Method and system for detecting running state of hydrogen pipeline flame arrester

Abstract

The invention provides a hydrogen pipeline flame arrester operation state detection method and system, and the key point of the technical scheme comprises the steps: obtaining historical normal data, and setting a threshold parameter corresponding to each detection information; acquiring detection information of each sensor in real time and preprocessing the detection information; performing time-frequency domain analysis processing on the preprocessed vibration information to obtain a plurality of IMF components and residual errors, and extracting vibration features to perform preliminary vibration abnormality judgment; in response to the vibration characteristic abnormity, calculating a dynamic vibration threshold value corresponding to the current sampling moment; performing abnormity judgment on the running state of the flame arrester at the current sampling moment; according to the invention, abnormal information of the flame arrester can be monitored in real time, the current state of the flame arrester can be accurately judged, potential safety hazards caused by failure of the flame arrester can be avoided, tedious operation of manual detection can be reduced, and omission of manual detection can be avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of flame arrester detection, and particularly to a method and system for detecting the operating state of a hydrogen pipeline flame arrester. Background Art

[0002] As a chemical raw material, hydrogen is widely used in fields such as metallurgy, aerospace, electronics, glass, fine chemicals, and energy. Since hydrogen is an inflammable and explosive gas, in order to maintain the safety of the lives of workers and the production safety of enterprises in the actual production work in the field of hydrogen application, a flame arrester needs to be installed in the hydrogen pipeline to prevent serious accidents caused by backfire, deflagration or leakage. At the same time, it is necessary to conduct daily inspections on the flame arrester to ensure that the flame arrester is in a normal working state.

[0003] A flame arrester is a safety device used to prevent the spread of combustion flames of flammable gases and flammable liquid vapors. It is generally installed in pipelines for transporting combustible gases or on ventilated tanks to prevent the spread of flames and deflagration or detonation phenomena caused by flames. The flame arrester mainly consists of two parts: a shell and a flame arrestor core. In the daily inspection and maintenance of the flame arrester, it is usually necessary to disassemble it and conduct visual inspection manually to check whether there are defects such as blockage, deformation, and corrosion in the flame arrestor layer and holes of the flame arrestor core.

[0004] When there are relatively large abnormal defects in the flame arrester, relevant staff need to be able to detect them in time and carry out repairs or replacements to ensure that the flame arrester in the hydrogen pipeline can effectively prevent fire in a high-risk environment, avoid serious accidents caused by backfire, deflagration or leakage, and avoid potential safety hazards caused by the failure of the flame arrester. Therefore, it is very important to inspect and monitor the state of the flame arrester. Summary of the Invention

[0005] In order to monitor the operating state of the flame arrester in real time, avoid potential safety hazards caused by the failure of the flame arrester, at the same time, reduce the cumbersome operation of manual detection and avoid omissions in manual detection, the present invention provides a method and system for detecting the operating state of a hydrogen pipeline flame arrester.

[0006] In a first aspect, the present invention provides a method for detecting the operating state of a hydrogen pipeline flame arrester, and its technical solution is as follows: A method for detecting the operating state of a hydrogen pipeline flame arrester, the steps of which include: obtaining historical normal data and setting threshold parameters corresponding to each detection information; obtaining the detection information of each sensor in real time and performing preprocessing; performing time-frequency domain analysis on the preprocessed vibration information to obtain a plurality of IMF components and residuals, extracting vibration characteristics for preliminary vibration anomaly judgment; in response to abnormal vibration characteristics, calculating the dynamic vibration threshold corresponding to the current sampling moment; and performing anomaly determination on the operating state of the flame arrester at the current sampling moment. Among them, the vibration characteristics include the energy value and the total energy entropy of the current vibration information. The energy of the i-th IMF component at the t-th sampling moment is ; the total energy entropy of the vibration information at the t-th sampling moment is , and the calculation formula is: , where, represents the energy proportion of the i-th IMF component at the t-th sampling moment, ln represents the natural logarithm, and n represents the number of IMF components; The dynamic vibration threshold is , and the calculation formula is: , where k is the error correction coefficient, represents the theoretical energy value of the i-th IMF component at the current sampling moment.

[0007] Preferably, historical normal data is obtained, and threshold parameters corresponding to various detection information are set, including adaptively installing the corresponding sensors for detecting vibration, hydrogen flow rate, temperature, and hydrogen concentration respectively. According to the performance standards of the corresponding type of flame arrester and historical normal data, threshold parameters corresponding to various detection information are set; Among them, the threshold parameters of temperature and hydrogen concentration can be directly set through the performance standards of the flame arrester. The dynamic vibration threshold for judging the mechanical state of the flame arrester needs to obtain historical normal data on vibration and hydrogen flow rate for calculation. If relevant historical normal data is saved, it is directly imported and referenced. If no relevant historical data is saved, it is necessary to take the data record for a period of time under normal working conditions as historical normal data after manually detecting and confirming that the flame arrester is in a normal state.

