Valve performance detection system and method thereof

Abstract

The invention discloses a valve performance detection system and method, relates to the field of valve detection, and is used for detecting the performance of a high-pressure hydrogen valve, and the method comprises the steps: arranging vibration sensors at a valve seat of the high-pressure hydrogen valve and valve outlets at the two sides, vibration signals of different opening strokes are respectively detected by applying knocking or not applying knocking to the valve, data processing is carried out, and the hydrogen cavitation degree is evaluated based on a physical information fusion LSTM model, so that a valve material performance detection result is obtained. According to the invention, effective long-term evaluation of the hydrogen embrittlement condition of the high-pressure hydrogen valve is realized; in evaluation, data enhancement of vibration signals is realized based on spatial correlation represented by valve material embrittlement in vibration spectrum presentation of multi-point vibration signals after hammering, so that the accuracy of evaluation results is effectively improved; and the hydrogen cavitation degree is evaluated based on the physical information fusion LSTM model, so that the problem that an evaluation result of a traditional model does not conform to a physical rule is solved.

Description

Technical Field

[0001] The present invention relates to the field of valve performance, and in particular to a valve performance detection system and method thereof. Background Art

[0002] The detection of high-pressure hydrogen valves is a key link to ensure the safe and reliable operation of hydrogen energy systems. Due to the flammable and explosive characteristics of hydrogen and the hydrogen embrittlement effect, the sealing performance and material stability of valves under high-pressure conditions are directly related to the safety of the entire system. The main failure reasons of high-pressure hydrogen valves are cavitation and erosion wear. Hydrogen embrittlement may occur under the combined action of a complex multi-directional stress state and a high-pressure hydrogen environment, which may accelerate the deterioration of the mechanical properties of materials and lead to the premature occurrence of fatigue failure.

[0003] However, the current valve performance inspection methods in the prior art still adopt the method of single performance detection and cannot achieve long-term and effective monitoring of the hydrogen embrittlement effect. Summary of the Invention

[0004] In order to solve the technical problem that long-term and effective monitoring of the hydrogen embrittlement effect cannot be achieved in the prior art, the present invention provides a valve performance detection system and method thereof.

[0005] The present invention is achieved by the following technical solutions: A valve performance detection method for detecting the performance of high-pressure hydrogen valves, comprising: Arranging a first vibration sensor, a second vibration sensor, and a third vibration sensor on the valve seat and the two valve outlets on both sides of the high-pressure hydrogen valve; During the valve opening stroke, the valve is opened from closed to fully open in N steps at a step of 1 / N; the vibration signals at different opening strokes are respectively detected to obtain a first vibration signal; During the valve opening stroke, the valve is opened from closed to fully open in N steps at a step of 1 / N, and a knocking force with a certain frequency and strength is applied to the valve seat to obtain a second vibration signal; Performing data processing on the first vibration signal and the second vibration signal, including data preprocessing and feature extraction; Evaluating the degree of hydrogen cavitation based on the LSTM model of physical information fusion, so as to obtain the detection result of the valve material performance.

[0006] Furthermore, the first vibration signal includes a first sub-signal of the first vibration signal corresponding to the first vibration sensor, a second sub-signal of the first vibration signal corresponding to the second vibration sensor, and a third sub-signal of the first vibration signal corresponding to the third vibration sensor; the second vibration signal includes a first sub-signal of the second vibration signal corresponding to the first vibration sensor, a second sub-signal of the second vibration signal corresponding to the second vibration sensor, and a third sub-signal of the second vibration signal corresponding to the third vibration sensor.

[0007] Further, the data preprocessing includes subtracting the first vibration signal from the second vibration signal to obtain a third vibration signal, and the third vibration signal includes three sub-signals.

[0008] Further, the data preprocessing further includes enhancing the data space correlation of the three sub-signals of the third vibration signal based on a graph neural network.

[0009] Further, the data space correlation enhancement processing includes adjusting edge connections based on an edge adjustment module to form a new graph structure. Specifically, a random forest model is embedded to adjust the weight adjustment coefficient of the edge correlation strength in the topological relationship.

