Vibration test and excitation load feedback analysis method for underground powerhouse structure of pumped storage power station

Abstract

The invention discloses a pumped storage power station underground powerhouse structure vibration test and excitation load feedback analysis method comprising the following steps: S1, combining time domain and frequency domain analysis to extract vibration characteristics of each structure of a powerhouse; s2, calculating three types of vibration sources, namely contribution degrees of machinery, hydraulic power and electricity to structural vibration; s3, adopting a finite element numerical simulation method and combining field actual measurement data to perform unit excitation load feedback analysis; the technical problems that the contribution of various vibration sources to plant vibration cannot be accurately judged in the prior art, so that the design of vibration control measures is lack of pertinence, and the transformation effect cannot be accurately predicted when the plant structure is subjected to anti-vibration optimization and transformation are solved.

Description

Technical Field

[0001] The invention relates to the technical field of underground powerhouse structures of pumped storage power stations, and in particular to a vibration test and excitation load feedback analysis method for underground powerhouse structures of pumped storage power stations. Background Art

[0002] A pumped storage power station is a special power station that uses the surplus electricity during the low-load period of the power grid to pump water to the upstream reservoir, store the electricity between peak and valley periods in the reservoir, and then release the water in the reservoir to generate electricity during the peak load period of the power grid, thereby realizing the storage and release of peak and valley dispatching electricity in the power system, and at the same time performing regulating power services such as frequency regulation and standby. However, the underground powerhouse of a pumped storage power station is prone to vibration problems such as fatigue cracking and structural damage under various operating modes due to the influence of multiple vibration sources.

[0003] In addition, the dynamic environment of the underground powerhouse of a pumped-storage power station is complex, and mechanical equipment, systems and structures are often affected by multiple vibration sources such as mechanical, electrical and hydraulic. Without proper treatment and control, vibration can cause equipment failure, damage the structural safety of the powerhouse, and reduce the operating efficiency of power equipment. Therefore, how to accurately detect and analyze the vibration characteristics of the powerhouse, know the contribution of vibration sources to structural vibration, and how to formulate and implement effective vibration control measures have become important issues in the current operation and maintenance of pumped-storage power stations.

[0004] Previous research on vibration has mostly focused on the analysis and control of a single vibration source, while ignoring the interaction between multiple vibration sources, making it impossible to accurately and comprehensively analyze the actual vibration conditions on site. This shows that the empirical research method based on a single vibration source cannot meet the analysis needs of the complex vibration phenomena of the underground powerhouse of a pumped-storage power station, and a comprehensive, accurate and practical method for vibration testing and excitation load feedback analysis of the underground powerhouse structure of a pumped-storage power station is needed. Summary of the invention

[0005] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a method for vibration testing and excitation load feedback analysis of the underground powerhouse structure of a pumped-storage power station, so as to solve the technical problems that the prior art cannot accurately judge the contribution of various vibration sources to the vibration of the powerhouse, resulting in the lack of pertinence in the design of vibration control measures, and cannot accurately predict the transformation effect when the vibration-resistant optimization and transformation of the powerhouse structure are carried out.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for vibration testing and excitation load feedback analysis of underground powerhouse structure of a pumped storage power station, which comprises the following steps:

[0007] S1. Combine time domain and frequency domain analysis to extract the vibration characteristics of each structure of the plant;

[0008] S2. Calculate the contribution of three types of vibration sources, namely mechanical, hydraulic and electrical, to structural vibration;

[0009] S3. Use finite element numerical simulation method and combine it with field measured data to conduct unit excitation load feedback analysis.

[0010] Preferably, the step S1 specifically includes the following process:

[0011] S101, sensor arrangement: First, arrange sensors for each plant structure to be tested. The location and number of sensors are selected according to the characteristics, size, shape and expected vibration mode of the structure to be tested;

[0012] S102, data collection: under normal plant conditions or by applying a known excitation signal, collect vibration data of each sensor;

[0013] S103, time domain analysis: perform time domain analysis on the vibration data of each sensor, including calculating the peak value, mean value, and effective value;

[0014] S104, frequency domain analysis uses Fourier transform tools to convert vibration signals into frequency domain and perform frequency domain analysis; calculates spectrum peaks, harmonic components, and frequency domain characteristics of frequency band energy distribution;

[0015] S105. Structural vibration characteristics analysis uses the results of time domain and frequency domain analysis to determine the vibration characteristics of each structure. At the same time, by comparing the vibration characteristics of each structure, the dominant vibration mode is found.

