Fluid cavitation state identification method based on variational mode self-optimization reconstruction

Abstract

The invention discloses a fluid cavitation state identification method based on variational mode self-optimization reconstruction. The method comprises the following steps of: 1, based on development of a VMD self-optimization reconstruction method, decomposing an original fluid pressure signal directly acquired by an experiment into a high-quality signal decomposition result; and 2, based on the high-quality signal decomposition result obtained in the step 1, analyzing pressure pulsation signals of a plurality of measuring points by adopting a pore plate cavitation experiment, and realizing accurate identification of different cavitation stages by comparing changes of main frequency components of the measuring points under the same working condition. According to the method, the transient pressure signal is converted into a main frequency component, and the cavitation intensity in the complex flow working condition is judged through the distribution change of the frequency component.

Description

Technical Field

[0001] The present invention relates to the technical field of cavitation identification, and particularly to a method for identifying the cavitation state of fluid by variational mode self-optimizing reconstruction. Background Art

[0002] Cavitation is a common hydrodynamic phenomenon that usually occurs when the liquid flows at high speed or the pressure fluctuates sharply. When the local pressure is lower than the saturated vapor pressure of the liquid, vaporization occurs, forming bubbles or cavities. These bubbles will impact the surrounding solid surfaces during the generation and rupture processes, thereby causing adverse effects such as vibration and noise. Although cavitation phenomena under normal temperature and pressure have been widely studied, in extreme or complex working conditions (such as in nuclear power plant cooling circuits, high-pressure steam systems, and other harsh environments), the cavitation behavior of water is more complex, and there are obvious differences in its formation conditions and control methods. Therefore, accurately detecting the cavitation state under different working conditions is of great significance for ensuring the safe and stable operation of the system. In recent years, with the development of multi-sensor fusion and intelligent signal processing technologies, significant progress has been made in cavitation state identification technology, but existing methods still have certain limitations.

[0003] For example, the cavitation identification technology based on deep learning image processing with the publication number CN116563219A can visually present the cavitation state, but due to the complex equipment and high computational cost, it is difficult to meet the requirements of real-time identification; while the cavitation identification method based on ultrasonic waves is easily affected by the medium characteristics, environmental noise, and arrangement methods, resulting in insufficient signal acquisition stability; the analysis based on vibration signals is also easily affected by external interference in a high-noise environment, affecting the accurate judgment of the cavitation state. In addition, although the cavitation state can be identified through pressure signals, due to the complexity of the original pressure signals themselves, it is difficult to accurately extract cavitation characteristics by direct analysis. Summary of the Invention

[0004] To overcome the above defects existing in the prior art, the present invention provides a method for identifying the cavitation state of fluid by variational mode self-optimizing reconstruction. This method converts the transient pressure signal into main frequency components and judges the cavitation intensity in complex flow conditions through the distribution change of the frequency components.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A method for identifying the cavitation state of fluid by variational mode self-optimizing reconstruction, comprising the following steps;

[0007] Step 1: Developed based on the VMD self-optimizing reconstruction method (SOVR, Self-Optimizing VMD Reconstruction), decompose the original fluid pressure signal (complex pressure pulsation signal) directly collected in the experiment into a high-quality signal decomposition result;

[0008] The high-quality signal decomposition result is the intrinsic mode functions (IMFs) that reflect the characteristics of fluid flow, and the intrinsic mode functions contain spectral information closely related to the cavitation state;

[0009] Step 2: Based on the high-quality signal decomposition result obtained in Step 1, an orifice plate cavitation experiment is adopted to analyze the pressure pulsation signals at multiple measuring points. By comparing the changes in the main frequency components of each measuring point under the same working conditions, the accurate identification of different cavitation stages is realized.

[0010] Apply the VMD self-optimizing reconstruction method to the cavitation state detection of high-temperature and high-pressure working fluids.