[0010] Further, the loss function expression of the LSTM model based on physical information fusion is as follows:

[0011] where N is the number of samples, yi is the observed value, is the predicted value, ρ is the regularization parameter, and Ui is the predicted value based on the physical model of the knocking vibration signal conducted in the valve.

[0012] Further, the physical model of the knocking vibration signal conducted in the valve is expressed as follows:

[0013] E is the total energy, P in is the input power, determined by the energy applied by the knocking hammer, P dis is the dissipated power, S is the energy flux vector, representing the energy flowing out, and A is the average cross-sectional area of the energy flow path.

[0014] Further, the energy flux vector expression is as follows:

[0015] where w is the angular frequency, E is the elastic modulus of the material, U is the vibration amplitude, α is the attenuation coefficient, and x is the distance from the knocking point to the measurement point.

[0016] The present invention also provides a valve performance detection system, based on the valve performance detection method described above, which includes: A first vibration sensor, a second vibration sensor, and a third vibration sensor, which are respectively arranged at the valve seat of the high-pressure hydrogen valve and the two valve outlets on both sides; A valve control module, which is used to control the valve opening; A vibration signal acquisition module, which is used to acquire vibration signals from the three vibration sensors; A vibration signal processing module for preprocessing and feature extraction of vibration signals, including data cleaning and denoising, differential operation of vibration signals, and data augmentation. A hydrogen cavitation degree evaluation and performance detection result acquisition module for evaluating the hydrogen cavitation degree based on a physical information fusion LSTM model to obtain the valve material performance detection result.

[0017] In addition, to achieve the above object, the present invention also provides a computer-readable storage medium, on which program instructions for the valve performance detection method are stored. The program instructions for the valve performance detection method can be executed by one or more processors to implement the steps of the valve performance detection method as described above.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention realizes an effective long-term evaluation of the hydrogen embrittlement of high-pressure hydrogen valves based on applying knocking and detecting vibrations. In the evaluation of the hydrogen embrittlement of high-pressure hydrogen valves, data augmentation of vibration signals is realized based on the spatial correlation presented by the vibration spectra of multi-point vibration signals after the valve material embrittlement is reflected by hammering. And the problem that the evaluation results of traditional models do not conform to physical laws is solved by evaluating the hydrogen cavitation degree based on a physical information fusion LSTM model. Description of the Drawings

[0019] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments and descriptions thereof are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings: Figure 1 It is a schematic flowchart of a valve performance detection method according to an embodiment of the present application. Detailed Embodiments

[0020] The embodiments of the present invention will be described in detail below with reference to the drawings.

[0021] The following illustrates the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, without conflict, the following embodiments and the features in the embodiments can be combined with each other. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0022] It should also be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner. The diagrams only show the components related to the present invention, rather than being drawn according to the number, shape, and size of the components in actual implementation. The form, quantity, and ratio of each component in actual implementation can be arbitrarily changed, and the layout form of its components may also be more complex.

[0023] See Figure 1 , a valve performance detection method for high-pressure hydrogen valve detection, comprising the following steps: S1: Set a first vibration sensor, a second vibration sensor, and a third vibration sensor on the valve seat and the two valve outlets of the high-pressure hydrogen valve; S2: During the valve opening stroke, the valve is opened from closed to fully open in N steps at a step of 1 / N; the vibration signals at different opening strokes are respectively detected to obtain a first vibration signal, which includes a first sub-signal of the first vibration signal corresponding to the first vibration sensor, a second sub-signal of the first vibration signal corresponding to the second vibration sensor, and a third sub-signal of the first vibration signal corresponding to the third vibration sensor; S3: During the valve opening stroke, the valve is opened from closed to fully open in N steps at a step of 1 / N, and a knocking force with a certain frequency and intensity is applied to the valve seat to obtain a second vibration signal, which includes a first sub-signal of the second vibration signal corresponding to the first vibration sensor, a second sub-signal of the second vibration signal corresponding to the second vibration sensor, and a third sub-signal of the second vibration signal corresponding to the third vibration sensor; Optionally, a pneumatic hammer or an electric hammer is used to apply the knocking force.

