous velocity component in the z direction; the N q Equal to the number of snapshot grids multiplied by the number of variables.

[0012] Furthermore, the step of transforming and segmenting the matrix set using a discrete Fourier transform method to obtain covariance matrices for multiple time periods includes:

[0013] Segmenting the matrix set according to the acquisition time to form matrices of multiple time periods;

[0014] The matrix of each time period is transformed by using a discrete Fourier transform method to generate a covariance matrix of the corresponding time period.

[0015] Furthermore, the step of decomposing the covariance matrix of each time period by using the spectral intrinsic orthogonal decomposition method and drawing the spectral intrinsic orthogonal decomposition mode shape comprises:

[0016] Using a spectral eigenorthogonal decomposition method to perform eigenvalue decomposition on the covariance matrix of the time period to obtain an eigenvector matrix and an eigenvalue diagonal matrix;

[0017] Arrange the eigenvalue diagonal matrix of the covariance matrix of the same time period according to the frequency;

[0018] The spectral eigen-orthogonal decomposition mode shapes at different frequencies are plotted based on the eigenvector matrix.

[0019] Furthermore, the covariance matrix set C of the time periods arranged in order of frequency corresponding to the matrix set Q is:

[0020]

[0021] Among them, N bis the number of segments, that is, the number of modes expected to be decomposed, N q is the number of rows in the matrix collection.

[0022] Furthermore, the step of performing eigenvalue decomposition on the covariance matrix of the time period by using the spectral eigenorthogonal decomposition method to obtain an eigenvector matrix and an eigenvalue diagonal matrix includes:

[0023]

[0024] Among them, C is the covariance matrix set; W is the weight matrix; is the eigenvector matrix obtained by decomposition, that is, the cross-sectional modal shape corresponding to different frequencies; It is the decomposed diagonal matrix of eigenvalues ​​arranged by frequency, that is, the modal energy corresponding to different frequencies.

[0025] Furthermore, the step of plotting the spectral intrinsic orthogonal decomposition mode shapes at different frequencies according to the eigenvector matrix includes:

[0026] Plotting a modal energy spectrum according to the eigenvalue diagonal matrix, and determining a corresponding frequency of a key flow field feature in the flow field according to a peak value of the modal energy spectrum;

[0027] The main modes of the flow field at the frequency corresponding to the key flow field characteristics are extracted, and the spectral intrinsic orthogonal decomposition mode shape is plotted.

[0028] Furthermore, the key flow field characteristics include velocity pulsation waves, wake pulsation waves, local flow separation and leakage secondary vortex loss areas.

[0029] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0030] (1) The spectral intrinsic orthogonal decomposition method can efficiently extract information from flow field data and directly give the energy peak at a specific frequency and the corresponding key feature shape, avoiding the manual screening and identification of a large amount of unsteady and nonlinear flow data in traditional methods, reducing dependence on manual experience, and significantly improving the recognition accuracy of key features;

[0031] (2) The data snapshot method can specify the flow field physical quantities and a limited number of snapshots, and give the key flow field characteristic shapes based on modal analysis. It does not require a global flow field detection box to obtain all physical quantities, which reduces the recognition time and improves the recognition efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings, in which:

[0033] Figure 1 is a flow chart of a key flow field feature extraction method for a pump turbine according to the present invention;

[0034] Figure 2 It is a schematic diagram of a data snapshot in the key flow field feature extraction method for a pump turbine according to the present invention;

[0035] Figure 3 It is a schematic diagram of processing data snapshots in the key flow field feature extraction method for a pump turbine according to the present invention;

[0036] Figure 4 It is a schematic diagram of the modal energy spectrum in the key flow field feature extraction method for a pump turbine according to the present invention;

[0037] Figure 5 It is a schematic diagram of the modal shape and the corresponding key flow field features in the key flow field feature extraction method for a water pump turbine according to the present invention. DETAILED DESCRIPTION

[0038] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0039] It should be noted that in the description of the present invention, the terms "upper", "lower", "left", "right", "inner", "outer" and the like indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings, which are only for the convenience of description, and do not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0040] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installation", "setting", and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a direct connection, an indirect connection through an intermediate medium, or the internal connection of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0041] In order to improve or solve the technical problem of insufficient accuracy of flow field feature extraction in the prior art, the present invention provides a key flow field feature extraction method for a pump-turbine. The feature extraction method includes: obtaining flow field data of the pump-turbine under different working conditions, constructing a matrix set for spectral intrinsic orthogonal decomposition; using a discrete Fourier transform method to transform and segment the matrix set to obtain a covariance matrix for multiple time periods; using a spectral intrinsic orthogonal decomposition method to decompose the covariance matrix of each time period, and drawing the spectral intrinsic orthogonal decomposition mode shape; analyzing the mode shape to identify key flow field features that have a significant impact on the performance of the pump-turbine.