[0016] Preferably, step S2 specifically includes the following process:

[0017] S201, screening characteristic frequency segments: First, through Fourier transform of vibration test data, the spectrum is obtained, and the characteristic frequency segments of each vibration source are found; the frequency corresponding to the rotation speed of the mechanical equipment is the mechanical vibration source, the characteristic frequency of the hydraulic power comes from the working frequency of the water pump equipment, and the electrical vibration source comes from the electromagnetic frequency of the electric equipment;

[0018] S202, calculating the energy contribution of the frequency segments of the vibration source, and then calculating the energy proportion of these characteristic frequency segments in the entire spectrum, that is, in the result after FFT conversion, comparing the energy of each characteristic frequency segment with the total energy to obtain the contribution of the vibration source to the total vibration;

[0019] If X(f) is the result of Fourier transform, f1 and f2 are the lower and upper limits of the characteristic frequency band, respectively, then the energy E corresponding to the frequency band can be expressed as: E = ∑|x(f)| 2 , where f∈[f1,f2], and the total energy can be expressed as: Et=∑|x(f t )| 2, where f t In all frequency ranges, the contribution B of the vibration source is: B = E / Et;

[0020] S203, summarizing the contribution of vibration sources: Calculate the contribution of each vibration source in the above manner, and add up the contributions of each vibration source, which should be close to or equal to 1.

[0021] Preferably, in step S3, the finite element numerical simulation method includes:

[0022] Geometric modeling: Measure the design drawings or physical objects to obtain the geometric shape of the structure to be analyzed, and then create a corresponding three-dimensional or two-dimensional model in the finite element software based on the collected data;

[0023] Meshing: Meshing the geometric model;

[0024] Material property definition: assign corresponding material properties to the model according to the material properties of the actual structure;

[0025] Boundary conditions and load settings: Determine the fixed boundaries in the model and the loads applied to the model. These boundary conditions and loads are set according to the actual working conditions and states;

[0026] In the finite element numerical model creation stage, the transfer learning method was introduced, as follows:

[0027] First, assuming that we already have a trained finite element model Mo, we still create a new finite element model for a new similar model Mn, but when setting boundary conditions, load settings, and meshing, we can use the transfer learning method to partially transfer the information of the Mo model;

[0028] The trained finite element model Mo refers to the existing vibration characteristic model of a specific large-scale plant or production facility. This model is developed based on comprehensive consideration of the operating conditions; and it has been verified to accurately simulate the vibration characteristics of the building or facility;

[0029] The trained finite element model Mo is as follows:

[0030] The geometry and dimensions of the equipment / system: can be obtained directly from design drawings or CAD models, or through field measurements;

[0031] Material properties: density, elastic modulus, Poisson's ratio of the vibration transmission medium can be derived from material properties such as temperature and material;

[0032] Loads and boundary conditions: the specific operating parameters of the mechanical equipment under normal working conditions, as well as the fixing and support conditions of the structure itself;

[0033] Perform a reasonable finite element mesh on all of these data and input all of this information into the finite element analysis software;

[0034] Conduct finite element simulation to obtain the vibration response of the structure, extract and analyze the vibration mode, and thus obtain the finite element model Mo;

[0035] Carry out model transfer learning, as follows:

[0036] Use some components or parameters of the Mo model as initial conditions or reference points to reduce analysis and design time; use the finite element mesh from Mo as the initial mesh for new locations, and then make necessary adjustments and optimizations;

[0037] For the migration of material properties, a weight parameter is used for adjustment; the material property Eni of Mn and the material property Eoi of Mo, assuming that their relationship is a linear relationship, Eni = w*Eoi, where w is the weight to be learned;

[0038] Similarly, other boundary conditions and load settings can also be migrated by finding the mapping relationship between them. It can be considered that there is an f relationship between them, expressed as: Bni=f(Boi), Lni=f(Loi), where Bni, Lni are the boundary conditions and loads of model Mn, and Boi, Loi are the boundary conditions and loads of model Mo; the acquisition of the f function can be achieved by observing the data distribution or fitting it using the neural network method;

[0039] Finally, when performing finite element simulation calculations on the new similar model Mn, the migrated boundary conditions, loads, and material properties are directly used, saving model creation and analysis time;

[0040] The numerical feedback analysis of the unit excitation load uses a finite element model to perform modal analysis on the unit to obtain the free vibration frequency and vibration modes of the structure; the finite element model is also used to apply a resonant load corresponding to the vibration frequency, and a harmonic response analysis is performed to obtain the response amplitude of the structure; the amplitude result is compared with the measured value to verify the accuracy of the model, and vibration control and optimization are performed based on the calculated results.