[0008] Preferably, the detection information of each sensor is obtained in real time, and the collected vibration information and hydrogen flow rate information are preprocessed, including synchronously collecting vibration information data and hydrogen flow rate information data at the same sampling frequency. Among them, the initial vibration information data at the t-th sampling moment is , and the flow rate information data at the t-th sampling moment is ; The vibration information is preprocessed, including noise reduction processing for eliminating high-frequency electromagnetic interference in the vibration information; detrending processing for eliminating low-frequency drift in the vibration information, such as sensor baseline drift and temperature influence. Among them, after preprocessing the vibration information data at the t-th sampling moment, the standard vibration information data at the t-th sampling moment is obtained; The flow rate information is preprocessed, including synchronizing the time stamps for aligning the time stamps of the flow rate information with those of the vibration information.

[0009] Preferably, perform time-frequency domain analysis on the preprocessed vibration information to obtain multiple IMF components and residuals, including analyzing the vibration information through the EMD algorithm to obtain the decomposition data of the vibration information. Among them, the preprocessed standard vibration information data at the t-th sampling moment is decomposed from high frequency to low frequency into n IMF components and residuals. The expression formula of the decomposition result is: In the formula, represents the i-th IMF component of the vibration information at the t-th sampling moment, represents the residual signal of the vibration information at the t-th sampling moment, and n represents the number of IMF components or the decomposition layer number.

[0010] Preferably, extract vibration characteristics for preliminary vibration anomaly judgment, including calculating the energy value and energy proportion of each IMF component at the t-th sampling moment and the total energy entropy of the vibration information at the t-th sampling moment; represents the energy of the i-th IMF component at the t-th sampling moment, and the calculation formula is: In the formula, represents the m-th decomposition data of the i-th IMF component, represents the summation function, and M represents the number of data of the vibration information at the current sampling moment; represents the energy proportion of the i-th IMF component at the t-th sampling moment, and the calculation formula is: Set a threshold F. When ≤F, it is determined that the vibration information at the t-th sampling moment is normal and the mechanical state of the flame arrester at the t-th sampling moment is normal. When >F, it is preliminarily determined that the vibration information at the t-th sampling moment is abnormal, and it is necessary to further combine the flow information at the t-th sampling moment to perform abnormal judgment on the mechanical state of the flame arrester.

[0011] Preferably, in response to the abnormal vibration characteristics, calculate the dynamic vibration threshold corresponding to the current sampling moment, including obtaining the correlation regression coefficient between the vibration energy value and the flow information based on historical normal data, obtaining the decomposition signal of the vibration information in the historical normal data through the EMD algorithm, obtaining n historical IMF components and residuals, and correspondingly calculating the energy value of each historical IMF component; The historical vibration energy Convert it into the form of a matrix sequence and use it as the dependent variable of the linear regression model respectively. The i-th historical vibration energy The matrix sequence expression is as follows: Convert the corresponding historical flow information data into the form of a matrix sequence as the independent variable. The corresponding historical flow information data The matrix sequence expression is as follows: Establish the relationship between the historical vibration information characteristics and the historical flow information characteristics through the linear regression model, and obtain the historical vibration energy corresponding to each historical IMF component and the historical hydrogen flow The regression coefficient between them; Establish a linear regression model separately for each historical IMF component in the historical normal data. Among them, the linear regression model of the i-th historical vibration energy is as follows: where the value of i ranges from 1 to n, and represent the regression coefficients corresponding to the i-th historical IMF component; Use the least squares formula to calculate the regression coefficients corresponding to each historical IMF component respectively. Among them, the specific calculation formula for the regression coefficient corresponding to the i-th historical IMF component is: In the formula, represents the m-th data in the matrix sequence of the i-th historical vibration energy , represents the average value of M data in the matrix sequence of the i-th historical vibration energy , represents the m-th data in the matrix sequence of the corresponding historical flow information data , represents the average value of M data in the corresponding historical flow information .

[0012] Preferably, in response to the abnormal vibration characteristics, calculate the dynamic vibration threshold corresponding to the current sampling moment, and also include calculating the theoretical energy value corresponding to each IMF component at the current sampling moment according to the regression coefficients corresponding to each historical IMF component. Among them, the theoretical energy value of the i-th IMF component at the t-th sampling moment is calculated as follows: In the formula, represents the flow information data at the t-th sampling moment; Obtain the average energy value of each IMF component at the current sampling moment as the dynamic vibration threshold at the current sampling moment The comparison object, and the calculation formula is: , where represents the average energy value of n IMF components at the t-th sampling moment.

[0013] Preferably, an abnormality determination is performed on the operating state of the flame arrester at the current sampling moment, including judging the mechanical state of the flame arrester at the current sampling moment. Taking the t-th sampling moment as an example, when , it indicates that the vibration information at the current sampling moment is within the normal fluctuation range, and the mechanical state of the flame arrester is normal. When , it indicates that the vibration information at the current sampling moment is abnormal, and the mechanical state of the flame arrester is abnormal, then it is determined that the current state of the flame arrester is abnormal; Combined with the detection information of temperature and hydrogen concentration, the temperature and hydrogen concentration information detected at the current sampling moment are respectively compared with the set thresholds for judgment. When any one of the temperature or hydrogen concentration is judged to be abnormal, it is determined that the current state of the flame arrester is abnormal; when the current state of the flame arrester is determined to be abnormal, an abnormal alarm is triggered and the corresponding abnormal information data is recorded.

