A method for monitoring the state of a hydrogen compressor based on vibration parameters

Abstract

The present invention belongs to the technical field of hydrogen compressor monitoring, and particularly relates to a method for monitoring the state of a hydrogen compressor based on vibration parameters. The monitoring method includes: for any vibration parameter of the hydrogen compressor, according to the window parameters of the current sliding window, obtaining the vibration data of the current sliding window to calculate the statistical data and vibration feature vectors of the current sliding window; the window parameters, vibration data, statistical data, and vibration feature vectors of the current sliding window together constitute the current sliding window information; according to the sliding window information of each vibration parameter between adjacent monitoring times and the exhaust pressure at the current monitoring time, each clustering algorithm outputs the cluster result at the current monitoring time; at the same time, calculating the weight coefficients of each clustering algorithm at the current monitoring time; adding up the weight coefficients of the clustering algorithms corresponding to the same cluster result as the score of the corresponding cluster result, and then taking the cluster result with the highest score as the operating state of the hydrogen compressor at the current monitoring time. The present invention can improve the accuracy of monitoring the operating state of a hydrogen compressor based on vibration parameters.

Description

Technical Field

[0001] The present invention belongs to the technical field of hydrogen compressor monitoring, and particularly relates to a method for monitoring the state of a hydrogen compressor based on vibration parameters. Background Art

[0002] The hydrogen compressor (hereinafter referred to as "hydrogen compressor") is the core pressurization equipment for hydrogen refueling stations and hydrogen filling, and is an important link in the wide application of hydrogen energy. The design pressure of the hydrogen compressor is high, up to more than 90 MPa, and the gas circuit, water circuit, and oil circuit inside the hydrogen compressor are complex and intertwined, with a large number of various valves.

[0003] Due to the influence of assembly technology, transportation bumps, and the on-site operating environment, even a hydrogen compressor that has passed the inspection at the factory may malfunction during actual operation. This will not only affect the operating efficiency of the hydrogen compressor but also pose a threat to the personal safety of on-site technicians. Therefore, in addition to regularly overhauling and maintaining the key components in the hydrogen compressor, technicians will also, during the operation of the hydrogen compressor, map the condition of the hydrogen compressor based on the exhaust pressure collected by sensors through a certain clustering algorithm, that is, obtain the current working condition of the hydrogen compressor, so as to determine whether there is a problem with the current hydrogen compressor.

[0004] However, the relative movement and vibration of the internal components of the hydrogen compressor during operation will further exacerbate the performance degradation and internal part aging of the hydrogen compressor.

[0005] Therefore, in the existing technical methods for determining whether there is a problem with the hydrogen compressor, the following problems exist:

[0006] ① There are mutual influences among various vibration parameters of the hydrogen compressor, and they will affect the exhaust pressure in the future for a period of time. This makes the existing technical methods only able to obtain the current working condition of the hydrogen compressor, but unable to accurately predict the working condition of the hydrogen compressor in the future for a period of time.

[0007] ② When the clustering algorithm does not have the ability to autonomously generate cluster results (that is, the clustering algorithm is completely set by technicians for the cluster results, and the clustering algorithm will not generate new clustering clusters and corresponding cluster results): If a new working condition appears during the actual operation of the hydrogen compressor, and any of the set cluster results does not match the current working condition of the hydrogen compressor, then the current clustering algorithm will still output a cluster result set by technicians, that is, the flexibility of this clustering algorithm is poor and it cannot output the truly accurate working condition of the hydrogen compressor.

[0008] ③When the clustering algorithm has the ability to generate cluster results independently (that is, on the basis of the cluster results set by the technician, the clustering algorithm will also independently generate new clustering clusters and corresponding cluster results): as the number of collected samples increases, the cluster results will also continue to increase, even showing an explosive growth. Based on the set cluster results, the technician can directly take corresponding measures for the hydrogen compressor; however, the new cluster results need to be verified and analyzed by the technician before it can be confirmed whether measures need to be taken for the hydrogen compressor and what measures to take; and the verification and analysis by the technician are lagging, and this greatly increases the workload of the technician. In practical applications, after verification and analysis by the technician, it is found that: some new cluster results are completely wrong; some new cluster results are correct, but they can completely belong to a set cluster result, and only a small number of new cluster results are accurate and different from the set cluster results.

[0009] Therefore, how to improve the accuracy of monitoring the operating state of the hydrogen compressor based on vibration parameters has become an urgent problem to be solved in the monitoring technology of the hydrogen compressor. Summary of the Invention

[0010] The object of the present invention is to overcome the above-mentioned deficiencies of the prior art and provide a method for monitoring the state of a hydrogen compressor based on vibration parameters, which can improve the accuracy of monitoring the operating state of the hydrogen compressor based on vibration parameters.

[0011] To achieve the above object, the present invention adopts the following technical solutions:

[0012] A method for monitoring the state of a hydrogen compressor based on vibration parameters includes the following steps:

[0013] S1. For any vibration parameter of the hydrogen compressor: according to the window parameters of the current sliding window, obtain the vibration data of the current sliding window; the sliding windows are arranged in chronological order;

[0014] S2. Calculate the statistical data and vibration feature vectors of the current sliding window according to the vibration data of the current sliding window; the window parameters, vibration data, statistical data and vibration feature vectors of the current sliding window together constitute the current sliding window information;

[0015] S3. According to the sliding window information of each vibration parameter between the current monitoring moment and the previous monitoring moment and the exhaust pressure at the current monitoring moment, each clustering algorithm outputs the cluster result corresponding to the current monitoring moment; at the same time, calculate the weight coefficient of each clustering algorithm at the current monitoring moment;

[0016] S4. Accumulate the weight coefficients of the clustering algorithms corresponding to the same cluster result as the score of the corresponding cluster result, and then take the cluster result with the highest score as the operating state of the hydrogen compressor at the current monitoring moment.

[0017] Preferably, after S2, S2' is further included: S2', calculating window parameters of the next sliding window according to statistical data of the current sliding window and window parameters.

[0018] Preferably, when monitoring the current hydrogen compressor for the first time or during the first operation after replacing more than 1 key component of the current hydrogen compressor, for any vibration parameter of the current hydrogen compressor, a new monitoring cycle is entered, and the window parameters of the first sliding window in the new monitoring cycle are initial values; the window parameters include window length and sliding step; there are several vibration data arranged in time of only one vibration parameter in each sliding window.

[0019] Preferably, the following content is further included in S2':

[0020] Denote the current sliding window with window length , t being a positive integer, and the previous sliding window of the current sliding window is denoted as , then the window length of the next sliding window :

[0021] ;

[0022] ;

[0023] ;

[0024] ;

[0025] ;

[0026] wherein, represents the window length adjustment amount of the current sliding window ; represents the window length adjustment amount of the next sliding window ; f represents the rotational frequency of the component where the vibration parameter corresponding to the right boundary moment of the current sliding window is located; represents rounding up; is the first adjustment coefficient; is the second adjustment coefficient; represents the weight coefficient of the i-th clustering algorithm, and a total of n clustering algorithms are adopted, that is, 1 ≤ i ≤ n; represents the within-cluster entropy of the sliding window in the i-th clustering algorithm; represents the historical maximum entropy value of m sliding windows counted forward from the sliding window including the sliding window , m being a positive integer and m ≤ t; Denote the sliding window as the mean vector, the modulus length of which is the difference between the sliding window and the average value of the vibration data of the sliding window ; the direction of is from the average value of the vibration data of the sliding window to the average value of the vibration data of the sliding window ; Denote as the 2-norm; is the maximum value function; Denote the sliding window as the within-cluster entropy in the i-th clustering algorithm, Denote the sliding window as the within-cluster entropy in the i-th clustering algorithm; Denote the within-cluster entropy obtained in the corresponding clustering algorithm during the process of taking m sliding windows forward starting from the sliding window and including the sliding window when i ranges from 1 to n; Denote the covariance; Denote the variance; Denote the time constant of the hydrogen compressor system; Denote the current sliding window as the sampling period of the vibration data in it, which is determined by the acquisition frequency of the corresponding sensor; Denote the current sliding window as the maximum allowable drift rate of the vibration parameters; Denote the current sliding window as the rated value of the vibration parameters.