[0011] In the above-mentioned Step 1, the VMD self-optimizing reconstruction method realizes the automatic and reasonable selection of the hyperparameters (decomposition layer number, quadratic penalty factor, and update rate) required for VMD decomposition, and through noise reduction and reconstruction, effective IMFs are obtained, spectral calculation is carried out, and the decomposition process of the input original signal is completed;

[0012] Its calculation process:

[0013] Step 1: Determine the VMD decomposition layer number by the central frequency method;

[0014] The specific method is that during the process of gradually increasing the decomposition layer number, calculate the change of the central frequency of the IMF component (intrinsic mode function component) with the largest energy proportion with the layer number; when the relative error of the central frequency in two adjacent decompositions is less than 5%, the previous decomposition layer number is considered the optimal layer number;

[0015] Step 2: Optimize the quadratic penalty factor and update rate through the genetic algorithm:

[0016] Through the genetic operations of selection, crossover, and mutation, use the fitness function composed of the reconstruction error between a group of intrinsic mode functions IMFs and the input original fluid pressure signal (i.e., the signal to be processed) to find the optimal parameter combination; that is, obtain the optimal combination of the quadratic penalty factor and update rate through the genetic algorithm to realize the reasonable selection of VMD decomposition parameters;

[0017] The genetic algorithm parameters are set as the population size of 20, the selection method is roulette wheel selection, the crossover method is single-point crossover, and the mutation method is uniform mutation.

[0018] Step 3: Distinguish the effective components and noise components by calculating the power spectrum entropy, and perform denoising and reconstruction on the IMFs:

[0019] The original fluid pressure signal is decomposed by VMD using the decomposition parameters (number of decomposition layers, quadratic penalty factor, and update rate) obtained in Step 1 and Step 2 to obtain preliminary IMFs. Calculate the power spectrum entropy and relative energy of the IMF components, sort them in ascending order of power spectrum entropy, and then accumulate the relative energy of the IMF components in sequence until the cumulative sum exceeds 0.95 for the first time when adding to the i-th component. At this time, retain the first i IMF components as the final effective IMFs;

[0020] Step 4: Calculate the frequency spectrum of each reconstructed IMF component to obtain the center frequency of each IMF component, complete the entire VMD self-optimizing reconstruction process, and the obtained IMFs are the final decomposition results of the pressure signal, thereby realizing the processing of the input signal.

[0021] In the second step:

[0022] The orifice cavitation experimental component consists of an upstream section, an orifice plate, and a downstream section;

[0023] The upstream section is used as a control reference, and a pressure measuring point is set to collect the pressure pulsation information when cavitation does not occur under the current temperature, pressure, and flow rate conditions;

[0024] Three pressure measuring points are set in the downstream section to capture the pressure pulsation characteristics of the fluid after orifice cavitation, and the cavitation stage is judged based on the pressure pulsation signals at different positions.

[0025] Among them, the diameters of the upstream and downstream sections of the orifice plate shape are both D, the length of the upstream section is 10D, and the length of the downstream section is 30D; the orifice diameter of the orifice plate is d, and the length is l;

[0026] Pressure transmitters are installed at both the inlet and outlet of the orifice experimental component to measure the inlet pressure (P in ) and the outlet pressure (P out ); At the same time, transient pressure sensors are arranged at 5D upstream of the orifice plate and at 2.5D, 7.5D, and 15D downstream, respectively, and are marked as P1, P2, P3, and P4 to monitor the fluid pressure pulsation signal;

[0027] The transient pressure sensor uses high-temperature pressure sensors (Rosemount3051 and Kistler601) to measure the fluid pressure;

[0028] Among them, the range of Kistler601 is 0~2Mpa, the sampling frequency is 20000Hz, and the uncertainty is 0.1%; the range of Rosemount3051 is 0~12Mpa, the sampling frequency is 1Hz, and the uncertainty is 0.22%;

[0029] The transient pressure sensor is connected to the experimental section through a base designed as a double saddle shape, aiming to minimize the influence of the opening on the flow field as much as possible.

[0030] In this orifice cavitation experiment, high-temperature and high-pressure deionized water flows through the experimental section with an orifice at a constant flow rate, so that pressure pulsation signals are collected at multiple measurement points.