[0024] S4: Perform data processing on the first vibration signal and the second vibration signal, including data preprocessing and feature extraction; Specifically, the data preprocessing includes: S41: Data cleaning and denoising; Optionally, noise is filtered based on an adaptive filtering method, and outliers are removed through a sliding window.

[0025] S42: Subtract the second vibration signal from the first vibration signal to obtain a third vibration signal. Specifically, the signals obtained by the first vibration sensor, the second vibration sensor, and the third vibration sensor are respectively subtracted to obtain a first sub-signal of the third vibration signal, a second sub-signal of the third vibration signal, and a third sub-signal of the third vibration signal; the first sub-signal of the third vibration signal is obtained by subtracting the first sub-signal of the second vibration signal from the first sub-signal of the first vibration signal, the second sub-signal of the third vibration signal is obtained by subtracting the second sub-signal of the second vibration signal from the second sub-signal of the first vibration signal, and the third sub-signal of the third vibration signal is obtained by subtracting the third sub-signal of the second vibration signal from the third sub-signal of the first vibration signal.

[0026] The difference between two signals is used to remove the vibration caused by the gas flow of the valve, and the vibration signal obtained by knocking with a hammer is obtained, so as to highlight the abnormal situation in the vibration signal caused by the embrittlement of the material due to the hydrogen embrittlement effect of the valve.

[0027] S43: Data enhancement; Since the embrittlement of the valve material is reflected in the spatial correlation of the vibration spectra of the multi-point vibration signals after being hammered, and the traditional vibration signal processing methods ignore the spatial correlation of the multi-scale vibration signals, in order to enhance the spatial correlation of the vibration signal features caused by the embrittlement of the valve material, the present invention performs data spatial correlation enhancement processing on the three sub-signals of the third vibration signal. Specifically, the sub-signals are regarded as nodes in the graph structure, and the graph neural network is used to model the spatial correlation between the nodes to generate enhanced data that conforms to the topological relationship, specifically including: a. Obtain the setting position coordinates of the three sensors; b. Construct a graph structure: The nodes are the feature vectors corresponding to the three sub-signals; the edge weights are the correlation strengths calculated based on the physical distances between the sub-signals; Specifically, it is assumed that the correlation strength decays in a Gaussian distribution with the distance, and the calculation formula of the correlation strength is as follows:

[0028] σ is the smoothing degree coefficient of attenuation; d is the distance between two sensors, is the initial value of the weight adjustment coefficient.

[0029] c. Graph data enhancement: Since the feature vectors of the corresponding vibration signals have different degrees of influence on the prediction of the embrittlement of the valve material, the present invention realizes graph data enhancement based on dynamically adjusting the weight adjustment coefficient; Including: Based on the edge adjustment module, adjust the edge connections to form a new graph structure. Specifically, embed a random forest model to adjust the weight adjustment coefficient: Since in the prediction of valve hydrogen embrittlement, the vibration spectrum features can effectively distinguish different material embrittlement situations, therefore, the present invention uses the vibration spectrum features to adjust the weight adjustment coefficient. Specifically, the random forest model inputs the vibration spectrum features and outputs the weight adjustment coefficient. Optionally, the vibration spectrum features include spectrum peak value, harmonic component, frequency band energy ratio, and vibration spectrum entropy change rate.

[0030] When the correlation strength is less than the set threshold, delete the edge connection; Traverse all the edges in turn to reset the correlation strength to obtain an updated graph structure.

[0031] By the above method, the problem of mixing of adjacent nodes in the graph network is effectively solved. d. Inverse mapping to the time domain: Convert the enhanced graph node features back to the time domain signal.