[0042] Figure 1 Flowchart of the key flow field feature extraction method for pump turbine of the present invention. Figure 1 As shown, the key flow field sign extraction method for a water pump turbine of the present invention comprises:

[0043] Step S1, obtaining flow field data of a pump-turbine under different working conditions, and constructing a matrix set for spectral intrinsic orthogonal decomposition; wherein the flow field data can be obtained by experiment or numerical calculation, and the specific method to be used can be determined according to the actual situation;

[0044] Step S2, using a discrete Fourier transform method to transform and segment the matrix set to obtain covariance matrices for multiple time periods; wherein discrete Fourier transform is a commonly used signal analysis method, which can transform the sampling of time domain signals into sampling in the frequency domain;

[0045] Step S3, decomposing the covariance matrix of each time period by using the spectral intrinsic orthogonal decomposition method, and drawing the spectral intrinsic orthogonal decomposition mode shape; wherein, the spectral intrinsic orthogonal decomposition is a Matlab (matrixlaboratory matrix factory) implementation of the intrinsic orthogonal decomposition in the frequency domain, which can extract the characteristic modes in frequency and space from the time series data;

[0046] Step S4, analyzing the modal shape to identify key flow field features that significantly affect the performance of the pump-turbine; by observing the characteristics and distribution of the modal shape and its association with the pump-turbine, identifying the key flow field features that significantly affect the performance of the pump-turbine; illustratively, the key flow field features include velocity pulsation waves, wake pulsation waves, local flow separation and leakage secondary vortex loss areas; Figure 5 The wake pulsation wave and the leakage vortex pulsation wave are illustrated in FIG. 1 ; among them, the leakage vortex pulsation wave is one of the reasons for the formation of the leakage secondary vortex loss zone.

[0047] Figure 2 is a schematic diagram of a data snapshot in the key flow field feature extraction method for a pump turbine according to the present invention, Figure 3FIG. 1 is a schematic diagram of processing data snapshots in the key flow field feature extraction method for a pump turbine according to the present invention. Figure 2 and Figure 3 As shown, in one or more embodiments, the step of constructing a matrix set for spectral eigenorthogonal decomposition in step S1 includes:

[0048] Step S11, intercepting a cross section of the flow field data covering the main operating area of ​​the pump-turbine, and obtaining a data snapshot of the velocity parameters arranged in chronological order; wherein, velocity pulsation is a key parameter affecting the performance of the pump-turbine, and therefore specifying the flow field velocity for analysis does not require global flow field monitoring and acquisition of all physical quantities, which can reduce recognition time and improve recognition efficiency; it can be understood that the key parameters affecting the performance of the pump-turbine also include pressure gradient, temperature gradient, vorticity, etc., and the key flow field feature extraction method for the pump-turbine of the present invention can also specify key features such as pressure gradient, temperature gradient, vorticity, etc. for analysis, and the following description will only take the analysis of the specified flow field velocity as an example.

[0049] Step S12, arranging the data snapshots into a matrix set in the order of collection time; wherein the data snapshots can be collected at preset intervals to form a limited number of data snapshots, which can record the change information of the flow field in the time dimension.

[0050] Specifically, the data snapshot can be saved in dat file format. The data snapshot at each time point includes the x, y, z coordinates of the internal section of the pump turbine and the u, v, w instantaneous velocity components; u is the instantaneous velocity component in the x direction, v is the instantaneous velocity component in the y direction, and w is the instantaneous velocity component in the z direction. The number of coordinates is the number of snapshot grids. Figure 2 Each row in is a grid; u, v, w transient velocity components are the number of variables, that is, the number of variables is 3. The matrix q formed by each data snapshot has N q OK, N q Equal to the number of snapshot grids multiplied by the number of variables.

[0051] Furthermore, the matrix set Q formed by the matrices q arranged in the order of acquisition time is:

[0052]

[0053] in, is the speed snapshot at the Kth time point; N q is the number of rows in the matrix set, which is equal to the number of snapshot grids multiplied by the number of variables; N t is the number of data snapshots.