[0031] According to the VMD self-optimizing reconstruction method, the measured pressure pulsation signals are decomposed. The X-axis is the cavitation number, and the Y-axis is the center frequency corresponding to each IMF component. The dimensionless cavitation number is used to characterize the cavitation intensity in the pipe. The smaller the cavitation number, the greater the cavitation intensity, and its definition is as follows:

[0032]

[0033] In the formula, P v is the saturated vapor pressure corresponding to the working fluid inlet temperature, P in is the inlet pressure of the experimental section, and P out is the outlet pressure of the experimental section.

[0034] At the same time, the relative energy ratio RE is used to describe the contribution degree of each IMF component in the original signal. Its calculation formula is as follows. In the formula, E i is the energy of each IMF component, and E total is the total energy of the reconstructed signal. The mode with a relatively high energy indicates that it occupies a larger weight in the original signal;

[0035]

[0036] After decomposing the pressure pulsation signals at each measurement point collected in the experiment by using the VMD self-optimizing reconstruction method, a series of IMF components will be obtained for each measurement point, and the number of these components will vary with the change of working conditions. At the same time, each IMF component corresponds to a specific center frequency.

[0037] Under different cavitation states, the number of IMF components decomposed at each measurement point and the corresponding center frequency distribution show obvious differences, and these differences are used to judge the incipience of cavitation and distinguish different cavitation states.

[0038] When cavitation does not occur, the frequency spectrum at each measurement point only shows a main frequency component. When the cavitation number is large, only one IMF component is detected at all four measurement points, and this stage corresponds to the non-cavitation state;

[0039] When cavitation just occurs, a new frequency component will appear in the frequency spectrum, and the P2 measurement point first shows an additional IMF component, marking the onset of cavitation;

[0040] When the cavitation intensity increases, although the total number of frequency components in the spectrum remains the same as in the initial stage of cavitation, two IMF components have appeared at P2, while there is still only one at P3 and P4, indicating that cavitation has entered the development stage;

[0041] When the cavitation degree is extremely strong, more frequency components will appear in the spectrum. High-frequency IMF components are detected at the P2, P3, and P4 measuring points downstream of the orifice plate, and each measuring point contains two IMF components. This stage is the severe cavitation stage.

[0042] Advantages of the present invention:

[0043] 1. Accurate identification of cavitation state. The adopted VMD self-optimizing reconstruction method can efficiently separate and extract key information from complex original transient pressure pulsation signals. Based on the change in the distribution of frequency components, this method not only realizes the accurate judgment of the cavitation state but also can capture the subtle changes during the inception and development of cavitation, thus greatly improving the detection accuracy and reliability.

[0044] 2. Excellent anti-interference ability and high stability. The present invention can still maintain excellent detection effects under complex working conditions, especially in the environment of high-temperature and high-pressure working fluids. Under high-temperature and high-pressure conditions, the pressure acquisition technology is mature and sensors are widely used, and high-quality pressure pulsation signals can be stably obtained in extreme working conditions. Since the cavitation state identification of the present invention only depends on this signal, the applicability of the method can be ensured even in harsh environments. At the same time, the noise reduction reconstruction algorithm effectively suppresses interference signals, ensuring the accurate identification of frequency components at each measuring point, thus realizing the long-term stable operation and high-precision detection of the system.

[0045] 3. Simple structure, easy to implement and popularize. This detection system only depends on standard devices such as conventional pressure sensors and does not require expensive or complex instrument support. The system design is simple, easy to integrate and maintain, which is conducive to rapid deployment and wide range promotion in various practical projects, further reducing the application cost. Description of the drawings

[0046] Figure 1 It is the flow chart of the cavitation state detection method based on variational mode self-optimizing reconstruction.

[0047] Figure 2 It is the result comparison diagram of the simulation signal verifying the SOVR algorithm.

[0048] Figure 3 It is the schematic diagram of the experimental section.