[0014] In a second aspect, the present invention provides a detection system for the operating state of a hydrogen pipeline flame arrester, and its technical solution is as follows: A detection system for the operating state of a hydrogen pipeline flame arrester, which is used to implement the above-mentioned operating state detection method. Its hardware structure includes: a processor for processing data, a data memory for storing abnormal information data, and a data buffer for caching various sensing information data. The processor is burned with computer program instructions for implementing the above-mentioned detection method. The data buffer is electrically connected to a vibration sensor for capturing the high-frequency mechanical vibration information of the flame arrester, a flow meter for measuring the hydrogen flow rate in the pipeline in real time, a temperature sensor for detecting the ambient temperature of the flame arrester accessories, and a hydrogen sensor for detecting the hydrogen concentration of the flame arrester accessories.

[0015] Preferably, the vibration sensor is installed at the vibration-sensitive point of the flame arrester shell, and the sampling frequency of the vibration sensor is greater than or equal to 10 kHz; the flow meter is installed on the hydrogen pipeline corresponding to the flame arrester, and the sampling frequency of the flow meter is the same as that of the vibration sensor.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. A method for detecting the operating state of a hydrogen pipeline flame arrester. By detecting and analyzing the vibration information and hydrogen flow information of the flame arrester, the mechanical state of the flame arrester can be accurately judged. Then, combined with the detection and judgment of temperature and hydrogen concentration, the current operating state of the flame arrester can be detected and determined in real time and effectively. Furthermore, when the operating state of the flame arrester is abnormal, relevant staff can be reminded in time to repair or replace the flame arrester, effectively avoiding potential safety hazards caused by the failure of the flame arrester.

[0017] 2. A system for detecting the operating state of a hydrogen pipeline flame arrester can, through a variety of detection sensors, monitor the operating state of the flame arrester in real time and promptly feedback the abnormal information of the flame arrester, thereby effectively reducing the cumbersome operations of manual detection and avoiding omissions in manual detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] By referring to the following detailed description with reference to the accompanying drawings, the above and other objects, features, and advantages of the exemplary embodiments of the present invention will become readily understandable. In the drawings, several embodiments of the present invention are shown in an exemplary rather than restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 is the flowchart of the operating state detection method; Figure 2 is the structural block diagram of the operating state detection system; In the figure: 1 is a processor; 2 is a data memory; 3 is a data buffer; 4 is a vibration sensor; 5 is a flowmeter; 6 is a temperature sensor; 7 is a hydrogen sensor. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The following further details the technical features of the present invention with reference to the accompanying drawings for the convenience of those skilled in the art to understand.

[0020] In actual production activities related to hydrogen, the installation and application scenarios of the flame arrester are as follows: For example, in a nozzle structure using hydrogen as fuel, a flame arrester with an appropriate diameter should be installed on the hydrogen transmission pipe behind it; when hydrogen is used as fuel for welding or flame cutting, etc., a flame arrester should be set on the branch pipe of each hydrogen-using device or each group of hydrogen-using devices; a flame arrester should be set on the transmission pipeline of the hydrogen station; at the same time, in the production field using hydrogen as a chemical raw material, a flame arrester should be set on the hydrogen transmission pipeline connecting the production equipment of the chemical plant; In summary, in actual application scenarios, the environmental factors affecting the state judgment of the flame arrester are often relatively complex. Therefore, multi-dimensional detection and judgment are required from aspects such as vibration information, hydrogen flow information, temperature information, and hydrogen concentration information.

[0021] Moreover, in the actual application process of detecting and judging the operating state of the flame arrester, complex conditions such as multi-fault coupling, strong background noise, and dynamic changes in the hydrogen flow rate in the pipeline usually need to be considered. The vibration information of the flame arrester can effectively reflect the mechanical state of the flame arrester under complex environmental conditions. Therefore, by detecting and analyzing the vibration information of the flame arrester, considering the influence of hydrogen flow rate changes on the vibration information, and extracting relevant features, the mechanical state of the flame arrester can be judged. Combining with the detection of temperature and hydrogen concentration, the current operating state of the flame arrester can be detected and determined in real time and effectively.

[0022] A method for detecting the operating state of a hydrogen pipeline flame arrester, and the operating logic of the operating state detection method is as Figure 1 shown, and the specific operating steps are as follows: Step S1: Obtain historical data under normal working conditions, and set threshold parameters corresponding to each detection information according to the performance standards of the flame arrester of the corresponding model and the historical normal data; Specifically, the corresponding sensors for detecting vibration, hydrogen flow rate, temperature, and hydrogen concentration are respectively installed in a compatible manner. Among them, the threshold parameters of temperature and hydrogen concentration can be directly set through the performance standards of the flame arrester, and are respectively used to judge whether the environmental temperature of the flame arrester is safe and whether there is hydrogen leakage; In addition, the corresponding threshold parameters for judging the mechanical state of the flame arrester are calculated based on the historical normal data of vibration information and hydrogen flow rate information. Therefore, it is necessary to obtain the historical normal data on vibration and hydrogen flow rate for calculation. If the relevant historical normal data is saved, it can be directly imported and referenced. If the relevant historical data is not saved, it is necessary to take the data record for a period of time under normal working conditions as the historical normal data after manually detecting and confirming that the flame arrester is in a normal state. The sampling time of the historical normal data is generally 10 minutes to 24 hours.