[0027] Preferably, in S2, calculating the vibration feature vector of the current sliding window based on the vibration data of the current sliding window further includes the following sub-steps:

[0028] S21, perform local mean decomposition on the vibration parameter signal corresponding to the vibration parameters in the current sliding window:

[0029] ;

[0030] wherein, T represents time; X(T) represents the vibration parameter signal in the current sliding window. Fit the N vibration data included in the current sliding window into a vibration parameter signal curve, and the vibration parameter signal represented by this vibration parameter signal curve is X(T); Denote the k-th PF component, k = 1,..., K, and K is a positive integer greater than 1; Denote the K-th residue, is a monotonic function;

[0031] S22. Extract the characteristic parameters of each PF component to form the vibration feature vector of the current sliding window:

[0032] ;

[0033] ;

[0034] ;

[0035] ;

[0036] Among them, represents the skewness coefficient of the k-th PF component ; represents the kurtosis coefficient of the k-th PF component ; represents the energy of the k-th PF component ; represents the energy ratio of the k-th PF component in the vibration parameter signal X(T); represents the h-th vibration data arranged in chronological order in the current sliding window; represents the average value of the N PF component values with the same time points as the N vibration data in the current sliding window; represents the standard deviation of the N PF component values with the same time points as the N vibration data in the current sliding window; a represents the time point of the left boundary of the current sliding window; b represents the time point of the right boundary of the current sliding window; E represents the energy of the vibration parameter signal X(T);

[0037] S23. Based on the characteristic parameters of the K PF components, obtain the K vibration feature vectors of the current sliding window:

[0038] ;

[0039] Among them, represents the k-th vibration feature vector of the current sliding window.

[0040] Preferably, the sliding step of the next sliding window is:

[0041] ;

[0042] ;

[0043] ;

[0044] Among them, represents the next sliding window​ The overlapping ratio; Indicates the current sliding window The overlapping ratio; Indicates the next sliding window The attenuation coefficient; Indicates the target overlapping ratio; Indicates the total number of sliding windows that need to be transitioned currently; and Both are known quantities determined by the operating state of the hydrogen compressor at the previous monitoring moment; Indicates including the next sliding window Inside, and the remaining number of sliding windows that need to be transitioned, , Indicates including the current sliding window Inside, and the remaining number of sliding windows that need to be transitioned.

[0045] Preferably, in S3, the following content is also included:

[0046] Adopt n different clustering algorithms. Among these n clustering algorithms, some clustering algorithms do not have the ability to independently generate cluster results, and some clustering algorithms have the ability to independently generate cluster results; input the sliding window information of each vibration parameter between the current monitoring moment and the previous monitoring moment and the exhaust pressure at the current monitoring moment into each clustering algorithm, and each clustering algorithm respectively outputs the cluster result corresponding to the hydrogen compressor at the current monitoring moment; at the same time, calculate the weight coefficients of each clustering algorithm at the current monitoring moment , 1 ≤ i ≤ n and i is a positive integer: If the current monitoring moment is the first monitoring moment within the current monitoring period, then obtain the weight coefficients of each clustering algorithm at the current monitoring moment based on solving the multi-objective optimization function; if the current monitoring moment is not the first monitoring moment within the current monitoring period, then obtain the weight coefficients of each clustering algorithm at the current monitoring moment based on the weight coefficients of each clustering algorithm at the previous monitoring moment.

[0047] Preferably, to obtain the weight coefficients of each clustering algorithm at the current monitoring moment based on solving the multi-objective optimization function, the following content is also included: After constructing the multi-objective optimization function F and solving it, obtain the values of the weight coefficients of each clustering algorithm at the current monitoring moment:

[0048] ;

[0049] ;

[0050] The constraint condition is: ;

[0051] Among them, Is the second constant parameter; Indicates the weight coefficient corresponding to the i-th clustering algorithm; Denote the sum of squared errors of the \(i\)-th clustering algorithm; Denote the first weight matrix, which is a column vector composed of weight coefficients; Denote the second weight matrix, which is the transpose of the first weight matrix and is a row vector; Denote the first weight matrix and the quadratic form of the correlation matrix ; is the regularization parameter.

[0052] Preferably, based on the weight coefficients of each clustering algorithm at the previous monitoring moment, to obtain the weight coefficients of each clustering algorithm at the current monitoring moment, it further includes the following content:

[0053] Denote the current monitoring moment as , the previous monitoring moment as , \(z\geq2\) and \(z\) is a positive integer, then the weight coefficient of the \(i\)-th clustering algorithm at the current monitoring moment is:

[0054] ;

[0055] wherein, denotes the weight coefficient of the \(i\)-th clustering algorithm at the previous monitoring moment; denotes the weight coefficient of the \(j\)-th clustering algorithm at the previous monitoring moment, \(1\leq j\leq n\) and \(j\) is a positive integer; denotes the learning rate; denotes the global distribution at the previous monitoring moment; denotes the distribution of the \(i\)-th clustering algorithm at the previous monitoring moment; denotes the distribution of the \(j\)-th clustering algorithm at the previous monitoring moment; denotes the JS divergence; denotes the JS divergence between the distribution and the global distribution ; denotes the JS divergence between the distribution and the global distribution ;

[0056] Preferably, after S4, S5 is further included: S5, if the cluster result obtained by the clustering algorithm with the ability to autonomously generate cluster results does not become the cluster result with the highest score, then the cluster result obtained by the clustering algorithm with the ability to autonomously generate cluster results is recorded as a set to be verified. After technicians regularly verify and analyze the set to be verified, if the cluster result obtained by a certain clustering algorithm in the set to be verified is considered correct by technicians for y consecutive times and is different from the cluster result set by technicians, then the technicians increase the weight coefficient of the corresponding clustering algorithm.

[0057] The beneficial effects of the present invention are as follows:

[0058] (1) The hydrogen compressor status monitoring method of the present invention can improve the accuracy of monitoring the operating status of the hydrogen compressor based on vibration parameters.

[0059] (2) A single parameter cannot effectively achieve the monitoring and early warning of equipment. The hydrogen compressor status monitoring method of the present invention does not directly process the single type of data collected by the sensor, but takes the sliding window as the unit for collecting data and performs clustering processing on the sliding window information between two adjacent monitoring moments. For any vibration parameter, the vibration data in a sliding window is time-sequential, and there is also time-sequentiality between adjacent sliding windows (a new sliding window can only be born as time goes by); there is also partial overlap of vibration data between adjacent sliding windows, which enables the clustering algorithm to well mine and learn the potential relationships and correlations between adjacent sliding windows when processing the time-sequential sliding window information, and improve the accuracy of the cluster results obtained by each clustering algorithm in the present invention.

[0060] (3) In the hydrogen compressor status monitoring method of the present invention, a sliding window only contains one type of vibration parameter. The sliding windows of different vibration parameters within the same time period overlap in time because the sensor collects data at the same time, so the sliding windows also overlap in time. The clustering algorithm in the present invention is not based on the isolated vibration data at a certain moment; instead, it is based on this time-sequential, time-related and overlapping vibration data sliding window information, and there is also time overlap between the sliding windows of different vibration parameters. The clustering algorithm will also mine and learn the potential relationships and correlations between the sliding window information of these different vibration parameters with time overlap and the exhaust pressure, which enables each clustering algorithm in the present invention to more accurately predict the working conditions of the current hydrogen compressor in the next period of time, thereby improving the accuracy of the cluster results output by each clustering algorithm; this also greatly improves the accuracy of the new clustering clusters and the corresponding cluster results obtained by the clustering algorithm with the ability to autonomously generate cluster results.