[0049] Figure 4 It is the diagram of the change in the distribution of IMF components at four measuring points under multiple working conditions. Detailed implementation manners

[0050] The present invention will be further described in detail below with reference to the accompanying drawings.

[0051] A method for identifying the cavitation state of fluid by variational mode self-optimal reconstruction is divided into two parts. The first part is the development of the VMD self-optimal reconstruction method (SOVR), and the second part is the application of this algorithm to the cavitation state detection of high-temperature and high-pressure working fluids.

[0052] The whole process is as Figure 1 shown. First, reasonably set the pressure measurement points and collect the pressure pulsation signals of the fluid from different positions. Subsequently, use the VMD self-optimal reconstruction method to process the pressure pulsation signals of the fluid collected at each measurement point, and identify different stages of cavitation by comparing the changes in the main frequency components of different measurement points under the same working conditions. The specific content is as follows:

[0053] The following is the content of the first part, the calculation process of the VMD self-optimal reconstruction method.

[0054] First, determine the decomposition layer number by the center frequency method. The specific method is to calculate the change of the center frequency of the IMF component with the largest energy proportion during the process of gradually increasing the decomposition layer number. When the relative error of the center frequency in two adjacent decompositions is less than 5%, the previous decomposition layer number is considered the optimal layer number.

[0055] Secondly, optimize the quadratic penalty factor and update rate through the genetic algorithm. This process uses genetic operations such as selection, crossover, and mutation to find the optimal parameter combination using the fitness function composed of the reconstruction error between the IMF component and the original signal. The genetic algorithm parameters are set as the population size of 20, the selection method is roulette wheel selection, the crossover method is single-point crossover, and the mutation method is uniform mutation.

[0056] Then, distinguish the effective components and noise components by calculating the power spectrum entropy, and perform denoising reconstruction on the IMFs. First, calculate the power spectrum entropy and relative energy of the IMF components, and sort them in ascending order of the power spectrum entropy. Then, accumulate the relative energy of the IMF components in turn until the cumulative sum exceeds 0.95 for the first time when adding to the i-th component. At this time, retain the first i IMF components as the final effective IMFs.

[0057] Finally, perform spectrum calculation on each reconstructed IMF component to complete the entire VMD self-optimal reconstruction process, and the obtained IMFs are the final decomposition results.

[0058] At the same time, use a group of simulated pressure pulsation signals to verify the effectiveness of the SOVR algorithm. This experiment compares and analyzes the self-programmed VMD self-optimal reconstruction method with the built-in VMD method in MATLAB, and all codes are run on the MATLABR2023b platform.

[0059] The expression of the simulation signal f(t) is:

[0060] f(t) = sin(20πt) + 0.2sin(256πt) + 0.5sin(672πt)

[0061] The simulated signal f(t) consists of three frequency components, namely 10 Hz, 128 Hz, and 336 Hz. The time t ranges from 0 to 1 second with a sampling interval of 0.0005 seconds, and the pressure pulsation amplitude is denoted as A (unit: Pa). Considering the noise interference in the actual collected signal, Gaussian distributed random noise is added to the simulated signal, and the noise levels are set to 0.05, 0.1, 0.2, 0.3, and 0.4 respectively. The schematic diagrams of each frequency component and the simulated signal are shown as Figure 2 (a).

[0062] The VMD self-optimizing reconstruction method can automatically determine three key hyperparameters, while the traditional VMD method requires manual parameter configuration. Given that the simulated signal contains three frequency components and noise components, the decomposition layer number in the traditional VMD is set to 4, and the quadratic penalty factor and update rate are taken as 1000 and 0.01 respectively (i.e., the default values of the built-in VMD algorithm in MATLAB). Since the traditional VMD usually matches the decomposition layer number with the number of signal components, during the verification process, instead of comparing the performance of the two methods in preventing mode mixing or under-decomposition, the focus is on examining their ability to identify effective components.