[0054] See also Figure 3 In step S2, the discrete Fourier transform method is used to transform and segment the matrix set to obtain the covariance matrix of multiple time periods, which includes:

[0055] Step S21, segmenting the matrix set according to the acquisition time to form matrices of multiple time periods; wherein the matrix of each time period includes at least one matrix q formed by a data snapshot, so as to facilitate the analysis of flow field characteristics of different time periods;

[0056] Step S22, using the discrete Fourier transform method to transform the matrix of each time period to generate the covariance matrix of the corresponding time period; wherein each matrix in the matrix of each time period is subjected to a discrete Fourier transform to convert the time domain signal into a frequency domain signal, revealing the characteristics of the flow field in the frequency domain; the matrix of each time period is a real number set, and the corresponding covariance matrix generated is a complex number set; the covariance matrix can reflect the correlation between flow field variables.

[0057] Specifically, the matrix set can be divided into N b The number of data snapshots in each time period can be the same or different. The number of segments is also the number of modes obtained by the eigenorthogonal decomposition of the expected spectrum.

[0058] Figure 4 is a schematic diagram of the modal energy spectrum in the key flow field feature extraction method for a pump turbine according to the present invention, Figure 5 Schematic diagram of the modal shape and the corresponding key flow field features in the key flow field feature extraction method for a pump turbine according to the present invention. Figure 3 , Figure 4 and Figure 5 As shown, in one or more embodiments, the covariance matrix of each time period is decomposed by using the spectral intrinsic orthogonal decomposition method in step S3, and the step of drawing the spectral intrinsic orthogonal decomposition mode shape includes:

[0059] Step S31, using the spectral intrinsic orthogonal decomposition method to perform eigenvalue decomposition on the covariance matrix of the time period to obtain an eigenvector matrix and an eigenvalue diagonal matrix; wherein, using the spectral intrinsic orthogonal decomposition method to perform eigenvalue decomposition on each covariance matrix in each time period to obtain a corresponding eigenvector matrix and a corresponding eigenvalue diagonal matrix, the eigenvector represents the shape of the modal vector obtained by the decomposition, and the eigenvalue diagonal matrix reflects the modal energy size of different features;

[0060] Step S32, arranging the eigenvalue diagonal matrix of the covariance matrix of the same time period according to the frequency; that is, arranging all the eigenvalue diagonal matrices decomposed in the same time period according to the frequency, the frequency corresponding to the fast appearance time is high, and the frequency corresponding to the slow appearance speed is low; the covariance matrix set C of the time period arranged according to the frequency corresponding to the matrix set Q is:

[0061]

[0062] in, is the frequency K at the Nth b The covariance matrix corresponding to each segment; N b is the number of segments, that is, the expected number of modes.

[0063] Step S33: plotting the spectral eigenorthogonal decomposition mode shapes at different frequencies according to the eigenvector matrix.

[0064] Furthermore, in step S31, the steps of performing eigenvalue decomposition on the covariance matrix of the time period by using the spectral eigenorthogonal decomposition method to obtain the eigenvector matrix and the eigenvalue diagonal matrix include:

[0065]

[0066] Among them, C is the covariance matrix set; W is the weight matrix; is the eigenvector matrix obtained by decomposition, that is, the cross-sectional modal shape corresponding to different frequencies; It is the decomposed diagonal matrix of eigenvalues ​​arranged by frequency, that is, the modal energy corresponding to different frequencies.

[0067] Further, step S33 of plotting the spectral eigenorthogonal decomposition mode shapes at different frequencies according to the eigenvector matrix includes:

[0068] Step S331, plotting a modal energy spectrum according to the eigenvalue diagonal matrix, and determining the corresponding frequencies of key flow field features in the flow field according to the peak values ​​of the modal energy spectrum; specifically, the eigenvalues ​​in the eigenvalue diagonal matrix reflect the energy of the mode corresponding to the corresponding eigenvector, and the modal energy spectrum is plotted with the frequency as the abscissa and the eigenvalue as the ordinate; the modal energy spectrum can display the energy contained in different modes, wherein the frequencies corresponding to the peak points are the frequencies corresponding to the modes with higher energy in the flow field, and they represent the parts of the flow field where the energy is more concentrated, and the corresponding frequencies of the key features in the flow field are determined by observing the peak values ​​of the modal energy spectrum;

[0069] Step S332, extract the main modes of the flow field at the frequencies corresponding to the key flow field characteristics, and draw the spectral intrinsic orthogonal decomposition modal shape; the spectral intrinsic orthogonal decomposition modal shape can clearly show the specific form and distribution of the key flow field characteristics; according to the corresponding frequencies of the key flow field characteristics determined in step S331, extract the eigenvectors corresponding to the above frequencies from the eigenvector matrix, and the modes represented by these eigenvectors are the main modes of the flow field at the frequencies corresponding to the key flow field characteristics; use the eigenvectors of the main modes in combination with the flow field data to draw the shape of the main modes in the flow field; exemplarily, it can be drawn through Matlab to display the modal shape in the form of a graph, and different colors, lines or shadows can be used to represent different parameters and features, so as to clearly show the structure and characteristics of the mode.