[0032] Specifically, feature extraction includes time domain features, frequency domain features, and time-frequency domain features. The time domain features include root mean square value, peak value, and kurtosis; the frequency domain features include spectral peak value, harmonic components, and frequency band energy ratio; to strengthen the change of vibration spectrum entropy of multi-point vibration signals after the valve material is embrittled and hammered, the present invention is based on the HHT transform to obtain the Hilbert spectrum of the structural nonlinear vibration signal, and further calculates the spectrum entropy as the time-frequency domain feature.

[0033] S5: Evaluate the degree of hydrogen cavitation by the LSTM model based on physical information fusion, so as to obtain the detection result of the valve material performance, specifically as follows: In the prior art, when using the traditional LSTM model for evaluation, there are problems such as slow convergence of the general model, low accuracy, and violation of physical laws, and it is impossible to effectively fuse the time-frequency characteristics of the vibration signal in the cavitation detection method of the present invention. Therefore, the present invention uses the LSTM model based on physical information fusion to detect the degree of hydrogen cavitation, and adds the physical constraint of the vibration signal as a regularization term to the loss function to solve the problems existing in the traditional model. Specifically, the expression of the loss function is as follows:

[0034] Among them, N is the number of samples, yi is the observed value, is the predicted value, ρ is the regularization parameter, and Ui is the predicted value based on the physical model of the knocking vibration signal conducted in the valve.

[0035] The boundary effect is considered in the physical model, including the redistribution of energy caused by reflection, transmission, and scattering, such as the scattering caused by defects such as cracks and holes. The physical model of the knocking vibration signal conducted in the steel material pipeline is expressed as follows:

[0036] E is the total energy, P in is the input power, which is determined by the energy applied by the knocking hammer, P dis is the dissipated power, S is the energy flux vector, represents the energy outflow, and A is the average cross-sectional area of the energy flow path.

[0037] The magnitude of the energy flux vector is proportional to the square of the vibration amplitude. In the valve system model, the expression of the energy flux vector is as follows:

[0038] Among them, w is the angular frequency, E is the elastic modulus of the material, U is the vibration amplitude, α is the attenuation coefficient, and x is the distance from the tapping point to the measurement point.

[0039] The dissipated power is determined according to the valve material.

[0040] Furthermore, corresponding valve material performance detection results are obtained based on the evaluation results of the hydrogen cavitation degree.

[0041] The present invention takes physical constraints as prior knowledge, guides the model to maintain reasonable output with the physical model of valve vibration, and reduces the dependence on pure data-driven.

[0042] In this embodiment, an effective long-term evaluation of the hydrogen embrittlement situation of high-pressure hydrogen valves is realized based on applying tapping and detecting vibrations; in the evaluation of the hydrogen embrittlement situation of high-pressure hydrogen valves, data enhancement of vibration signals is realized based on the spatial correlation presented by the vibration spectra of multi-point vibration signals after the valve material embrittlement is reflected under hammering; and the problem that the evaluation results of traditional models do not conform to physical laws is solved by evaluating the hydrogen cavitation degree based on the LSTM model of physical information fusion.

[0043] The embodiment of the present invention also proposes a valve performance detection system, which is used for high-pressure hydrogen valve detection based on the above-mentioned valve performance detection method, and includes: The first vibration sensor, the second vibration sensor, and the third vibration sensor are respectively arranged on the valve seat and the two valve outlets of the high-pressure hydrogen valve; The valve control module is used to control the valve opening; The vibration signal acquisition module is used to acquire vibration signals from the three vibration sensors; The vibration signal processing module is used for data preprocessing and feature extraction of vibration signals; including data cleaning and denoising, subtracting vibration signals, and data enhancement; The hydrogen cavitation degree evaluation and performance detection result acquisition module is used to evaluate the hydrogen cavitation degree based on the LSTM model of physical information fusion, so as to obtain the valve material performance detection result.

[0044] In addition, the embodiment of the present invention also proposes a computer-readable storage medium, on which program instructions of the valve performance detection method are stored, and the program instructions of the valve performance detection method can be executed by one or more processors to implement the steps of the above-mentioned valve performance detection method.