[0061] (4) The window length and sliding step of the sliding window are directly related to the sample data input into each clustering algorithm, and indirectly related to the accuracy of the cluster results output by each clustering algorithm. In each monitoring period of the present invention, the window length and sliding step are automatically adjusted based on the previous sliding window, according to the previous several sliding windows and the intra-cluster entropy obtained by each clustering algorithm based on the previous several sliding windows, without manual intervention. Technicians only need to set the window parameters (i.e., the initial values) of the first sliding window in each monitoring period, which greatly reduces the labor cost.

[0062] (5) In the entire monitoring process of the present invention, the adjustment of the window length and sliding step of the sliding window not only has a high degree of automation, but also balances the computational cost and storage resource occupancy during the process of processing the sliding window information. At the same time, it enables the clustering algorithm to better mine and learn the potential relationships and associations between sliding windows, allowing each clustering algorithm to better process the sliding window information and effectively perceive the long-term slowly deteriorating monitoring object.

[0063] (6) The sliding window information of the present invention includes the vibration feature vector of the sliding window, including the time-frequency distribution result of the vibration data within the sliding window, the distribution of signal energy at different spatial scales, and the sensitivity coefficient related to the surface damage fault of the machine body, further improving the sensitivity of each clustering algorithm to the state of the hydrogen compressor.

[0064] (7) The monitoring method of the present invention will automatically update the monitoring period, which is reflected in: when monitoring the current hydrogen compressor for the first time, or in the first run after replacing one or more key components of the current hydrogen compressor, for any vibration parameter of the current hydrogen compressor, a new monitoring period is entered. Even for hydrogen compressors of the same model, although they pass the pre-factory test, there are certain pre-factory test data; however, the period during which the hydrogen compressor is officially put into operation after being transported to the site and installed is the period monitored by the monitoring method of the present invention, that is, when monitoring a hydrogen compressor for the first time. Therefore, before monitoring a hydrogen compressor for the first time, the hydrogen compressor may still have some parts damaged or installation mismatches due to transportation and on-site installation; the same is true for the first run after replacing one or more key components of the current hydrogen compressor; these two situations mean that the hydrogen compressor as a whole needs to go through the running-in period, the stable period, and the deterioration period again. And in each monitoring period, the window length and sliding step are based on the previous sliding window. Therefore, the window length and sliding step of the new sliding window in these two situations cannot be obtained based on the previous sliding window. Therefore, we need to let the hydrogen compressor enter a new monitoring period in these two situations.

[0065] (8) The update of the window length and the sliding step of the sliding window is crucial for the subsequent clustering algorithm in terms of the frequency of sample input and the change in the amount of sample data. The monitoring method of the present invention makes the frequency of sample input and the amount of sample data adapt to the subsequent clustering algorithm processing process by automatically updating the monitoring period and automatically adjusting the window length and the sliding step of the sliding window within each monitoring period, further improving the efficiency of the entire monitoring process and the accuracy of the cluster results output by the clustering algorithm. Therefore, the monitoring method of the present invention can better adapt to each hydrogen compressor it monitors.

[0066] (9) The hydrogen compressor status monitoring method of the present invention does not use a single clustering algorithm to monitor the operating status of the hydrogen compressor, but uses multiple clustering algorithms to jointly monitor. Moreover, these clustering algorithms not only include clustering algorithms with the ability to independently generate cluster results, but also include clustering algorithms without the ability to independently generate cluster results. Some of the cluster results obtained by these clustering algorithms may be the same, or they may all be different. However, the present invention finally uses the cluster result with the highest score as the operating status of the hydrogen compressor at the current monitoring moment. The cluster result with the highest score is also the cluster result of most clustering algorithms. After being randomly inspected and verified by technicians, it is basically the operating status of the hydrogen compressor at the current monitoring moment.

[0067] (10) The score of the cluster result is obtained by accumulating the weight coefficients of the clustering algorithms corresponding to the same cluster result. The weight coefficients in the monitoring method of the present invention also change with the monitoring period and the monitoring moment within each monitoring period: the weight coefficients of each clustering algorithm at the first monitoring moment within each monitoring period are determined by solving a multi-objective optimization function; while the weight coefficients of each clustering algorithm at the remaining monitoring moments within each monitoring period need to be obtained based on the weight coefficients of each clustering algorithm at the previous monitoring moment. This makes the weight coefficients of each clustering algorithm not mutate while changing with the monitoring moment. And after being verified by technicians, only when a certain clustering algorithm continuously obtains several cluster results that are all the operating status of the hydrogen compressor at the corresponding monitoring moment, the weight coefficient of this clustering algorithm will gradually increase.

[0068] (11) The method for monitoring the state of the hydrogen compressor of the present invention also takes into account a special case with an extremely low probability: for a clustering algorithm with the ability to autonomously generate cluster results, if the cluster result corresponding to the current monitoring moment is a new cluster result and the weight coefficient of the clustering algorithm for generating this new cluster result is small, then this new cluster result cannot be used as the operating state of the hydrogen compressor at the current monitoring moment; and the content of S5 is correspondingly adopted. In this way, on the one hand, it can prevent the clustering algorithm with the ability to autonomously generate cluster results from generating multiple new cluster results without restraint to interfere with the technician's evaluation of the operating state of the hydrogen compressor, greatly reducing the workload of the technician. On the other hand, it can also ensure through a small amount of manual intervention that when the hydrogen compressor really shows a situation outside the cluster results preset by the technician, the clustering algorithm with a high weight coefficient and the ability to autonomously generate cluster results can accurately determine the operating state of the hydrogen compressor at the current monitoring moment.

[0069] (12) In the method for monitoring the state of the hydrogen compressor of the present invention, the operating state of the hydrogen compressor at the current monitoring moment can finally be accurately determined. The operating state of the hydrogen compressor not only includes the working condition of the hydrogen compressor at the current monitoring moment but also includes the prediction of the working condition of the current hydrogen compressor in a period of time in the future; therefore, the technician can preset a larger number of cluster results: for example, among multiple cluster results, the determination of the working condition of the hydrogen compressor at the current monitoring moment is the same, but the prediction of the working condition of the current hydrogen compressor in a period of time in the future is different. This is directly different from the prior art in which the working condition of the hydrogen compressor at the current monitoring moment is determined based on the exhaust pressure collected by the sensor; and the monitoring method of the present invention will not be prone to misjudgment as in the prior art due to the technician presetting too many cluster results. Description of the Drawings

[0070] Figure 1 is a flowchart of a method for monitoring the state of a hydrogen compressor based on vibration parameters according to the present invention;

[0071] Figure 2 is the clustering result obtained by using the GMM clustering algorithm;

[0072] Figure 3 is the clustering result obtained by using the DBSCAN clustering algorithm;

[0073] Figure 4 is the clustering result obtained by using the OPTICS clustering algorithm;

[0074] Figure 5 is the clustering result obtained by using the method for monitoring the state of the hydrogen compressor of the present invention. Detailed Embodiments

[0075] To make the technical solution of the present invention clearer and more definite, the present invention will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Any equivalent replacement and conventional reasoning of the technical features of the technical solution of the present invention by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0076] As Figure 1 shown, it is a flowchart of a hydrogen compressor status monitoring method based on vibration parameters according to the present invention, including the following steps:

[0077] S1. For any vibration parameter of the hydrogen compressor: obtain the vibration data of the current sliding window according to the window parameters of the current sliding window; the sliding windows are arranged in chronological order;

[0078] S2. Calculate the statistical data and vibration feature vectors of the current sliding window according to the vibration data of the current sliding window; the window parameters, vibration data, statistical data, and vibration feature vectors of the current sliding window together constitute the current sliding window information;

[0079] S3. According to the sliding window information of each vibration parameter between the current monitoring moment and the previous monitoring moment and the exhaust pressure at the current monitoring moment, each clustering algorithm outputs the cluster result corresponding to the current monitoring moment; at the same time, calculate the weight coefficient of each clustering algorithm at the current monitoring moment;

[0080] S4. Accumulate the weight coefficients of the clustering algorithms corresponding to the same cluster result as the score of the corresponding cluster result, and then take the cluster result with the highest score as the operating state of the hydrogen compressor at the current monitoring moment.