[0041] Preferably, in step S3, the specific process of performing the unit excitation load feedback analysis in combination with the field measured data is as follows:

[0042] S301. Use a finite element model to perform modal analysis on the unit to obtain the free vibration frequency and vibration modes of the structure;

[0043] S302, calculating the exciting frequency of the unit;

[0044] The amplitude is calculated using the finite element method and the solution is obtained using the harmonic response method;

[0045] The vibration frequency caused by the rotating parts of the unit;

[0046]

[0047] Where n is the rated speed of the generator, r / min; np is the runaway speed, r / min; m is 3 for low-frequency units and 2 for high-frequency units;

[0048] Vibration frequency caused by hydraulic shock:

[0049]

[0050] Where: x1, x2 are the number of guide vane blades and runner blades; a is the greatest common divisor of x1 and x2, ω is the circular frequency;

[0051] The solution process uses simple harmonic wave to apply the load to each action position; the equation is as follows:

[0052]

[0053] {u(t)}={u(t) max e iθ}e iΩt ;

[0054] Where: F(t) is the excitation load; u(t) is the displacement; F(t) max is the amplitude of the excitation load; is the phase angle of the exciting load; u(t) max is the displacement amplitude; θ is the displacement phase angle; Ω is the circular frequency of the excitation load; t represents time; i represents the imaginary unit; the frequencies of the excitation load and the response displacement are the same during calculation;

[0055] S303, also using the finite element model, using the formula Apply the resonant load corresponding to the vibration frequency and perform harmonic response analysis to obtain the response amplitude of the structure;

[0056] S304, compare the amplitude results with the measured values. This step not only verifies the accuracy of the model, but also realizes adaptive adjustment of vibration control parameters based on the comparison results, analyzes the influence of the three types of vibration sources at present, and can infer future vibration trends based on historical data. At the same time, dynamic adjustments are made based on the real-time operating status to improve the real-time performance and efficiency of structural vibration control.

[0057] First, the spectrum analysis is performed using the step-by-step Fourier transform to obtain the spectrum analysis results at different times, find the sensitive frequency points and frequency bands, and calculate the real-time vibration energy Es, that is, the real-time vibration intensity; the formula is as follows:

[0058] Es=∑|Xi(f)| 2 ;

[0059] f represents frequency, Xi is the STFT result;

[0060] Through multivariate linear regression modeling, the real-time excitation intensity of each vibration source is reversely derived, that is, the contribution of mechanical, hydraulic, and electrical to the real-time vibration energy Es, which is taken as El, Ew, and Ee; the specific formula is as follows:

[0061] Es=aEl+bEw+c*Ee;

[0062] Among them, a, b, c represent the correlation coefficients between mechanical, hydraulic, electrical excitation loads and vibration real-time energy, respectively. El, Ew, Ee can be solved in reverse.

[0063] Beneficial effects of the present invention: The present invention extracts the vibration characteristics of each structure of the plant, calculates the contribution of mechanical, hydraulic, electrical and other vibration sources to the structural vibration, and uses finite element numerical simulation technology to perform feedback analysis of the unit excitation load, and feeds back effective vibration control measures to the model. After multiple iterative optimizations, the best vibration control solution can be expected; this method can not only accurately reflect the dynamic characteristics of the plant structure under actual working conditions, but also find out the contribution of each vibration source to the total vibration, and formulate the most appropriate vibration control measures for different vibration sources; therefore, it is conducive to optimizing the design and operation of the underground plant of the pumped storage power station, improving the safety and stability of the power station, thereby reducing maintenance costs and increasing the economic benefits of the power station; it solves the technical problems that the prior art cannot accurately judge the contribution of various vibration sources to the plant vibration, resulting in the lack of pertinence in the design of vibration control measures, and cannot accurately predict the transformation effect when the plant structure is optimized and transformed for vibration resistance. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 The present invention is a flow chart of a vibration test and excitation load feedback analysis method for an underground powerhouse structure of a pumped storage power station. DETAILED DESCRIPTION

[0065] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0066] like Figure 1 As shown, a method for vibration testing and excitation load feedback analysis of underground powerhouse structure of a pumped storage power station comprises the following steps:

[0067] S1. Combine time domain and frequency domain analysis to extract the vibration characteristics of each structure of the plant;

[0068] S2. Calculate the contribution of three types of vibration sources, namely mechanical, hydraulic and electrical, to structural vibration;

[0069] S3. Use finite element numerical simulation method and combine it with field measured data to conduct unit excitation load feedback analysis.