[0023] Step S2: Obtain the detection information of each sensor in real time, and preprocess the collected vibration information and hydrogen flow rate information; Specifically, the vibration sensor is installed at the vibration-sensitive point of the flame arrester shell, and the sampling frequency needs to be ≥10 kHz for capturing high-frequency mechanical vibrations; the flowmeter is installed on the hydrogen pipeline corresponding to the flame arrester, and the sampling frequency needs to be ≥100 Hz for real-time measurement of the hydrogen flow rate; generally, since the high-frequency mechanical vibration information can effectively reflect the mechanical state of the flame arrester, the vibration sensor requires a relatively high sampling frequency to obtain sufficient detection information, while the flow rate detection of hydrogen does not require a particularly high sampling frequency to obtain the change characteristics of the flow rate; however, in the actual application scenario of this detection method, in order to calculate the correlation regression coefficient between the vibration characteristics and the flow information, it is necessary to align the data lengths of the vibration information and the flow information. Therefore, the sampling frequency of the flowmeter needs to be consistent with that of the vibration sensor; when the sampling frequency of the vibration sensor is set to 10 kHz, the sampling frequency of the flowmeter also needs to be set to 10 kHz. The vibration information segment and the flow rate value are synchronously recorded with a period of 1 second. Then, each segment of vibration information contains 10,000 sampling data points, and each segment of flow information also contains 10,000 sampling data points. Synchronously collect the vibration information data and the hydrogen flow information data at the same sampling frequency. Among them, the initial vibration information data at the t-th sampling moment is , and the flow information data at the t-th sampling moment is ; In addition, in order to reduce the influence of environmental factors on the vibration information, it is necessary to perform noise reduction processing and detrending processing on the collected vibration information. Preferably, this solution uses the adaptive wavelet threshold denoising method for noise reduction processing. The wavelet basis is selected as the Sym8 wavelet, and the decomposition level is 3 layers, which can eliminate the high-frequency electromagnetic interference in the vibration information; preferably, this solution uses the least squares method to fit and eliminate the trend term of the vibration information, which can eliminate the low-frequency drift in the vibration information, such as the sensor baseline drift and temperature influence, and avoid the interference of the trend term on the IMF decomposition; among them, after preprocessing the vibration information data at the t-th sampling moment, the standard vibration information data at the t-th sampling moment is obtained; the preprocessing of the flow information requires synchronous time scale alignment to align the time stamp of the flow information with the time stamp of the vibration information. In addition, for application scenarios with relatively complex environmental factors, the vibration information is relatively messy. When decomposing the vibration information by the EMD algorithm, since the endpoints of the signal cannot be at the maximum or minimum values simultaneously, the upper and lower envelopes will diverge at both ends of the data sequence, and this divergence will gradually move inward as the operation progresses, thus affecting the entire data sequence. This is the so-called endpoint effect of the EMD method; at this time, it is necessary to suppress the endpoint effect when preprocessing the vibration information, and the extreme value extension method or the endpoint mirror method can be used to alleviate the distortion generated at both ends of the EMD envelope interpolation and avoid the divergence at both ends of the decomposition result.

[0024] Step S3: Perform time-frequency domain analysis on the preprocessed vibration information to obtain multiple IMF components and residuals, extract vibration features, and make a preliminary judgment on vibration anomalies; Preferably, the EMD algorithm is used to analyze the vibration information in this solution because the vibration information may contain multiple frequency components. For example, the vibration frequencies of different components are different. The empirical mode decomposition algorithm, that is, the EMD algorithm, can decompose the intrinsic mode function IMF corresponding to each frequency component. After the vibration information is decomposed by EMD, each IMF component may correspond to the vibration mode of one of the components. Therefore, by extracting the relevant IMF component features, the characteristics of each vibration source can be analyzed. Therefore, the EMD algorithm can effectively analyze the vibration information to facilitate anomaly diagnosis.