[0081] After S2, it further includes S2':

[0082] S2'. Calculate the window parameters of the next sliding window according to the statistical data and window parameters of the current sliding window.

[0083] In S1, it further includes the following content:

[0084] The vibration parameters of the hydrogen compressor are classified by components and include several categories such as the vibration parameters of the south / north suction valves, the vibration parameters of the south / north exhaust valves, the vibration parameters of the crankshaft, the vibration parameters of the motor, the vibration parameters of the body, and the vibration parameters of the bearings. Specifically, it includes the rotational frequency of the compressor, the rotational angle of the crankshaft, the vibration amplitudes of the south exhaust valve in the x, y, and z directions, etc.

[0085] The key components of the hydrogen compressor include a diaphragm assembly, a valve assembly, a crankshaft mechanism, a sealing system, a cooling system, a driving motor, and a control system, etc. Among them, the diaphragm assembly is the core power transmission component. Any changes in the diaphragm material, thickness, number of layers, etc. are regarded as replacing the diaphragm assembly; the valve assembly is one of the most critical components, including an intake valve / exhaust valve. Any changes in the valve plate material, spring stiffness, valve clearance, etc. are regarded as replacing the valve assembly; the crankshaft mechanism is the core of power conversion. Any changes in its clearance, counterweight, etc. are regarded as replacing the crankshaft mechanism; the sealing system includes static seals and dynamic seals. Any changes in its packing material and size, etc. are regarded as replacing the sealing system; in the cooling system, a change in the nature of the cooling medium is regarded as replacing the cooling system; in the driving motor and control system, any changes in the motor efficiency, power, PID control parameters, etc. are regarded as replacing the driving motor and control system.

[0086] When the current hydrogen compressor is monitored for the first time, or during the first operation after replacing one or more key components of the current hydrogen compressor, for any vibration parameter of the current hydrogen compressor, a new monitoring cycle has started, and the window parameters of the first sliding window in the new monitoring cycle are all initial values. The window parameters include the window length and the sliding step size. The initial values of the window parameters are set by technicians, and the initial window values for different vibration parameters can be different.

[0087] Except for the above situations, no matter how many times the current hydrogen compressor is restarted after shutdown, for any vibration parameter of the hydrogen compressor, it is in the same monitoring cycle. In the same monitoring cycle, for a certain vibration parameter, assuming that the last sliding window during the previous operation of the current hydrogen compressor is the 51st sliding window in the current cycle, then the first sliding window after the current hydrogen compressor is restarted is the 52nd sliding window in the current cycle.

[0088] Each sliding window contains a number of vibration data arranged in time for only one vibration parameter. There is an overlap of a number of vibration data between adjacent sliding windows of the same vibration parameter.

[0089] The sliding direction of the sliding window is unique and synchronized with the time sequence; the sliding window slides one sliding step size each time. The sliding step size is a time quantity. When the sliding window completes one slide, it becomes a new sliding window. Each time the sliding window slides, its sliding step size is the sliding step size in the window parameters of the current sliding window.

[0090] For the same vibration parameter: the window length and the window sliding step size will directly affect the number of overlapping vibration data between adjacent sliding windows. The window length directly determines the number of vibration data in that sliding window.

[0091] Vibration data is collected regularly by the corresponding sensors, and the acquisition frequency of the sensors is generally fixed.

[0092] In S2:

[0093] The statistical data of the current sliding window include: the mean value of the vibration data, the variance of the vibration data, the standard deviation of the vibration data, the median of the vibration data, the first quartile of the vibration data, the third quartile of the vibration data, the interquartile range of the vibration data, the skewness of the vibration data, the kurtosis of the vibration data, the maximum value of the vibration data, the minimum value of the vibration data, and the coefficient of variation of the vibration data.

[0094] "Calculating the vibration feature vector of the current sliding window based on the vibration data of the current sliding window" further includes sub-steps S21 to S23:

[0095] S21, perform local mean decomposition on the vibration parameter signal corresponding to the vibration parameters in the current sliding window:

[0096] ;

[0097] Where, T represents time; X(T) represents the vibration parameter signal in the current sliding window. There are N vibration data in the current sliding window, and these N vibration data are fitted into a vibration parameter signal curve, then the vibration parameter signal represented by this vibration parameter signal curve is X(T); represents the k-th PF component, k = 1,..., K, and K is a positive integer greater than 1; represents the K-th residue; Determination of K: In the process of performing local mean decomposition on the vibration parameter signal X(T), the first PF component will be obtained first and the first residue , when the first residue is not a monotonic function, then continue to decompose the first residue into the second PF component and the second residue , until the K-th residue is a monotonic function.

[0098] Therefore, after local mean decomposition of the vibration parameter signal X(T), it is expressed as the sum of K PF components and 1 residue.

[0099] Performing local mean decomposition on the vibration parameter signal, obtaining multiple groups of PF components with instantaneous physical meanings, obtaining time-frequency distribution information, thereby reflecting the time-frequency distribution result of the vibration parameter signal and the distribution of signal energy at different spatial scales.

[0100] S22, extract the characteristic parameters of each PF component to form the vibration feature vector of the current sliding window:

[0101] ;

[0102] ;

[0103] ;

[0104] ;

[0105] Among them, represents the skewness coefficient of the k-th PF component ; represents the kurtosis coefficient of the k-th PF component ; represents the energy of the k-th PF component ; represents the energy ratio of the k-th PF component in the vibration parameter signal X(T); represents the h-th vibration data arranged in chronological order in the current sliding window; represents the average value of the N PF component values in that are the same as the time points of the N vibration data in the current sliding window; represents the standard deviation of the N PF component values in

[0106] that are the same as the time points of the N vibration data in the current sliding window; a represents the time point of the left boundary of the current sliding window; b represents the time point of the right boundary of the current sliding window; E represents the energy of the vibration parameter signal X(T).

[0107] S23, based on the characteristic parameters of K PF components, obtain K vibration feature vectors of the current sliding window:

[0108] ;

[0109] Among them, represents the k-th vibration feature vector of the current sliding window.

[0110] The vibration feature vector of the sliding window includes the time-frequency distribution result of the vibration data in the sliding window, the distribution of signal energy on different spatial scales, and the sensitivity coefficient (i.e., kurtosis coefficient) to the damage fault on the surface of the machine body.

[0111] In S2' also includes the following content:

[0112] The window length of the sliding window covers more than 1 integer period of the corresponding vibration parameter;

[0113] Denote the current sliding window The window length of is t, where t is a positive integer, and the current sliding window is denoted as , then the next sliding window has a window length :

[0114] ;

[0115] ;

[0116] ;

[0117] ;

[0118] ;

[0119] where represents the window length adjustment amount of the current sliding window ; represents the window length adjustment amount of the next sliding window ; f represents the rotational frequency of the component where the vibration parameter corresponding to the current sliding window is located, which is the rotational frequency of the component where the vibration parameter is located at the right boundary moment of the current sliding window ; represents rounding up; is the first adjustment coefficient; is the second adjustment coefficient; represents the weight coefficient of the i-th clustering algorithm. In the present invention, a total of n clustering algorithms are adopted, and 1 ≤ i ≤ n; represents the within-cluster entropy of the sliding window in the i-th clustering algorithm; represents the historical maximum entropy value of m sliding windows counted forward from the sliding window (including the sliding window ), where m is a positive integer and m ≤ t; represents the mean vector of the sliding window , and the modulus of is the difference between the sliding window and the average value of the vibration data of the sliding window , and the direction of points from the average value of the vibration data of the sliding window to the average value of the vibration data of the sliding window ; represents the 2-norm of is the maximum value function; represents the within-cluster entropy of the sliding window in the i-th clustering algorithm, represents the sliding window Intra - cluster entropy in the i - th clustering algorithm It represents the intra - cluster entropy in the corresponding clustering algorithm obtained during the process of taking i from 1 to n for m sliding windows counted forward from the start of the sliding window (including the sliding window). It represents covariance It represents variance It represents the time constant of the hydrogen compressor system It represents the current sliding window The sampling period of the vibration data in it, which is determined by the acquisition frequency of the corresponding sensor It represents the current sliding window The maximum allowable drift rate of the vibration parameters of it, which is set by technicians according to experience It represents the current sliding window The rated value of the vibration parameters of it

[0120] If 1 = t < m, that is, there are only t sliding windows counted forward from the start of the sliding window (including the sliding window ), then ; if 2 = t < m, then ; if 3 = t < m, then ; if 3 = t < m, then .