[0070] Preferably, the step S1 specifically includes the following process:

[0071] S101, sensor arrangement: First, arrange sensors for each plant structure to be tested (such as floor, machine pier wind shield, stairs and columns). The location and number of sensors are selected according to the characteristics, size, shape and expected vibration mode of the structure to be tested;

[0072] S102, data collection: under normal plant conditions or by applying a known excitation signal, collect vibration data of each sensor;

[0073] S103, time domain analysis: perform time domain analysis on the vibration data of each sensor, including calculating the peak value (AbsMax=max│x(t)│), mean value (Mean=1 / N*∑x(t)), effective value

[0074] S104, frequency domain analysis uses Fourier transform tools to convert vibration signals into frequency domain and perform frequency domain analysis; calculates spectrum peaks, harmonic components, and frequency domain characteristics of frequency band energy distribution;

[0075] S105. Structural vibration characteristics analysis uses the results of time domain and frequency domain analysis to determine the vibration characteristics of each structure, such as main frequency, vibration amplitude, harmonic components, etc. At the same time, by comparing the vibration characteristics of each structure, the dominant vibration mode is found.

[0076] Preferably, step S2 specifically includes the following process:

[0077] S201, screening characteristic frequency segments: First, through Fourier transform of vibration test data, the spectrum is obtained, and the characteristic frequency segments of each vibration source are found; the frequency corresponding to the rotation speed of the mechanical equipment is the mechanical vibration source, the characteristic frequency of the hydraulic power comes from the working frequency of the water pump equipment, and the electrical vibration source comes from the electromagnetic frequency of the electric equipment;

[0078] S202, calculating the energy contribution of the frequency segments of the vibration source, and then calculating the energy proportion of these characteristic frequency segments in the entire spectrum, that is, in the result after FFT conversion, comparing the energy of each characteristic frequency segment with the total energy to obtain the contribution of the vibration source to the total vibration;

[0079] If X(f) is the result of Fourier transform, f1 and f2 are the lower and upper limits of the characteristic frequency band, respectively, then the energy E corresponding to the frequency band can be expressed as: E = ∑|x(f)| 2, where f∈[f1,f2], and the total energy can be expressed as: Et=∑|x(f t )| 2 , where f t In all frequency ranges, the contribution B of the vibration source is: B = E / Et;

[0080] S203. Summarize the contribution of vibration sources: Calculate the contribution of each vibration source in the above manner, and add up the contributions of each vibration source, which should be close to or equal to 1 (there may be some slight errors caused by testing and processing errors).

[0081] Preferably, in step S3, the finite element numerical simulation method includes:

[0082] Geometric modeling: Measure the design drawings or physical objects to obtain the geometric shape of the structure to be analyzed, and then create a corresponding three-dimensional or two-dimensional model in the finite element software based on the collected data;

[0083] Meshing: Meshing the geometric model; the mesh needs to fit the actual structure as closely as possible to simulate the actual situation more accurately. At the same time, attention should be paid to the density of the mesh. A mesh that is too coarse may result in inaccurate results, while a mesh that is too fine may result in a long calculation time.

[0084] Material property definition: According to the material properties of the actual structure, the corresponding material properties (such as elastic modulus, Poisson's ratio, etc.) are assigned to the model;

[0085] Boundary conditions and load settings: Determine the fixed boundaries in the model and the loads applied to the model. These boundary conditions and loads are set according to the actual working conditions and states;

[0086] During the finite element numerical model creation stage, the transfer learning method was introduced to enable analysis and design of other similar structures based on one model training, greatly saving the time for model creation and analysis.

[0087] If there is an existing finite element model of a similar structure, we can use the concept of transfer learning to quickly create a new model by modifying the parameters of the existing model. This not only saves time in model creation and analysis, but also allows us to quickly obtain preliminary analysis results based on the experience of the existing model, thus improving work efficiency. The details are as follows:

[0088] First, assuming that we already have a trained finite element model Mo, we still create a new finite element model for a new similar model Mn, but when setting boundary conditions, load settings, and meshing, we can use the transfer learning method to partially transfer the information of the Mo model;

[0089] The trained finite element model Mo refers to the existing vibration characteristic model of a specific large-scale plant or production facility. This model is developed by comprehensive consideration of the operating conditions (such as equipment type, equipment layout and operating parameters); and it has been verified to accurately simulate the vibration characteristics of the building or facility;