[0025] Specifically, the vibration information is analyzed by the EMD algorithm to obtain the decomposition data of the vibration information, including the preprocessed standard vibration information data at the t-th sampling moment Decompose from high frequency to low frequency into n IMF components and residuals. The expression formula of the decomposition result is: Among them, IMF is the intrinsic mode function, represents the i-th IMF component of the vibration information at the t-th sampling moment, represents the residual signal of the vibration information at the t-th sampling moment, and n represents the number of IMF components or the decomposition layer number; Generally, the vibration information data of the flame arrester needs to be decomposed into 4 to 6 layers. Taking 4-layer decomposition as an example in this solution, the vibration information data of the flame arrester is decomposed from high frequency to low frequency into 4 IMF components and residuals. Among them, the high-frequency IMF components can reflect mechanical problems such as local looseness, micro-impact noise, structural resonance, periodic wear, and pipeline fluid pulsation coupling vibration; while the low-frequency IMF components generally represent factors such as background noise and low-frequency mechanical vibration. Therefore, in practical applications, it is generally only necessary to consider the relevant features of the IMF1 component, IMF2 component, and IMF3 component to judge the current mechanical state of the flame arrester; In addition, for the vibration information, the CEEMDAN algorithm, that is, the adaptive noise-assisted decomposition algorithm, can also be used to suppress mode mixing and optimize the decomposed signal, or the ICA algorithm or PCA algorithm can be used to extract effective features to optimize the decomposed signal. However, in actual applications, the calculation processes of these algorithms are too cumbersome, requiring a large amount of computing resources and hardware resources for processing data, and the cost is too high to be suitable for popularization and application. Moreover, if the calculation time is too long, it is difficult to ensure the real-time nature of the detection information. Therefore, the EMD algorithm is used to process and analyze the vibration information in this solution, which can balance the amount of data calculation and the accuracy of judgment.

[0026] Specifically, extract the features of each IMF component, including calculating the energy value of each IMF component at the t-th sampling moment. , the energy proportion and the total energy entropy of the vibration information at the t-th sampling moment ; According to the sampling characteristics of the vibration sensor, the collected vibration information is discrete data. Since each step of the EMD algorithm processes the original signal or its residual, each decomposed IMF component will maintain the same number of data points as the original signal. Therefore, the data lengths of each IMF component are equal and the same as the length of the original signal. At the same time, to avoid excessive calculation values, normalization processing can be performed by dividing by the total number M of vibration information data at the current sampling moment; Among them, represents the energy of the i-th IMF component at the t-th sampling moment, and the calculation formula is: In the formula, represents the summation function, M represents the number of data of the vibration information at the t-th sampling moment, and each IMF component contains M decomposed data, represents the m-th decomposed data of the i-th IMF component; In addition, represents the energy proportion of the i-th IMF component at the t-th sampling moment, and the calculation formula is: In addition, represents the total energy entropy of the vibration information at the t-th sampling moment, and the calculation formula is: In the formula, ln represents the natural logarithm, and n represents the number of IMF components; Specifically, the total energy entropy of the current vibration information can be used to quantify the stability of the energy distribution of each IMF component. When the flame arrester is in a normal state, the energy distribution of the vibration information is stable and the entropy value of the total energy entropy is low. When the flame arrester has a mechanical fault, the energy of the high-frequency component of the vibration information suddenly increases and the entropy value of the total energy entropy increases. Therefore, through the change of the entropy value of the total energy entropy, a preliminary abnormal determination of the vibration information at the current sampling moment can be made; To make a preliminary vibration abnormality judgment, a threshold F needs to be set. When ≤F, it is determined that the vibration information at the current sampling moment is normal and the mechanical state of the flame arrester at the current sampling moment is normal. When When it is greater than F, it is preliminarily determined that the vibration information at the current sampling moment is abnormal, and it is necessary to further combine the flow information at the current sampling moment to judge the abnormal mechanical state of the flame arrester; in relevant experimental tests, the normal value range of the total energy entropy is between 1.0 and 1.5. Therefore, in actual detection applications, the threshold F can be set to 1.8; In addition, since the vibration of the flame arrester in the pipeline is closely related to the gas flow, generally, the vibration intensity usually increases with the increase of the flow or flow velocity. Therefore, historical normal data when the hydrogen flow is zero or the hydrogen flow is small can be selected, and the threshold F can be set by calculating the corresponding energy entropy, so that the threshold F can reasonably screen the vibration information; when the total energy entropy is less than or equal to the threshold F, it means that the vibration information at the current sampling moment is similar to the historical normal data. Therefore, it can be determined that the vibration information at the current sampling moment is normal and the mechanical state of the flame arrester at the current sampling moment is normal, so that the relevant calculation process of the flow information at the current sampling moment can be omitted to reduce the amount of calculation and save computing resources.

[0027] Step S4: In response to the abnormal vibration characteristics, calculate the dynamic vibration threshold corresponding to the current sampling moment; When >F, it is preliminarily determined that the vibration information at the current sampling moment is abnormal. Since the change of the hydrogen flow will affect the detection of the vibration information, it is necessary to introduce the flow information at the current sampling moment to further judge the abnormal vibration information initially determined to be abnormal, exclude the influence of the flow change at the current sampling moment on the vibration information, and avoid false triggering of alarms caused by flow changes, so as to more accurately judge the mechanical state of the flame arrester at the current sampling moment; At the same time, since the vibration of the flame arrester is closely related to the hydrogen flow in the pipeline, generally, the vibration intensity usually increases with the increase of the flow or flow velocity, the vibration energy is positively correlated with the vibration intensity, and the vibration intensity is positively correlated with the hydrogen flow. Therefore, the relationship between the vibration energy E and the hydrogen flow Q can be established through a linear regression model; Specifically, to judge the abnormal mechanical state of the flame arrester by combining the flow information at the current sampling moment, it is necessary to obtain the relevant regression coefficients between the vibration energy value and the flow information based on historical normal data, including obtaining the decomposed signals of the vibration information in the historical normal data through the EMD algorithm, obtaining n historical IMF components and residuals, and calculating the historical energy values of each historical IMF component accordingly, and then establishing the relationship between the historical vibration information characteristics and the historical flow information characteristics through a linear regression model, so as to obtain the historical vibration energy corresponding to each historical IMF component and the historical hydrogen flow between the regression coefficients; According to the calculation formula of the energy value, since the energy value is obtained by normalizing the sum of the squares of M data, therefore, the M data corresponding to the energy value can be expanded into the form of a matrix sequence. At the same time, the flow data also has M data due to the same sampling frequency, and the flow data can be converted into the corresponding matrix sequence to facilitate the calculation of the regression coefficient.