[0045] The embodiments described above are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A valve performance detection method, characterized in that, For detecting the performance of high-pressure hydrogen valves, including: A first vibration sensor, a second vibration sensor, and a third vibration sensor are arranged at the valve seat of the high-pressure hydrogen valve and the valve outlets on both sides; Within the valve opening stroke, the valve is opened from closed to fully open in N steps at a step of 1 / N; the vibration signals at different opening strokes are respectively detected to obtain a first vibration signal; Within the valve opening stroke, the valve is opened from closed to fully open in N steps at a step of 1 / N, and a knocking force with a certain frequency and intensity is applied to the valve seat to obtain a second vibration signal; Data processing is performed on the first vibration signal and the second vibration signal, including data preprocessing and feature extraction; The degree of hydrogen cavitation is evaluated based on the LSTM model of physical information fusion, so as to obtain the detection result of the valve material performance.

2. The valve performance detection method according to claim 1, wherein The first vibration signal includes a first sub-signal of the first vibration signal corresponding to the first vibration sensor, a second sub-signal of the first vibration signal corresponding to the second vibration sensor, and a third sub-signal of the first vibration signal corresponding to the third vibration sensor; the second vibration signal includes a first sub-signal of the second vibration signal corresponding to the first vibration sensor, a second sub-signal of the second vibration signal corresponding to the second vibration sensor, and a third sub-signal of the second vibration signal corresponding to the third vibration sensor.

3. The valve performance detection method according to claim 2, characterized in that The data preprocessing includes subtracting the first vibration signal from the second vibration signal to obtain a third vibration signal, and the third vibration signal includes three sub-signals.

4. The valve performance detection method according to claim 3, characterized in that The data preprocessing further includes enhancing the data spatial correlation of the three sub-signals of the third vibration signal based on a graph neural network.

5. The valve performance detection method according to claim 4, characterized in that, The data spatial correlation enhancement processing includes adjusting the edge connection based on an edge adjustment module to form a new graph structure. Specifically, a weight adjustment coefficient for adjusting the edge correlation strength in the topological relationship is embedded in a random forest model.

6. The valve performance detection method according to claim 3, wherein The expression of the loss function of the LSTM model based on physical information fusion is as follows: where N is the number of samples, yi is the observed value, is the predicted value, ρ is the regularization parameter, and Ui is the predicted value based on the physical model of the valve conduction of the knocking vibration signal.

7. The valve performance detection method according to claim 6, characterized in that, The physical model of the knocking vibration signal conducted in the valve is expressed as follows: E is the total energy, P in is the input power, determined by the energy applied by the percussion hammer, P dis is the dissipated power, S is the energy flux vector, indicating the energy outflow, A is the average cross-sectional area of the energy flow path.

8. The valve performance detection method according to claim 7, characterized in that, The expression of the energy flow vector is as follows: Wherein, w is the angular frequency, E is the elastic modulus of the material, U is the vibration amplitude, α is the attenuation coefficient, and x is the distance from the percussion point to the measurement point.

9. A valve performance detection system, based on the valve performance detection method according to any one of claims 1 to 8, includes: A first vibration sensor, a second vibration sensor, and a third vibration sensor are respectively arranged at the valve seat of the high-pressure hydrogen valve and the valve outlets on both sides; A valve control module, which is used to control the valve opening; A vibration signal acquisition module, which is used to acquire vibration signals from the three vibration sensors; A vibration signal processing module, which is used for data preprocessing and feature extraction of vibration signals; including data cleaning and denoising, subtracting vibration signals, and data enhancement; A hydrogen cavitation degree evaluation and performance detection result acquisition module, which is used to evaluate the hydrogen cavitation degree based on the LSTM model of physical information fusion, so as to obtain the detection result of the valve material performance.

10. A computer-readable storage medium, characterized in that, Program instructions for the valve performance detection method are stored on the computer-readable storage medium, and the program instructions for the valve performance detection method can be executed by one or more processors to implement the steps of the valve performance detection method according to any one of claims 1 to 8.

Images