[0090] The trained finite element model Mo is as follows:

[0091] The geometry and dimensions of the equipment / system: can be obtained directly from design drawings or CAD models, or through field measurements;

[0092] Material properties: density, elastic modulus, Poisson's ratio of the vibration transmission medium can be derived from material properties such as temperature and material;

[0093] Loads and boundary conditions: the specific operating parameters of the mechanical equipment under normal working conditions (such as rotation speed, electrical load, etc.), as well as the fixing and support conditions of the structure itself;

[0094] Perform reasonable finite element meshing on all these data and input all this information into finite element analysis software (such as ANSYS, ABAQUS, etc.);

[0095] Conduct finite element simulation to obtain the vibration response of the structure, extract and analyze the vibration mode, and thus obtain the finite element model Mo;

[0096] Carry out model transfer learning, as follows:

[0097] Utilize some components or parameters of the Mo model as initial conditions or reference points to reduce analysis and design time; use the finite element mesh from Mo as the initial mesh for new locations, and then make necessary adjustments and optimizations; in this way, the development time of the new model MN can be significantly reduced while ensuring the accuracy and reliability of its results.

[0098] For the migration of material properties, a weight parameter is used for adjustment; the material property Eni of Mn and the material property Eoi of Mo, assuming that their relationship is a linear relationship, Eni = w*Eoi, where w is the weight to be learned;

[0099] Similarly, other boundary conditions and load settings can also be migrated by finding the mapping relationship between them. It can be considered that there is an f relationship between them, expressed as: Bni=f(Boi), Lni=f(Loi), where Bni, Lni are the boundary conditions and loads of model Mn, and Boi, Loi are the boundary conditions and loads of model Mo; the acquisition of the f function can be achieved by observing the data distribution or fitting it using the neural network method;

[0100] Finally, when performing finite element simulation calculations on a new similar model Mn, the migrated boundary conditions, loads, and material properties are directly used, which greatly saves model creation and analysis time;

[0101] The numerical feedback analysis of the unit excitation load uses a finite element model to perform modal analysis on the unit to obtain the free vibration frequency and vibration modes of the structure; the finite element model is also used to apply a resonant load corresponding to the vibration frequency, and a harmonic response analysis is performed to obtain the response amplitude of the structure; the amplitude result is compared with the measured value to verify the accuracy of the model, and vibration control and optimization are performed based on the calculated results.

[0102] Preferably, in step S3, the specific process of performing the unit excitation load feedback analysis in combination with the field measured data is as follows:

[0103] S301. Use a finite element model to perform modal analysis on the unit to obtain the free vibration frequency and vibration modes of the structure;

[0104] S302, calculating the exciting frequency of the unit;

[0105] The amplitude is calculated using the finite element method and the solution is obtained using the harmonic response method;

[0106] The vibration frequency caused by the rotating parts of the unit;

[0107]

[0108] Where n is the rated speed of the generator, r / min; np is the runaway speed, r / min; m is 3 for low-frequency units and 2 for high-frequency units;

[0109] Vibration frequency caused by hydraulic shock:

[0110]

[0111] Where: x1, x2 are the number of guide vane blades and runner blades; a is the greatest common divisor of x1 and x2, ω is the circular frequency;

[0112] The solution process uses simple harmonic wave to apply the load to each action position; the equation is as follows:

[0113]

[0114] {u(t)}={u(t) max e iθ}e iΩt ;

[0115] Where: F(t) is the excitation load; u(t) is the displacement; F(t) maxis the amplitude of the excitation load; is the phase angle of the exciting load; u(t) max is the displacement amplitude; θ is the displacement phase angle; Ω is the circular frequency of the excitation load; t represents time; i represents the imaginary unit; the frequencies of the excitation load and the response displacement are the same during calculation;

[0116] S303, also using the finite element model, using the formula Apply the resonant load corresponding to the vibration frequency and perform harmonic response analysis to obtain the response amplitude of the structure;

[0117] S304, compare the amplitude results with the measured values. This step not only verifies the accuracy of the model, but also realizes adaptive adjustment of vibration control parameters based on the comparison results, analyzes the influence of the three types of vibration sources at present, and can infer future vibration trends based on historical data. At the same time, dynamic adjustments are made based on the real-time operating status to improve the real-time performance and efficiency of structural vibration control.