[0121] Optionally, calculate the sliding step size of the next sliding window : = , where γ is the first constant parameter, which is defined by technicians themselves. In this embodiment, γ = 0.2

[0122] Optionally, calculate the sliding step size of the next sliding window :

[0123] ;

[0124] ;

[0125] ;

[0126] where represents the overlap ratio of the next sliding window represents the overlap ratio of the current sliding window represents the attenuation coefficient of the next sliding window represents the target overlap ratio represents the total number of sliding windows to be transitioned currently and All are known quantities determined by the operating state of the hydrogen compressor at the previous monitoring moment; Indicates including the next sliding window Among them, and the remaining number of sliding windows to be transitioned, , Indicates including the current sliding window Among them, and the remaining number of sliding windows to be transitioned.

[0127] In the process of the present invention determining "the sliding step of the next sliding window ": Based on the smooth transition algorithm of state transition, a state transition equation is constructed, that is And And ; Timely adjustment of the sliding step of the next sliding window is to timely adjust the number of overlapping vibration data between adjacent sliding windows; and through the transition of continuously adjusting several sliding steps, that is, several sliding windows gradually change the sliding step, a huge adjustment of the sliding step can be achieved in stages, avoiding the influence on the mining and learning of the potential relationship and relevance between adjacent sliding windows by the subsequent clustering algorithm due to the too large change in the sliding step between adjacent sliding windows.

[0128] The sliding window Is the first sliding window.

[0129] In this example, if the average value of the vibration data of the sliding window Is greater than the average value of the vibration data of the sliding window , then The direction is positive, otherwise it is negative.

[0130] During each sliding of the sliding window, the window length is adjusted, that is, when the sliding window Completes one sliding, the window length is also adjusted, and the window length of the next sliding window Formed is . Only by adjusting the right boundary of the sliding window can the window length of the sliding window be adjusted; the left boundary of the sliding window does not need to be adjusted, and the distance between the left boundaries of adjacent sliding windows is a sliding step.

[0131] In the calculation of the intra-cluster entropy, if the clustering algorithm is the Kmeans++ clustering algorithm, hierarchical clustering algorithm, DBSCAN clustering algorithm, GMM clustering algorithm, OPTICS clustering algorithm, etc., then calculate the intra-cluster entropy based on distance; while for the spectral clustering algorithm, calculate the intra-cluster entropy based on similarity.

[0132] In S3, the following content is also included:

[0133] Adopt n different clustering algorithms. Among these n clustering algorithms, some do not have the ability to autonomously generate cluster results, while others do. Input the sliding window information of each vibration state parameter between the current monitoring moment and the previous monitoring moment, as well as the exhaust pressure at the current monitoring moment, into each clustering algorithm. Each clustering algorithm respectively outputs the cluster result corresponding to the hydrogen compressor at the current monitoring moment.

[0134] Simultaneously calculate the weight coefficients of each clustering algorithm at the current monitoring moment. , 1 ≤ i ≤ n and i is a positive integer: If the current monitoring moment is the first monitoring moment within the current monitoring period, then obtain the weight coefficients of each clustering algorithm at the current monitoring moment based on solving the multi-objective optimization function. If the current monitoring moment is not the first monitoring moment within the current monitoring period, then obtain the weight coefficients of each clustering algorithm at the current monitoring moment based on the weight coefficients of each clustering algorithm at the previous monitoring moment.

[0135] The exhaust pressure is collected by a pressure sensor.

[0136] In this embodiment, the sliding window information of each vibration parameter between the current monitoring moment and the previous monitoring moment is respectively input into each clustering algorithm. This sliding window refers to the sliding window whose right boundary is between the current monitoring moment and the previous monitoring moment.

[0137] In this embodiment, 10 different clustering algorithms are adopted, namely the AP clustering algorithm (Affinity Propagation clustering algorithm), Agglomerative clustering algorithm (Agglomerative hierarchical clustering algorithm), BIRCH clustering algorithm (Balanced Iterative Reducing and Clustering using Hierarchies algorithm), DBSCAN clustering algorithm (Density-Based Spatial Clustering of Applications with Noise algorithm), GMM clustering algorithm (Gaussian Mixture Model clustering algorithm), Divisive clustering algorithm (Divisive hierarchical clustering algorithm), Kmeans++ clustering algorithm (K-Means++ algorithm), MeanShift clustering algorithm (Mean Shift algorithm), OPTICS clustering algorithm (Ordering Points To Identify the Clustering Structure algorithm), and Spectral clustering algorithm (Spectral clustering algorithm); corresponding to the weight coefficients in turn ~ 。

[0138] Among them, the Agglomerative clustering algorithm, GMM clustering algorithm, Divisive clustering algorithm, Kmeans++ clustering algorithm, and Spectral clustering algorithm do not have the ability to autonomously generate cluster results; the AP clustering algorithm, BIRCH clustering algorithm, DBSCAN clustering algorithm, MeanShift clustering algorithm, and OPTICS clustering algorithm have the ability to autonomously generate cluster results.

[0139] "Obtaining the weight coefficients of each clustering algorithm at the current monitoring moment based on solving a multi-objective optimization function" further includes the following content:

[0140] Based on the Lyapunov stability theory, after constructing the multi-objective optimization function F and solving it, the values of the weight coefficients of each clustering algorithm at the current monitoring moment are obtained:

[0141] ;

[0142] ;

[0143] The constraint conditions are: ;

[0144] Among them, is the second constant parameter; represents the weight coefficient corresponding to the i-th clustering algorithm; represents the sum of squared errors of the i-th clustering algorithm; represents the first weight matrix, which is a column vector composed of weight coefficients; represents the second weight matrix, which is the transpose of the first weight matrix and is a row vector; represents the first weight matrix and the correlation matrix of the quadratic form; is the regularization parameter.

[0145] The correlation matrix is used to describe the relationship between different clustering algorithms; the regularization parameter is set by the technical personnel and is used to control the influence of the correlation between clustering algorithms.

[0146] In this embodiment, for the multi-objective optimization function F, the CVXPY tool is used to convert it into a convex optimization, and then the built-in solvers ECOS and SCS are called for solution.

[0147] "Obtaining the weight coefficients of each clustering algorithm at the current monitoring moment based on the weight coefficients of each clustering algorithm at the previous monitoring moment" further includes the following content:

[0148] Denote the current monitoring moment as , the previous monitoring moment as , z≥2 and z is a positive integer, then the weight coefficient of the i-th clustering algorithm at the current monitoring moment is:

[0149] ;

[0150] Among them, represents the previous monitoring moment The weight coefficient of the i-th clustering algorithm; Indicates the previous monitoring moment The weight coefficient of the j-th clustering algorithm; 1 ≤ j ≤ n and j is a positive integer; Indicates the learning rate, which is set by the technician and used to control the update speed of the algorithm weight coefficient. When it increases, it means that the algorithm adjusts the weight coefficient more actively to adapt to the distribution faster; Indicates the global distribution at the previous monitoring moment. The global distribution at the previous monitoring moment can be obtained by synthesizing the vibration data recorded during the current monitoring period and the weight coefficients of each algorithm, and it is a known quantity; Indicates the distribution of the i-th clustering algorithm at the previous monitoring moment; Indicates the distribution of the j-th clustering algorithm at the previous monitoring moment; Indicates the JS divergence; Indicates the distribution And the global distribution The JS divergence of, which is used to measure And The difference. The more similar the distribution And the global distribution The more similar, then The value is larger; Indicates the distribution And the global distribution The JS divergence of, which is used to measure And The difference. The more similar the distribution And the global distribution The more similar, then The value is larger.