[0028] The specific calculation process of the regression coefficient is as follows: Convert the historical vibration energy corresponding to each historical IMF component into the form of a matrix sequence and use them as the dependent variables of the linear regression model respectively. The matrix sequence expression of the i-th historical vibration energy is: Convert the corresponding historical flow information data into the form of a matrix sequence as the independent variable. The matrix sequence expression of the corresponding historical flow information data is: Establish the relationship between the historical vibration information characteristics and the historical flow information characteristics through the linear regression model, and obtain the regression coefficient between the historical vibration energy corresponding to each historical IMF component and the historical hydrogen flow ; Establish a linear regression model separately for each historical IMF component in the historical normal data. Among them, the linear regression model of the i-th historical vibration energy is: , where the value of i ranges from 1 to n, and represent the regression coefficients corresponding to the i-th historical IMF component; Use the least squares formula to calculate the regression coefficients corresponding to each historical IMF component respectively. Among them, the specific calculation formula for the regression coefficient corresponding to the i-th historical IMF component is: In the formula, represents the m-th data in the matrix sequence of the i-th historical vibration energy , represents the average value of the M data in the matrix sequence of the i-th historical vibration energy , represents the m-th data in the matrix sequence of the corresponding historical flow information data , represents the average value of the M data in the corresponding historical flow information .

[0029] In addition, after obtaining the regression coefficients corresponding to each historical IMF component, the theoretical energy value corresponding to each IMF component at the current sampling moment can be calculated based on the regression coefficients corresponding to each historical IMF component, so as to obtain the dynamic vibration threshold at the current sampling moment; The calculation process of the dynamic vibration threshold at the current sampling moment is as follows: According to the obtained regression coefficients, assuming that the current sampling moment is the t-th sampling moment, calculate the theoretical energy value of each IMF component at the t-th sampling moment. The calculation formula is: In the formula, represents the theoretical energy value of the i-th IMF component at the t-th sampling moment, represents the flow information data at the t-th sampling moment, and represents the regression coefficient corresponding to the i-th IMF component obtained based on historical normal data; The dynamic vibration threshold at the t-th sampling moment is , and the calculation formula is: where k is an error correction coefficient. Generally, the value range of k is 1.0 - 1.2. The value of k is related to the environment in the actual application scenario. The greater the environmental noise, the greater the value of k. On the contrary, the smaller the environmental noise, the closer the value of k is to 1; After that, since the flow information and the vibration information are positively correlated, in order to avoid the problem of logical confusion in multiple comparison and judgment, in the way of average value processing, compare and judge the average value of the energy values of each IMF component at the t-th sampling moment with the corresponding dynamic vibration threshold, so as to determine the mechanical state of the flame arrester at the current sampling moment; Obtain the average energy value of each IMF component at the current sampling moment as the comparison object of the dynamic vibration threshold at the current sampling moment. The calculation formula is: In the formula, represents the average energy value of n IMF components at the t-th sampling moment.

[0030] Step S5: Determine whether there is an abnormality in the operating state of the flame arrester at the current sampling moment; Specifically, judge the mechanical state of the flame arrester at the current sampling moment. Taking the t-th sampling moment as an example, when , it means that the vibration information at the current sampling moment is within the normal fluctuation range, and the mechanical state of the flame arrester is normal. When When it indicates that the vibration information at the current sampling moment is abnormal and the mechanical state of the flame arrester is abnormal, it is determined that the current state of the flame arrester is abnormal; Then, combining the detection information of temperature and hydrogen concentration, respectively judge whether the ambient temperature of the flame arrester is safe and whether there is hydrogen leakage. Compare the temperature and hydrogen concentration information detected at the current sampling moment with the set thresholds respectively. When any one of the temperature or hydrogen concentration is judged to be abnormal, it is determined that the current state of the flame arrester is abnormal.

[0031] Step S6: When the current state of the flame arrester is determined to be abnormal, trigger an abnormal alarm and record the corresponding abnormal information data; Specifically, when the abnormal alarm is triggered within a short time or continuously for S times, remind the relevant staff to perform manual processing. Generally, the value of S is 3. When the abnormal alarm is triggered within a short time or continuously for 3 times or more, remind the relevant staff that emergency processing is required. The relevant staff disassemble the flame arrester for manual inspection and confirmation, and then perform maintenance or replacement operations.