[0118] First, the spectrum analysis is performed using the step-by-step Fourier transform (STFT) to obtain the spectrum analysis results at different times, find the sensitive frequency points and frequency bands, and calculate the real-time vibration energy Es, that is, the real-time vibration intensity; the formula is as follows:

[0119] Es=∑|Xi(f)| 2 ;

[0120] f represents frequency, Xi is the STFT result;

[0121] Through multivariate linear regression modeling, the real-time excitation intensity of each vibration source is reversely derived, that is, the contribution of mechanical, hydraulic, and electrical to the real-time vibration energy Es, which is taken as El, Ew, and Ee; the specific formula is as follows:

[0122] Es=aEl+bEw+c*Ee;

[0123] Among them, a, b, c represent the correlation coefficients between mechanical, hydraulic, electrical excitation loads and vibration real-time energy, respectively. El, Ew, Ee can be solved in reverse.

[0124] The embodiments of the present invention have the following beneficial effects:

[0125] 1. Fusion of actual scenarios: This method integrates field measured data with finite element numerical simulation technology, and combines theoretical data with actual conditions through actual scenarios, making the analysis results more in line with the actual situation and reducing the deviation between theoretical analysis and actual conditions.

[0126] 2. Calculation of vibration source contribution: Through detailed calculation and analysis of the three main vibration sources of the underground plant (mechanical, hydraulic, and electrical), find out the vibration source that contributes most to the structural vibration.

[0127] 3. Comprehensive vibration control: The method is not just aimed at simply conducting vibration testing and analysis, but integrates a series of work processes such as vibration testing, analysis, and vibration control optimization design to provide full-service for the operation of pumped storage power stations.

[0128] 4. Introduction of transfer learning: In the finite element numerical model creation stage, the concept of transfer learning was introduced, which enabled the analysis and design of other similar structures based on one model training, greatly saving the time of model creation and analysis.

[0129] 5. Adaptive feedback adjustment: Adaptive adjustment of vibration control parameters is realized. Not only can the parameters be adjusted according to historical data analysis, but also dynamic adjustment can be made according to real-time operating status, which greatly improves the real-time performance and efficiency of structural vibration control.

[0130] Embodiment 2:

[0131] First, a systematic vibration test was conducted on an underground powerhouse. Various types of sensors, including acceleration sensors, displacement sensors, and velocity sensors, were placed on each key structure in the powerhouse, including the floor, machine pier wind shield, stairs, and columns. The final number and location of the sensors were determined based on the characteristics, size, shape, and expected vibration of the structure to be tested. Here, no less than 5 effective sensors were placed on each structure.

[0132] During the normal operation of the plant, the vibration data of the sensor was collected by data acquisition equipment, and then time domain analysis and frequency domain analysis were performed. Considering the large amount of data and the wide distribution of data, the time domain data was statistically analyzed using methods including peak value, mean value, and effective value. At the same time, Fourier transform was used for frequency domain conversion, and the spectrum peak, harmonic components, and frequency domain characteristics of frequency band energy distribution were recorded.

[0133] Again, for the three basic vibration sources of mechanical, hydraulic, and electrical, first find out their characteristic frequency segments in the spectrum through Fourier transform. For example, the main frequency of the mechanical vibration source in the sample plant is below 30Hz, the main frequency of the hydraulic vibration source is 128Hz and its multiples, and the electrical vibration source is 50Hz and its multiples. Then calculate the energy proportion of these characteristic frequency segments in the overall spectrum, and obtain the contribution of the three types of vibration sources to the total vibration. It is preliminarily concluded that the contribution ratio of mechanical vibration sources is about 10%, the contribution ratio of hydraulic vibration sources is about 70%, and the contribution ratio of electrical vibration sources is about 20%.

[0134] Again, based on the field measured data, a finite element model was constructed. Taking the pier wind shield as an example, its geometric dimensions are: bottom area 8m×8m, height 12m. The material selected is C30 concrete, with an elastic modulus of 35GPa and a Poisson's ratio of 0.2. The structure was meshed. Similarly, key parts such as floor slabs, stairs and columns were also processed similarly. The total finite element model contains nearly 250,000 nodes and 270,000 finite elements.

[0135] Various types of boundary conditions such as measured displacement, velocity and acceleration were applied to each node, and three types of measured loads were also applied: mechanical, hydraulic and electrical. Taking mechanical load as an example, considering that the vibration frequency of the wind shield is below 30Hz, a sinusoidal load with an amplitude of 5μm was applied.