[0151] After S4, it further includes S5:

[0152] S5. If the cluster result obtained by the clustering algorithm with the ability to autonomously generate cluster results does not become the cluster result with the highest score, then the cluster result obtained by the clustering algorithm with the ability to autonomously generate cluster results is recorded as a set to be verified. After the technician regularly verifies and analyzes the set to be verified, if the cluster result obtained by a certain clustering algorithm in the set to be verified is continuously considered correct by the technician for y times and is different from the cluster result set by the technician, then the technician increases the weight coefficient of the corresponding clustering algorithm.

[0153] In this embodiment, y = 5.

[0154] In the prior art, a single type of data collected by a sensor, such as exhaust pressure, is directly processed. Since the data at the current time point is processed through clustering to obtain a cluster result, and the corresponding cluster result is output as the current working condition of the hydrogen compressor. Therefore, the prior art generally can only be used to determine the current working condition of a certain hydrogen compressor; and this determination of the current working condition of a certain hydrogen compressor is also affected by the clustering algorithm itself. This influence will directly manifest as inaccurate judgment of the current working condition of the hydrogen compressor, which we have already explained in the background art and will not be elaborated here.

[0155] Suppose the set cluster result 1 by the technician is "the hydrogen compressor is continuously operating normally", and the cluster result 2 is "the hydrogen compressor is currently operating normally, but in a low-pressure operating state". It can be seen that cluster result 1 not only contains the determination of the current working condition "the hydrogen compressor is operating normally", but also contains the prediction of the working condition of the current hydrogen compressor in a future period of time "continuously operating normally". In cluster result 2, it not only contains the determination of the current working condition of the hydrogen compressor "the hydrogen compressor is operating normally", but also contains the prediction of the working condition of the current hydrogen compressor in a future period of time "in a low-pressure operating state"; when one or several components in the hydrogen compressor are about to enter the fatigue state, it will cause the hydrogen compressor to currently operate normally, but in a future period of time, the corresponding components will enter the fatigue state, resulting in the hydrogen compressor entering the low-pressure operating state. Although the compressor in the low-pressure operating state is also working properly, the components in the fatigue state may be damaged, causing the hydrogen compressor to stop operating. In the prior art clustering algorithms (regardless of whether they have the ability to generate cluster results autonomously), since the data at the current time point is processed through clustering, it cannot well predict the working condition of the current hydrogen compressor in a future period of time (for example, the current hydrogen compressor is indeed "in a low-pressure operating state" in a future period of time, but the prior art determines that it will "continue to operate normally" in a future period of time). Then, even if the operating state of the current hydrogen compressor should more appropriately fall into the cluster where cluster result 2 is located, the clustering algorithm of the prior art will make it fall into cluster result 1.

[0156] Single parameters cannot effectively realize the monitoring and early warning of equipment. The hydrogen compressor state monitoring method of the present invention does not directly process a single type of data collected by the sensor, but takes the sliding window as the unit for collecting data, and performs clustering processing on the sliding window information between two adjacent monitoring moments. For any vibration parameter, the vibration data in a sliding window has time series, and there is also time series between adjacent sliding windows (new sliding windows can only be born as time goes by); there is also partial overlap of vibration data between adjacent sliding windows, which enables the clustering algorithm to well mine and learn the potential relationships and correlations between adjacent sliding windows when processing the time-series sliding window information, improving the accuracy of the cluster results obtained by each clustering algorithm in the present invention.

[0157] In the hydrogen compressor condition monitoring method of the present invention, a sliding window only contains one type of vibration parameter. The sliding windows of different vibration parameters within the same time period overlap in time due to the overlapping data acquisition times of the sensors, resulting in the sliding windows also overlapping in time. The clustering algorithm in the present invention is not based on the isolated vibration data at a certain moment; instead, it is based on the sliding window information of this sequential, temporally related and overlapping vibration data, and there is also temporal overlap between the sliding windows of different vibration parameters. The clustering algorithm will also explore and learn the potential relationships and associations between the sliding window information of these different vibration parameters with temporal overlap and the exhaust pressure. This enables the clustering algorithms in the present invention to accurately predict the operating conditions of the current hydrogen compressor in the next period of time, thereby improving the accuracy of the cluster results output by each clustering algorithm; this also greatly improves the accuracy rate of the new clustering clusters and the corresponding cluster results obtained by the clustering algorithm with the ability to autonomously generate cluster results.

[0158] As can be seen from the above analysis, the window length and sliding step size of the sliding window are directly related to the sample data input into each clustering algorithm and indirectly related to the accuracy of the cluster results output by each clustering algorithm. In each monitoring cycle of the present invention, the window length and sliding step size are automatically adjusted based on the previous sliding window, the previous several sliding windows, and the intra-cluster entropy obtained by each clustering algorithm based on the previous several sliding windows, without manual intervention. Technicians only need to set the window parameters (i.e., initial values) of the first sliding window in each monitoring cycle, which greatly reduces the labor cost. If the window length is too long or the sliding step size is too short, it will lead to a significant increase in the computational cost during the process of the clustering algorithm processing the sliding window information and occupy a large amount of storage resources; while if the window length is too short or the sliding step size is too long, although the computational cost and the occupation of storage resources are reduced, the overlapping vibration data between adjacent sliding windows is also greatly reduced or even zero, and the distribution of sliding windows with temporal overlap between the sliding windows of different vibration parameters will also change, which is very unfavorable for the clustering algorithm to explore and learn the potential relationships and associations between the sliding windows, and even affects the accuracy of the cluster results output by each clustering algorithm. Therefore, the adjustment of the window length and sliding step size of the sliding window in the entire monitoring process of the present invention not only has a high degree of automation, but also balances the computational cost and the amount of storage resources occupied during the process of processing the sliding window information, while enabling the clustering algorithm to better explore and learn the potential relationships and associations between the sliding windows, allowing each clustering algorithm to better process the sliding window information and effectively perceive the long-term slowly deteriorating monitoring object.

[0159] The sliding window information of the present invention includes the vibration feature vector of the sliding window, including the time-frequency distribution result of the vibration data within the sliding window, the distribution of signal energy at different spatial scales, and the sensitivity coefficient to the damage fault on the surface of the machine body, further improving the sensitivity of each clustering algorithm to the condition of the hydrogen compressor.

[0160] The monitoring method of the present invention automatically updates the monitoring period, which is reflected in that: when monitoring the current hydrogen compressor for the first time, or during the first operation after replacing one or more key components of the current hydrogen compressor, for any vibration parameter of the current hydrogen compressor, a new monitoring period has started. Even for hydrogen compressors of the same model that have passed the pre - factory test, there are certain pre - factory test data; however, the period of the present invention's monitoring method is the process after a hydrogen compressor is transported to the site, installed, and officially put into operation. That is, when monitoring a hydrogen compressor for the first time. Therefore, before the first monitoring of a certain hydrogen compressor, the hydrogen compressor may still have some internal parts damaged or installation maladjustment due to transportation and on - site installation; the same is true for the first operation after replacing one or more key components of the current hydrogen compressor; both of these situations mean that the hydrogen compressor as a whole needs to go through a new break - in period, stable period, and deterioration period again. And within each monitoring period, the window length and sliding step are based on the previous sliding window. So, it is inevitable that the window length and sliding step of the new sliding window in these two situations cannot be obtained based on the previous sliding window. Therefore, we need to let the hydrogen compressors in these two situations enter a new monitoring period.