[0063] Taking the noise level of 0.4 as an example, the decomposition results of SOVR and the traditional VMD are compared. The simulated signal f(t) consists of sine signals with frequencies of 10 Hz, 128 Hz, and 336 Hz, and their amplitudes are 1, 0.2, and 0.5 respectively. The obtained IMF components are shown as Figure 2 (b) and Figure 2 (c) respectively. Through SOVR decomposition, the central frequencies of the IMF components are 10.6 Hz, 142 Hz, and 335.7 Hz in sequence, and the amplitudes are 1, 0.3, and 0.6 respectively; while the decomposition results of the traditional VMD show that the central frequencies of its IMF components are 10.03 Hz, 153.65 Hz, and 351 Hz respectively, and the amplitudes are 1, 0.6, and 0.75 respectively.

[0064] Comparing with the original signal ( Figure 2 (a)), it can be seen that the decomposition result of SOVR shows better anti-noise performance and smaller relative error of the central frequency. In addition, under the condition of unknown decomposition layer number, SOVR can accurately extract the IMF components that match the original signal.

[0065] Based on the above comparison results, the effectiveness and superiority of the VMD self-optimizing reconstruction method in signal decomposition are verified.

[0066] The following is the content of the second part, applying this algorithm to the cavitation state detection of high-temperature and high-pressure working fluids.

[0067] In this part, the SOVR is used to decompose the fluid pressure pulsation signals collected at multiple measuring points. By comparing the changes in the main frequency components at different measuring points under the same working conditions, different stages of cavitation can be distinguished.

[0068] As a specific embodiment, this study adopted an orifice cavitation experiment. The orifice experimental assembly consists of an upstream section, an orifice plate, and a downstream section (see Figure 3 (a)), where the shape of the orifice plate is as shown in Figure 3 (b). The diameters of the upstream and downstream sections are both D, the length of the upstream section is 10D, and the length of the downstream section is 30D; the aperture of the orifice plate is d, and the length is l.

[0069] Pressure transmitters are installed at both the inlet and outlet of the experimental section to measure the inlet pressure (P in ) and the outlet pressure (P out ). At the same time, transient pressure sensors (marked as P1, P2, P3, and P4) are arranged at 5D upstream of the orifice plate and at 2.5D, 7.5D, and 15D downstream respectively to monitor the fluid pressure pulsation signals. High-temperature pressure sensors (Rosemount3051 and Kistler601) are used to measure the fluid pressure in the experiment. Among them, the measuring range of Kistler601 is 0 - 2 Mpa, the sampling frequency is 20000 Hz, and the uncertainty is 0.1%; the measuring range of Rosemount3051 is 0 - 12 Mpa, the sampling frequency is 1 Hz, and the uncertainty is 0.22%. The installation schematic diagram of the transient pressure sensors is as shown in Figure 3 (d). These sensors are connected to the experimental section through a base designed as a double saddle shape, aiming to minimize the influence of the opening on the flow field.

[0070] In this orifice cavitation experiment, high-temperature and high-pressure deionized water flows through the experimental section with the orifice plate at a constant flow rate, so that pressure pulsation signals are collected at multiple measuring points.

[0071] The pressure pulsation signals measured in the experiment are decomposed according to the VMD self-optimizing reconstruction method, and the results are as shown in Figure 4 . In the figure, the X-axis is the cavitation number, the Y-axis is the center frequency corresponding to each IMF component, (a), (b), (c), and (d) respectively represent the four measuring point positions, and the color reflects the relative energy magnitude of each IMF component.

[0072] The present invention uses a dimensionless cavitation number to characterize the cavitation intensity in the pipe. The smaller the cavitation number, the greater the cavitation intensity, and its definition is as follows:

[0073]

[0074] In the formula, P v is the saturated vapor pressure corresponding to the working fluid inlet temperature, P in is the inlet pressure of the experimental section, and P out is the outlet pressure of the experimental section.

[0075] Meanwhile, the relative energy ratio RE is used to describe the contribution degree of each IMF component in the original signal. Its calculation formula is as follows. In the formula, E i is the energy of each IMF component, and E total is the total energy of the reconstructed signal. The mode with a relatively higher energy indicates that it occupies a larger weight in the original signal.