[0136] Finite element analysis was performed using the explicit integration method with a time step of 0.01s, and basic mechanical responses such as regional strain and stress under various operating conditions such as cracking and failure were obtained. Finally, by experimentally adjusting the vibration control parameters one by one in the model, the prediction of passive vibration control effect based on the finite element numerical model was realized, thus providing a theoretical basis and technical guidance for the implementation of vibration control on site, and significantly improving the actual operation effect.

[0137] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A method for vibration testing and excitation load feedback analysis of underground powerhouse structure of a pumped storage power station, characterized in that: It includes the following steps: S1. Combine time domain and frequency domain analysis to extract the vibration characteristics of each structure of the plant; S2. Calculate the contribution of three types of vibration sources, namely mechanical, hydraulic and electrical, to structural vibration; S3. Use finite element numerical simulation method and combine it with field measured data to conduct unit vibration load feedback analysis.

2. A method for vibration testing and excitation load feedback analysis of underground powerhouse structure of a pumped storage power station according to claim 1, characterized in that: The step S1 specifically The process includes: S101, sensor arrangement: First, arrange sensors for each plant structure to be tested. The location and number of sensors are selected according to the characteristics, size, shape and expected vibration mode of the structure to be tested; S102, data collection: under normal plant conditions or by applying a known excitation signal, collect vibration data of each sensor; S103, time domain analysis: perform time domain analysis on the vibration data of each sensor, including calculating the peak value, mean value, and effective value; S104, frequency domain analysis uses Fourier transform tools to perform frequency domain conversion on the vibration signal and perform frequency domain analysis; Calculate the spectrum peak, harmonic components and frequency domain characteristics of band energy distribution; S105. Structural vibration characteristics analysis uses the results of time domain and frequency domain analysis to determine the vibration characteristics of each structure. At the same time, by comparing the vibration characteristics of each structure, the dominant vibration mode is found.

3. The method for vibration testing and excitation load feedback analysis of underground powerhouse structure of a pumped storage power station according to claim 1, characterized in that: The step S2 specifically includes the following process: S201, screening characteristic frequency bands: first, obtaining a frequency spectrum through Fourier transform of the vibration test data, and finding the characteristic frequency bands of each vibration source; The frequency corresponding to the rotation speed of the mechanical equipment is the mechanical vibration source, the characteristic frequency of the hydraulic force comes from the working frequency of the water pump equipment, and the electrical vibration source comes from the electromagnetic frequency of the electric equipment; S202, calculating the energy contribution of the frequency segments of the vibration source, and then calculating the energy proportion of these characteristic frequency segments in the entire spectrum, that is, in the result after FFT conversion, comparing the energy of each characteristic frequency segment with the total energy to obtain the contribution of the vibration source to the total vibration; If X(f) is the result of Fourier transform, f1 and f2 are the lower and upper limits of the characteristic frequency band, respectively, then the energy E corresponding to the frequency band can be expressed as: E = Σ|x(f)| 2 , where f∈[f1,f2], and the total energy can be expressed as: Et=Σ|x(f t )| 2 , where f t In all frequency ranges, the contribution B of the vibration source is: B = E / Et; S203, summarizing the contribution of vibration sources: Calculate the contribution of each vibration source in the above manner, and add up the contributions of each vibration source, which should be close to or equal to 1.