[0161] Based on the above analysis, it can also be seen that the update of the window length and the sliding step of the sliding window is crucial for the subsequent clustering algorithm in terms of the frequency of sample input and the change in the amount of sample data. The monitoring method of the present invention automatically updates the monitoring period and automatically adjusts the window length and the sliding step of the sliding window within each monitoring period, so that the frequency of sample input and the amount of sample data adapt to the subsequent clustering algorithm processing process, further improving the efficiency of the entire monitoring process and the accuracy of the cluster results output by the clustering algorithm; therefore, the monitoring method of the present invention can better adapt to each hydrogen compressor it monitors. For ease of understanding, for example: after entering a new monitoring period, in order to better understand the current situation of the current hydrogen compressor within the current monitoring period, the monitoring method of the present invention needs to frequently perform sample input and increase the amount of sample data input; after verification by technical personnel, within a period of time (i.e., the running-in period) after the monitoring period is updated, the window length of the sliding window indeed increases slowly, and the sliding step of the sliding window is short, to meet the need to frequently perform sample input and increase the amount of sample data input within the running-in period of the current monitoring period; while in the middle period of the new monitoring period (i.e., the stable period), the window length and the sliding step of the sliding window indeed increase significantly, and there is no need to frequently perform sample input or input a large amount of sample data within the stable period of the current monitoring period (reflected in the reduction of the overlapping vibration data between adjacent sliding windows and the increase of the sliding step of the sliding window); subsequently, as the clustering algorithm determines the operating condition of the hydrogen compressor, the frequency of sample input and the amount of sample data input will increase again due to the hydrogen compressor entering the deterioration period of the current monitoring period, which will not be elaborated here.

[0162] The hydrogen compressor condition monitoring method of the present invention does not use a single clustering algorithm to monitor the operating state of the hydrogen compressor, but uses multiple clustering algorithms to jointly monitor, and these clustering algorithms not only include clustering algorithms with the ability to independently generate cluster results, but also include clustering algorithms without the ability to independently generate cluster results; some of the cluster results obtained by these clustering algorithms may be the same, or they may all be different, but the present invention will ultimately use the cluster result with the highest score as the operating state of the hydrogen compressor at the current monitoring moment, and the cluster result with the highest score is also the cluster result of most clustering algorithms. After random inspection by technical personnel, it is basically the operating state of the hydrogen compressor at the current monitoring moment.

[0163] The score of the cluster result is obtained by accumulating the weight coefficients of the clustering algorithms corresponding to the same cluster result, and the weight coefficients in the monitoring method of the present invention also change with the monitoring period and the monitoring time within each monitoring period: the weight coefficients of each clustering algorithm at the first monitoring time in each monitoring period are determined by solving the multi-objective optimization function; and the weight coefficients of each clustering algorithm at the remaining monitoring times in each monitoring period need to be obtained based on the weight coefficients of each clustering algorithm at the previous monitoring time. This ensures that the weight coefficients of each clustering algorithm will not mutate as they change with the monitoring time; and it has been verified by technical personnel that only when a clustering algorithm obtains several cluster results in succession that are all the operating status of the hydrogen compressor at the corresponding monitoring time, the weight coefficient of the clustering algorithm will gradually increase.

[0164] Considering the special case with extremely small probability: if the clustering algorithm with the ability to autonomously generate cluster results obtains a new cluster result corresponding to the current monitoring moment, and the weight coefficient of the clustering algorithm that generates this new cluster result is small, then this new cluster result cannot be used as the operating state of the hydrogen compressor at the current monitoring moment. Therefore, the present invention records the cluster results obtained by all clustering algorithms with the ability to autonomously generate cluster results as a set to be verified, that is, in S5, after the technicians regularly verify and analyze the set to be verified, if the cluster results obtained by a certain clustering algorithm in the set to be verified are considered correct by the technicians for y consecutive times, and the cluster results are different from the cluster results set by the technicians, then the technicians increase the weight coefficient of the corresponding clustering algorithm, which means that although the cluster results output by the clustering algorithm are not the cluster results pre-set by the technicians, they are highly accurate. On the one hand, this can prevent the clustering algorithm with the ability to autonomously generate cluster results from indiscriminately generating multiple new cluster results to interfere with the technicians' evaluation of the operating status of the hydrogen compressor, greatly reducing the workload of the technicians. On the other hand, a small amount of manual intervention can be used to ensure that when the hydrogen compressor really has a situation other than the cluster results preset by the technicians, the clustering algorithm with a high weight coefficient and the ability to autonomously generate cluster results can accurately determine the operating status of the hydrogen compressor at the current monitoring moment.

[0165] In the hydrogen compressor condition monitoring method of the present invention, the operating state of the hydrogen compressor at the current monitoring moment can be accurately determined. The operating state of the hydrogen compressor not only includes the operating condition of the hydrogen compressor at the current monitoring moment, but also includes the prediction of the operating condition of the current hydrogen compressor in a period of time in the future. Therefore, the number of cluster results preset by technicians can be larger. For example, the determination of the operating condition of the hydrogen compressor at the current monitoring moment in multiple cluster results is the same, but the prediction of the operating condition of the current hydrogen compressor in a period of time in the future is different. This is directly different from the prior art in which the operating condition of the hydrogen compressor at the current monitoring moment is determined based on the exhaust pressure collected by the sensor. Moreover, the monitoring method of the present invention will not be prone to misjudgment as in the prior art due to technicians presetting too many cluster results.

[0166] As Figures 2 to 4 shown, for the clustering results obtained by technicians using the GMM clustering algorithm, DBSCAN clustering algorithm, and OPTICS clustering algorithm respectively during one revolution (i.e., 360 degrees) of the crankshaft, based on the inner ring vibration of the suction valve and the average exhaust pressure collected by the sensor; the abscissa is the serial number of the data segment, and the ordinate represents the vibration, that is, the vibration acceleration. In Figure 2 , the green data points represent condition 0, the purple data points represent condition 1, and the yellow data points represent condition 2. Among them, condition 0 means the hydrogen compressor is operating normally, condition 1 means the hydrogen compressor is operating at low pressure and no component is in a fatigue state, and condition 2 means the hydrogen compressor is operating at low pressure and the inner ring of the north exhaust valve is in a fatigue state. After verification by technicians: there are a small number of data points that should belong to condition 1 but are misidentified as green or yellow; the yellow data points are scattered among the other two conditions, and the probability of misidentification is very high. In Figure 3 , the green data points represent condition 0, and the yellow data points represent condition 1. Among them, condition 1 means the hydrogen compressor is operating at low pressure and the inner ring of the north exhaust valve is in a fatigue state, and condition 0 means the hydrogen compressor is operating normally. After verification by technicians: most of the data points in the interval with a large deviation from the peak value are marked as yellow, and a small number are marked as green; at the same time, a small number of data points at the end cycle that should belong to condition 0 are mislabeled as yellow in large quantities; due to the lack of rich condition types, the deviated data points cannot be correctly labeled. In Figure 4 , the green data points represent condition 0, and the yellow data points represent condition 1. Among them, condition 1 means the hydrogen compressor is operating at low pressure and the inner ring of the north exhaust valve is in a fatigue state, and condition 0 means the hydrogen compressor is operating normally. After verification by technicians: compared with Figure 3 , the number of a small number of data points at the end cycle that should belong to condition 0 and are mislabeled as yellow is greatly reduced; however, there is still a situation where due to the lack of rich condition types, the deviated data points cannot be correctly labeled.