[0032] A hydrogen pipeline flame arrester operating state detection system, as Figure 2 shown, is used to implement the above-mentioned operating state detection method. Its hardware structure includes a processor 1 for processing data, a data memory 2 for storing abnormal information data, and a data buffer 3 for caching various sensing information data. The processor 1 is burned with computer program instructions for implementing the above-mentioned detection method. The data buffer 3 is electrically connected to a vibration sensor 4 for capturing the high-frequency mechanical vibration information of the flame arrester, a flow meter 5 for measuring the hydrogen flow rate in the pipeline in real time, a temperature sensor 6 for detecting the ambient temperature of the flame arrester accessories, and a hydrogen sensor 7 for detecting the hydrogen concentration of the flame arrester accessories.

[0033] Specifically, the vibration sensor 4 is installed at the vibration sensitive point of the flame arrester shell, and the sampling frequency of the vibration sensor 4 is greater than or equal to 10 kHz; the flow meter 5 is installed on the hydrogen pipeline corresponding to the flame arrester, and the sampling frequency of the flow meter 5 is the same as that of the vibration sensor 4.

[0034] The embodiments described in the present invention are only descriptions of the preferred embodiments of the present invention, and are not limited to the exact structures already described and shown in the drawings. Various modifications and changes can be made without departing from its protection scope; without departing from the design concept of the present invention, various variations and improvements made by those skilled in the art to the technical solutions of the present invention should all fall within the protection scope of the present invention.

Claims

1. A method for detecting the operating state of a hydrogen pipeline flame arrester, characterized in that: Obtain historical normal data and set threshold parameters corresponding to each detection information; obtain the detection information of each sensor in real time and perform preprocessing; perform time-frequency domain analysis on the preprocessed vibration information to obtain multiple IMF components and residuals, extract vibration characteristics for preliminary vibration anomaly judgment; in response to abnormal vibration characteristics, calculate the dynamic vibration threshold corresponding to the current sampling moment; determine the abnormal operation state of the flame arrester at the current sampling moment. Among them, the vibration characteristics include the energy value and the total energy entropy of the current vibration information. The energy of the $i$-th IMF component at the $t$-th sampling moment is ; the total energy entropy of the vibration information at the $t$-th sampling moment is , and the calculation formula is: , where, represents the energy proportion of the $i$-th IMF component at the $t$-th sampling moment, $\ln$ represents the natural logarithm, and $n$ represents the number of IMF components; the dynamic vibration threshold is , and the calculation formula is: , where $k$ is the error correction coefficient, represents the theoretical energy value of the $i$-th IMF component at the current sampling moment.

2. The operating state detection method according to claim 1, wherein: Obtain historical normal data and set threshold parameters corresponding to each detection information, including adaptively installing the corresponding sensors for detecting vibration, hydrogen flow rate, temperature, and hydrogen concentration respectively, and setting threshold parameters corresponding to each detection information according to the performance standards of the corresponding type of flame arrester and historical normal data. Among them, the threshold parameters of temperature and hydrogen concentration can be directly set through the performance standards of the flame arrester. The dynamic vibration threshold for judging the mechanical state of the flame arrester needs to obtain historical normal data on vibration and hydrogen flow rate for calculation. If relevant historical normal data is saved, it is directly imported and referenced. If no relevant historical data is saved, it is necessary to take the data record for a period of time under normal working conditions as historical normal data after manually detecting and confirming that the flame arrester is in a normal state.

3. The operating state detection method according to claim 2, characterized in that: Obtain the detection information of each sensor in real time, and preprocess the collected vibration information and hydrogen flow information, including synchronously collecting vibration information data and hydrogen flow information data at the same sampling frequency. Among them, the initial vibration information data at the t-th sampling moment is , and the flow information data at the t-th sampling moment is ; Preprocess the vibration information, including noise reduction processing for eliminating high-frequency electromagnetic interference in the vibration information; removing the trend term processing for eliminating low-frequency drift in the vibration information, such as sensor baseline drift and temperature influence; wherein, for the vibration information data at the t-th sampling moment After preprocessing, the standard vibration information data at the t-th sampling moment is obtained ; Perform preprocessing on the flow information, including synchronizing the time scale alignment to align the time stamp of the flow information with the time stamp of the vibration information.

4. The operating state detection method according to claim 3, characterized in that: Perform time-frequency domain analysis on the preprocessed vibration information to obtain multiple IMF components and residuals, including parsing the vibration information through the EMD algorithm to obtain the decomposition data of the vibration information. Among them, the preprocessed standard vibration information data at the t-th sampling moment Is decomposed from high frequency to low frequency into n IMF components and residuals, and the expression formula of the decomposition result is: Wherein, represents the i-th IMF component of the vibration information at the t-th sampling moment, represents the residual signal of the vibration information at the t-th sampling moment, and n represents the number of IMF components or the decomposition layer number.