[0076]

[0077] Based on the change characteristics of the frequency component distribution, it can be used to judge the incipience of cavitation and distinguish different cavitation states.

[0078] When cavitation does not occur, the frequency spectrum of each measuring point only presents one main frequency component. As Figure 4 shown, when the cavitation number is relatively large, only one IMF component is detected at all four measuring points, and this stage corresponds to the non-cavitation state.

[0079] When cavitation just occurs, a new frequency component will appear in the frequency spectrum. Figure 4 shows that an additional IMF component first appears at the P2 measuring point, marking the initial stage of cavitation.

[0080] When the cavitation intensity increases, although the total number of frequency components in the frequency spectrum remains the same as that in the initial stage of cavitation, two IMF components have appeared at P2, while only one remains at P3 and P4, indicating that cavitation has entered the development stage.

[0081] When the cavitation degree is extremely strong, more frequency components will appear in the frequency spectrum. Figure 4 shows that high-frequency IMF components are detected at the P2, P3, and P4 measuring points downstream of the orifice plate, and each measuring point contains two IMF components. This stage is the severe cavitation state.

[0082] The above is the specific application example of this algorithm in the detection of cavitation state of high-temperature and high-pressure working fluids.

Claims

1. A method for identifying the fluid cavitation state with variational mode self-optimal reconstruction, characterized in that Including the following steps; Step 1: Based on the VMD self-optimizing reconstruction method, decompose the original fluid pressure signal directly collected from the experiment into a high-quality signal decomposition result; Step 2: Based on the high-quality signal decomposition result obtained in Step 1, use the orifice cavitation experiment to analyze the pressure pulsation signals at multiple measuring points. By comparing the changes in the main frequency components at each measuring point under the same working conditions, achieve accurate identification of different stages of cavitation.

2. A method for identifying the fluid cavitation state by variational mode self-optimal reconstruction according to claim 1, characterized in that In Step 1, the high-quality signal decomposition result is the intrinsic mode function components (IMFs) reflecting the fluid flow characteristics, and the intrinsic mode function components contain spectral information closely related to the cavitation state.

3. A method for identifying the fluid cavitation state by variational mode self-optimal reconstruction according to claim 2, characterized in that In Step 1, the VMD self-optimizing reconstruction method realizes VMD decomposition; Its calculation process: Step 1: During the process of gradually increasing the decomposition layer number, calculate the change of the central frequency of the IMF component with the largest energy proportion with the layer number; when the relative error of the central frequency in two adjacent decompositions is less than 5%, the previous decomposition layer number is considered the optimal layer number; Step 2: Through genetic operations of selection, crossover, and mutation, use the fitness function composed of the reconstruction error between a group of intrinsic mode functions IMFs and the input original fluid pressure signal to find the optimal parameter combination; that is, obtain the optimal combination of the quadratic penalty factor and the update rate through the genetic algorithm to realize the reasonable selection of VMD decomposition parameters; Step 3: Distinguish the effective components and noise components by calculating the power spectrum entropy, and perform denoising reconstruction on the IMFs; Step 4: Perform spectral calculation on each reconstructed IMF component to obtain the central frequency of each IMF component, complete the entire VMD self-optimizing reconstruction process, and the obtained IMFs are the final decomposition result of the pressure signal, thereby realizing the processing of the input signal.

4. A method for identifying the fluid cavitation state with variational mode self-optimal reconstruction according to claim 3, characterized in that In Step 3, by using the decomposition layer number, quadratic penalty factor, and update rate obtained in Step 1 and Step 2, perform VMD decomposition on the input original fluid pressure signal to obtain preliminary IMFs, calculate the power spectrum entropy and relative energy of the IMF components, sort them in ascending order of the power spectrum entropy, and then accumulate the relative energy of the IMF components in turn until the cumulative sum exceeds 0.95 for the first time when adding to the i-th component. At this time, retain the first i IMF components as the final effective IMFs.