4. The method for vibration testing and excitation load feedback analysis of underground powerhouse structure of a pumped storage power station according to claim 1, characterized in that: In step S3, the finite element numerical simulation method includes: Geometric modeling: Measure the design drawings or physical objects to obtain the geometric shape of the structure to be analyzed, and then create a corresponding three-dimensional or two-dimensional model in the finite element software based on the collected data; Meshing: Meshing the geometric model; Material property definition: assign corresponding material properties to the model according to the material properties of the actual structure; Boundary conditions and load settings: Determine the fixed boundaries in the model and the loads applied to the model. These boundary conditions and loads are set according to the actual working conditions and states; In the finite element numerical model creation stage, the transfer learning method was introduced, as follows: First, assuming that we already have a trained finite element model Mo, we still create a new finite element model for a new similar model Mn, but when setting boundary conditions, load settings, and meshing, we can use the transfer learning method to partially transfer the information of the Mo model; The trained finite element model Mo refers to the existing vibration characteristic model of a specific large-scale plant or production facility. This model is developed based on comprehensive consideration of the operating conditions; and it has been verified to accurately simulate the vibration characteristics of the building or facility; The trained finite element model Mo is as follows: The geometry and dimensions of the equipment / system: can be obtained directly from design drawings or CAD models, or through field measurements; Material properties: density, elastic modulus, Poisson's ratio of the vibration transmission medium can be derived from material properties such as temperature and material; Loads and boundary conditions: the specific operating parameters of the mechanical equipment under normal working conditions, as well as the fixing and support conditions of the structure itself; Perform a reasonable finite element mesh on all of these data and input all of this information into the finite element analysis software; Conduct finite element simulation to obtain the vibration response of the structure, extract and analyze the vibration mode, and thus obtain the finite element model Mo; Carry out model transfer learning, as follows: Use some components or parameters of the Mo model as initial conditions or reference points to reduce analysis and design time; use the finite element mesh from Mo as the initial mesh for new locations, and then make necessary adjustments and optimizations; For the migration of material properties, a weight parameter is used for adjustment; the material property Eni of Mn and the material property Eoi of Mo, assuming that their relationship is a linear relationship, Eni = w*Eoi, where w is the weight to be learned; Similarly, other boundary conditions and load settings can also be migrated by finding the mapping relationship between them. It can be considered that there is an f relationship between them, expressed as: Bni=f(Boi), Lni=f(Loi), where Bni, Lni are the boundary conditions and loads of model Mn, and Boi, Loi are the boundary conditions and loads of model Mo; the acquisition of the f function can be achieved by observing the data distribution or fitting it using the neural network method; Finally, when performing finite element simulation calculations on the new similar model Mn, the migrated boundary conditions, loads, and material properties are directly used, saving model creation and analysis time; The numerical feedback analysis of the unit excitation load uses a finite element model to perform modal analysis on the unit to obtain the free vibration frequency and vibration modes of the structure; the finite element model is also used to apply a resonant load corresponding to the vibration frequency, and a harmonic response analysis is performed to obtain the response amplitude of the structure; the amplitude result is compared with the measured value to verify the accuracy of the model, and vibration control and optimization are performed based on the calculated results.

5. A method for vibration testing and excitation load feedback analysis of underground powerhouse structure of a pumped storage power station according to claim 4, characterized in that: In step S3, the specific process of performing the unit excitation load feedback analysis in combination with the on-site measured data is as follows: S301. Use a finite element model to perform modal analysis on the unit to obtain the free vibration frequency and vibration modes of the structure; S302, calculating the exciting frequency of the unit; The amplitude is calculated using the finite element method and the solution is obtained using the harmonic response method; The vibration frequency caused by the rotating parts of the unit; Where n is the rated speed of the generator, r / min; np is the runaway speed, r / min; m is 3 for low-frequency units and 2 for high-frequency units; Vibration frequency caused by hydraulic shock: Where: x1, x2 are the number of guide vane blades and runner blades; a is the greatest common divisor of x1 and x2, ω is the circular frequency; The solution process uses simple harmonic wave to apply the load to each action position; the equation is as follows: {u(t)}={u(t) max e iθ }e iΩt ; Where: F(t) is the excitation load; u(t) is the displacement; F(t) max is the amplitude of the excitation load; is the phase angle of the exciting load; u(t) max is the displacement amplitude; θ is the displacement phase angle; Ω is the circular frequency of the excitation load; t represents time; i represents the imaginary unit; the frequencies of the excitation load and the response displacement are the same during calculation; S303, also using the finite element model, using the formula Apply the resonant load corresponding to the vibration frequency and perform harmonic response analysis to obtain the response amplitude of the structure; S304, compare the amplitude results with the measured values. This step not only verifies the accuracy of the model, but also realizes adaptive adjustment of vibration control parameters based on the comparison results, analyzes the influence of the three types of vibration sources at present, and can infer future vibration trends based on historical data. At the same time, dynamic adjustments are made based on the real-time operating status to improve the real-time performance and efficiency of structural vibration control. First, the spectrum analysis is performed using the step-by-step Fourier transform to obtain the spectrum analysis results at different times, find the sensitive frequency points and frequency bands, and calculate the real-time vibration energy Es, that is, the real-time vibration intensity; the formula is as follows: It is=∑|Xi(f)| 2 ; f represents frequency, Xi is the STFT result; Through multivariate linear regression modeling, the real-time excitation intensity of each vibration source is reversely derived, that is, the contribution of mechanical, hydraulic, and electrical to the real-time vibration energy Es, which is taken as El, Ew, and Ee; the specific formula is as follows: Es=aEl+bEw+c*Ee; Among them, a, b, c represent the correlation coefficients between mechanical, hydraulic, electrical excitation loads and vibration real-time energy, respectively. El, Ew, Ee can be solved in reverse.

Images