5. The operating state detection method according to claim 4, characterized in that: Extract vibration characteristics for preliminary vibration anomaly judgment, including calculating the energy value of each IMF component at the t-th sampling moment , energy proportion and total energy entropy ; Denote the energy of the \(i\)-th IMF component at the \(t\)-th sampling moment, and the calculation formula is as follows: In the formula, represents the m-th decomposition data of the i-th IMF component, represents the summation function, and M represents the number of vibration information data at the current sampling moment; It represents the energy proportion of the $i$-th IMF component at the $t$-th sampling moment, and the calculation formula is as follows: Set a threshold value F. When ≤F, it is determined that the vibration information at the t-th sampling moment is normal, and the mechanical state of the flame arrester at the t-th sampling moment is normal. When >F, it is preliminarily determined that there is an abnormality in the vibration information at the t-th sampling moment, and it is necessary to further combine the flow information at the t-th sampling moment to make an abnormal judgment on the mechanical state of the flame arrester.

6. The operating state detection method according to claim 5, wherein: In response to abnormal vibration characteristics, calculate the dynamic vibration threshold corresponding to the current sampling moment, including obtaining the correlation regression coefficient between the vibration energy value and the flow information based on historical normal data, obtaining the decomposed signal of the vibration information in the historical normal data through the EMD algorithm, obtaining n historical IMF components and residuals, and calculating the energy value of each historical IMF component accordingly. Convert the historical vibration energy corresponding to each historical IMF component into the form of a matrix sequence and use them as the dependent variables of the linear regression model respectively. The matrix sequence expression of the historical vibration energy of the i-th is: Similarly, convert the corresponding historical traffic information data into the form of a matrix sequence as the independent variable, and the corresponding historical traffic information data The matrix sequence expression of is: Establish the relationship between the historical vibration information characteristics and the historical flow information characteristics through a linear regression model, and obtain the historical vibration energy corresponding to each historical IMF component and the historical hydrogen flow rate between the regression coefficients; A linear regression model is established separately for each historical IMF component in the historical normal data. Among them, the linear regression model of the i-th historical vibration energy is as follows: , where the value range of i is from 1 to n, and represent the regression coefficients corresponding to the i-th historical IMF component; Use the least squares formula to obtain the regression coefficient corresponding to each historical IMF component respectively. Among them, the specific calculation formula for the regression coefficient corresponding to the i-th historical IMF component is: In the formula, represents the m-th data in the matrix sequence of the i-th historical vibration energy , represents the average value of M data in the matrix sequence of the i-th historical vibration energy , represents the m-th data in the matrix sequence of the corresponding historical flow information data , represents the average value of M data in the corresponding historical flow information .

7. The operating state detection method according to claim 6, characterized in that: In response to abnormal vibration characteristics, calculating a dynamic vibration threshold corresponding to the current sampling moment, further including calculating a theoretical energy value corresponding to each IMF component at the current sampling moment according to the regression coefficient corresponding to each historical IMF component, wherein the theoretical energy value of the i-th IMF component at the t-th sampling moment is calculated by the formula: In the formula, represents the flow information data at the t-th sampling moment; Obtain the average energy of each IMF component at the current sampling moment as the dynamic vibration threshold at the current sampling moment as the comparison object, and the calculation formula is: , where represents the average energy of n IMF components at the t-th sampling moment.

8. The operating state detection method according to claim 7, wherein: Perform an abnormality determination on the operating state of the flame arrester at the current sampling moment, including judging the mechanical state of the flame arrester at the current sampling moment. Taking the t-th sampling moment as an example, when it indicates that the vibration information at the current sampling moment is within the normal fluctuation range, and the mechanical state of the flame arrester is normal. When it indicates that the vibration information at the current sampling moment is abnormal, and the mechanical state of the flame arrester is abnormal, then it is determined that the current state of the flame arrester is abnormal; Combined with the detection information of temperature and hydrogen concentration, compare and judge the temperature and hydrogen concentration information detected at the current sampling moment with the set thresholds respectively. When any one of the temperature or hydrogen concentration is judged to be abnormal, it is determined that the current state of the flame arrester is abnormal; when the current state of the flame arrester is determined to be abnormal, trigger an abnormal alarm and record the corresponding abnormal information data.

9. A hydrogen pipeline flame arrester operating state detection system for implementing the operating state detection method according to any one of claims 1 to 8, characterized in that: It includes a processor (1) for processing data, a data memory (2) for storing abnormal information data, and a data buffer (3) for caching data of each sensing information. The processor (1) is burned with computer program instructions for implementing the above detection method. The data buffer (3) is electrically connected to a vibration sensor (4) for capturing high-frequency mechanical vibration information of the flame arrester, a flowmeter (5) for measuring the hydrogen flow rate in the pipeline in real time, a temperature sensor (6) for detecting the ambient temperature of the flame arrester accessories, and a hydrogen sensor (7) for detecting the hydrogen concentration of the flame arrester accessories.

10. The operating state detection system according to claim 9, characterized in that: The vibration sensor (4) is installed at the vibration-sensitive point of the flame arrester housing, and the sampling frequency of the vibration sensor (4) is greater than or equal to 10 kHz; the flowmeter (5) is installed on the hydrogen pipeline corresponding to the flame arrester, and the sampling frequency of the flowmeter (5) is the same as that of the vibration sensor (4).

Images