5. A method for identifying the fluid cavitation state with variational mode self-optimal reconstruction according to claim 1, characterized in that In Step 2: The orifice cavitation experiment component consists of an upstream section, an orifice plate, and a downstream section; The upstream section is used as a control reference, and a pressure measuring point is set to collect the pressure pulsation information when cavitation does not occur under the current temperature, pressure, and flow rate conditions; Three pressure measuring points are set in the downstream section to capture the pressure pulsation characteristics of the fluid after orifice cavitation, and judge the cavitation stage based on the pressure pulsation signals at different positions.

6. A method for identifying the fluid cavitation state with variational mode self-optimal reconstruction according to claim 5, characterized in that The pipe diameters of the upstream section and the downstream section are both D, the length of the upstream section is 10D, and the length of the downstream section is 30D; the orifice diameter of the orifice plate is d, and the length is l; Pressure transmitters are installed at both the inlet and outlet of the orifice plate experiment assembly to measure the inlet pressure (P in ) and the outlet pressure (P out ); meanwhile, transient pressure sensors are arranged at 5D upstream of the orifice plate and at 2.5D, 7.5D, and 15D downstream respectively, and are marked as P1, P2, P3, and P4 respectively to monitor the fluid pressure pulsation signal; The transient pressure sensor uses high-temperature pressure sensors Rosemount3051 and Kistler601 to measure the fluid pressure; The range of Kistler601 is 0-2Mpa, the sampling frequency is 20000Hz, and the uncertainty is 0.1%; the range of Rosemount3051 is 0-12Mpa, the sampling frequency is 1Hz, and the uncertainty is 0.22%; The transient pressure sensor is connected to the test section through a double saddle-shaped base.

7. The method for fluid cavitation state identification based on variational modal self-optimal reconstruction according to claim 6, characterized in that: The pressure pulsation signal measured experimentally is decomposed according to the VMD self-optimal reconstruction method. The X-axis is the cavitation number, and the Y-axis is the center frequency corresponding to each IMF component. The dimensionless cavitation number is used to characterize the cavitation intensity in the tube. The smaller the cavitation number, the greater the cavitation intensity. Its definition is as follows: where, P v is the saturation vapor pressure corresponding to the working fluid inlet temperature, P in is the inlet pressure of the experimental section, P out is the outlet pressure of the experimental section; Meanwhile, the relative energy ratio RE is used to describe the contribution degree of each IMF component in the original signal. Its calculation formula is as follows. In the formula, E i is the energy of each IMF component, and E total is the total energy of the reconstructed signal. The mode with a relatively high energy indicates that it occupies a larger weight in the original signal; After decomposing the pressure pulsation signals of each measuring point collected in the experiment using the VMD self-optimal reconstruction method, each measuring point will obtain a series of IMF components. The number of these components will vary with the change of working conditions. At the same time, each IMF component corresponds to a specific center frequency.

8. A method for identifying the fluid cavitation state by variational mode self-optimal reconstruction according to claim 7, characterized in that Under different cavitation states, the number of IMF components decomposed at each measuring point and the corresponding center frequency distribution show obvious differences. These differences are used to judge the initiation of cavitation and distinguish different cavitation states. When cavitation does not occur, the spectrum of each measuring point shows only one main frequency component. When the cavitation number is large, only one IMF component is detected at each of the four measuring points. This stage corresponds to the non-cavitation state. When cavitation just occurs, a new frequency component will appear in the spectrum, and the additional IMF component first appears at the P2 measuring point, marking the beginning of cavitation; When the cavitation intensity increases, although the total number of frequency components in the spectrum remains the same as in the early stage of cavitation, two IMF components appear at P2, while there is only one at P3 and P4, which indicates that cavitation has entered the development stage; When the cavitation degree is extremely strong, more frequency components will appear in the spectrum. High-frequency IMF components are detected at the P2, P3 and P4 measuring points downstream of the orifice plate. Each measuring point contains two IMF components. This stage is a severe cavitation stage.

Images