[0070] The key flow field feature extraction method for a water pump turbine of the present invention can accurately and quickly extract the key flow field features by applying spectral intrinsic orthogonal decomposition technology to experimental monitoring or numerical simulation results of the water pump turbine.

[0071] So far, the technical scheme of the present invention has been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical schemes after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A method for extracting key flow field features of a pump-turbine, characterized in that: include: Obtain flow field data of the pump-turbine under different working conditions and construct a matrix set for spectral eigenorthogonal decomposition; The matrix set is transformed and segmented using a discrete Fourier transform method to obtain covariance matrices for multiple time periods; Decomposing the covariance matrix of each time period by using a spectral intrinsic orthogonal decomposition method, and drawing a spectral intrinsic orthogonal decomposition mode shape; The modal shapes are analyzed to identify key flow field features that significantly affect the performance of the pump-turbine.

2. The key flow field feature extraction method for a pump-turbine according to claim 1, characterized in that: The step of constructing a matrix set for spectral eigenorthogonal decomposition comprises: intercepting a cross section of the flow field data covering a main operating area of ​​the pump-turbine to obtain a data snapshot in which velocity parameters are arranged in chronological order; According to the order of acquisition time, the data snapshots are arranged into a matrix set: in, is the speed snapshot at the Kth time point; N q is the number of rows in the matrix set, N t is the number of data snapshots.

3. The key flow field feature extraction method for a pump-turbine according to claim 2 is characterized in that: The data snapshot at each time point includes the x, y, z coordinates of the internal section of the pump turbine and the u, v, w transient velocity components; u is the instantaneous velocity component in the x direction, v is the instantaneous velocity component in the y direction, and w is the instantaneous velocity component in the z direction; the N q Equal to the number of snapshot grids multiplied by the number of variables.

4. The key flow field feature extraction method for a pump-turbine according to claim 1, characterized in that: The step of transforming and segmenting the matrix set by using the discrete Fourier transform method to obtain the covariance matrix of multiple time periods comprises: Segmenting the matrix set according to the acquisition time to form matrices of multiple time periods; The matrix of each time period is transformed by using a discrete Fourier transform method to generate a covariance matrix of the corresponding time period.

5. The key flow field feature extraction method for a pump-turbine according to claim 1, characterized in that: The step of decomposing the covariance matrix of each time period by using the spectral intrinsic orthogonal decomposition method and drawing the spectral intrinsic orthogonal decomposition mode shape comprises: Using a spectral eigenorthogonal decomposition method to perform eigenvalue decomposition on the covariance matrix of the time period to obtain an eigenvector matrix and an eigenvalue diagonal matrix; Arrange the eigenvalue diagonal matrix of the covariance matrix of the same time period according to the frequency; The spectral eigen-orthogonal decomposition mode shapes at different frequencies are plotted based on the eigenvector matrix.

6. The key flow field feature extraction method for a pump-turbine according to claim 5, characterized in that: The covariance matrix set C of the time periods arranged by frequency corresponding to the matrix set Q is: Among them, N b is the number of segments, that is, the number of modes expected to be decomposed, N q is the number of rows in the matrix collection.

7. The key flow field feature extraction method for a pump-turbine according to claim 6, characterized in that: The step of performing eigenvalue decomposition on the covariance matrix of the time period by using the spectral eigenorthogonal decomposition method to obtain an eigenvector matrix and an eigenvalue diagonal matrix comprises: Among them, C is the covariance matrix set; W is the weight matrix; is the eigenvector matrix obtained by decomposition, that is, the cross-sectional modal shape corresponding to different frequencies; It is the decomposed diagonal matrix of eigenvalues ​​arranged by frequency, that is, the modal energy corresponding to different frequencies.

8. The key flow field feature extraction method for a pump-turbine according to claim 5, characterized in that: The step of plotting the spectral intrinsic orthogonal decomposition mode shapes at different frequencies according to the eigenvector matrix comprises: Plotting a modal energy spectrum according to the eigenvalue diagonal matrix, and determining a corresponding frequency of a key flow field feature in the flow field according to a peak value of the modal energy spectrum; The main modes of the flow field at the frequency corresponding to the key flow field characteristics are extracted, and the spectral intrinsic orthogonal decomposition mode shape is plotted.

9. The key flow field feature extraction method for a pump-turbine according to claim 1, characterized in that: The key flow field features include velocity pulsation waves, wake pulsation waves, local flow separation and leakage secondary vortex loss areas.

Images