[0167] As Figure 5As shown, it is the clustering result obtained by technicians using the monitoring method of the present invention, based on multiple vibration parameters (such as the vibration parameters of the inner ring of the south exhaust valve, the vibration parameters of the inner ring of the north exhaust valve, the vibration parameters of the inner ring of the suction valve, and the vibration parameters of the inner ring of the exhaust valve, etc.) during one rotation period of the crankshaft, and based on the vibration of the inner ring of the suction valve and the average exhaust pressure collected by the sensor; the abscissa is the serial number of the data segment, and the ordinate represents the vibration, that is, the vibration acceleration. The green data points represent the operating state 0, the purple data points represent the operating state 1, and the yellow data points represent the operating state 2. Among them, the operating state 0 means the hydrogen compressor is operating normally, the operating state 1 means the hydrogen compressor is operating at low pressure and no component is in a fatigue state, and the operating state 2 means the hydrogen compressor is operating at low pressure and the inner ring of the north exhaust valve is in a fatigue state; verified by technicians: the recognition accuracy of data points of various colors is relatively high, and the false alarm rate is relatively low.

[0168] For the 3 cluster results preset by technicians, using the hydrogen compressor state monitoring method of the present invention, compared with the GMM clustering algorithm that does not have the ability to autonomously generate cluster results in the prior art, the same hydrogen compressor is continuously monitored for 133 hours, and the interval between adjacent monitoring times is 5 minutes. After manual verification, the accuracy rate of the monitoring method of the present invention is as high as 91.36%, while the highest accuracy rate of the prior art is only 75.16%.

[0169] The technologies, shapes, and structures not described in detail in the present invention are all well-known technologies. It should also be pointed out that the above are only the preferred embodiments of the present invention, and are not used to limit the present invention. The components or steps in the embodiments of the present invention can be decomposed and / or recombined, and these decompositions and / or recombinations should be regarded as equivalent solutions of this application and should all fall within the protection scope of the present invention.

Claims

1. A method for monitoring the state of a hydrogen compressor based on vibration parameters, characterized in that, It includes the following steps: S1. For any vibration parameter of the hydrogen compressor: Obtain the vibration data of the current sliding window according to the window parameters of the current sliding window; the sliding windows are arranged in chronological order; S2. Calculate the statistical data and vibration feature vectors of the current sliding window based on the vibration data of the current sliding window; the window parameters, vibration data, statistical data, and vibration feature vectors of the current sliding window together form the current sliding window information; S3. Based on the sliding window information of each vibration parameter between the current monitoring moment and the previous monitoring moment and the exhaust pressure at the current monitoring moment, each clustering algorithm outputs the cluster result corresponding to the current monitoring moment; meanwhile, calculate the weight coefficients of each clustering algorithm at the current monitoring moment; S4. Accumulate the weight coefficients of the clustering algorithms corresponding to the same cluster result as the score of the corresponding cluster result, and then use the cluster result with the highest score as the operating state of the hydrogen compressor at the current monitoring moment; After S2, it also includes S2': S2'. Calculate the window parameters of the next sliding window based on the statistical data and window parameters of the current sliding window; The following content is also included in S2': Record the current sliding window The window length of is, where t is a positive integer. The previous sliding window of the current sliding window is denoted as , then the window length of the next sliding window is : ， ， ， Among them, represents the window length adjustment amount of the current sliding window ; represents the window length adjustment amount of the next sliding window ; f represents the rotational frequency of the component where the vibration parameter corresponds to the right boundary moment of the current sliding window ; ⌉ represents rounding up; is the first adjustment coefficient; is the second adjustment coefficient; represents the weight coefficient of the i-th clustering algorithm. A total of n clustering algorithms are adopted, that is, 1 ≤ i ≤ n; represents the sliding window 's within-cluster entropy in the i-th clustering algorithm; represents starting from the sliding window , including the sliding window inside, the historical maximum entropy value of the previous m sliding windows. m is a positive integer, and m ≤ t; represents the sliding window 's mean vector, 's modulus length is the difference between the sliding window and the average value of the vibration data of the sliding window , 's direction points from the average value of the vibration data of the sliding window to the average value of the vibration data of the sliding window ; represents 's 2-norm; is the maximum value function; represents the sliding window 's within-cluster entropy in the i-th clustering algorithm, represents the sliding window 's within-cluster entropy in the i-th clustering algorithm; represents starting from the sliding window , including the sliding window inside, the within-cluster entropy obtained in the corresponding clustering algorithm during the process of i ranging from 1 to n for the previous m sliding windows; represents covariance; represents variance; represents the time constant of the hydrogen compressor system; represents the sampling period of the vibration data in the current sliding window , which is determined by the acquisition frequency of the corresponding sensor; represents the maximum allowable drift rate of the vibration parameter of the current sliding window ; represents the rated value of the vibration parameter of the current sliding window .

2. The method for monitoring the state of a hydrogen compressor based on vibration parameters according to claim 1, wherein: When monitoring the current hydrogen compressor for the first time or during the first operation after replacing more than 1 key component of the current hydrogen compressor, for any vibration parameter of the current hydrogen compressor, a new monitoring cycle has entered, and the window parameters of the first sliding window in the new monitoring cycle are initial values; The window parameters include the window length and the sliding step; there are several vibration data arranged in time of only one vibration parameter in each sliding window.

3. A method for monitoring the state of a hydrogen compressor based on vibration parameters according to claim 1, characterized in that, In S2, calculating the vibration feature vectors of the current sliding window based on the vibration data of the current sliding window further includes the following sub-steps: S21. Perform local mean decomposition on the vibration parameter signal corresponding to the vibration parameter in the current sliding window: ， Among them, \(T\) represents time; \(X(T)\) represents the vibration parameter signal in the current sliding window. By fitting the \(N\) vibration data included in the current sliding window into a vibration parameter signal curve, the vibration parameter signal represented by this vibration parameter signal curve is \(X(T)\); represents the \(k\)-th PF component, \(k = 1,\cdots,K\), and \(K\) is a positive integer greater than 1; represents the \(K\)-th residue, is a monotonic function; S22. Extract the characteristic parameters of each PF component to form the vibration feature vectors of the current sliding window: ， ， ， ， Among them, represents the skewness coefficient of the k-th PF component ; represents the kurtosis coefficient of the k-th PF component ; represents the energy of the k-th PF component ; represents the energy ratio of the k-th PF component in the vibration parameter signal X(T); represents the h-th vibration data arranged in chronological order in the current sliding window; represents the average value of the N PF component values in that are at the same time points as the N vibration data in the current sliding window; represents the standard deviation of the N PF component values in that are at the same time points as the N vibration data in the current sliding window; a represents the time point of the left boundary of the current sliding window; b represents the time point of the right boundary of the current sliding window; E represents the energy of the vibration parameter signal X(T). S23. Based on the characteristic parameters of the K PF components, obtain the K vibration feature vectors of the current sliding window: , where represents the k-th vibration feature vector of the current sliding window.

4. A method for monitoring the state of a hydrogen compressor based on vibration parameters according to claim 1, characterized in that Next sliding window Sliding step size is as follows: ， ， ； Among them, represents the overlapping ratio of the next sliding window ; represents the overlapping ratio of the current sliding window ; represents the attenuation coefficient of the next sliding window ; represents the target overlapping ratio; represents the total number of sliding windows that need to be transitioned currently; and are all known quantities determined by the operating state of the hydrogen compressor at the previous monitoring moment; represents the remaining number of sliding windows that need to be transitioned, including the next sliding window ; , represents the remaining number of sliding windows that need to be transitioned, including the current sliding window .

5. A method for monitoring the state of a hydrogen compressor based on vibration parameters according to claim 1, characterized in that, The following content is also included in S3: Adopt n different clustering algorithms. Among these n clustering algorithms, some clustering algorithms do not have the ability to generate cluster results independently, and some clustering algorithms have the ability to generate cluster results independently; input the sliding window information of each vibration parameter between the current monitoring moment and the previous monitoring moment and the exhaust pressure at the current monitoring moment into each clustering algorithm, and each clustering algorithm respectively outputs the cluster result corresponding to the hydrogen compressor at the current monitoring moment; Calculate the weight coefficients of each clustering algorithm at the current monitoring moment simultaneously , 1 ≤ i ≤ n and i is a positive integer: If the current monitoring moment is the first monitoring moment within the current monitoring period, then obtain the weight coefficients of each clustering algorithm at the current monitoring moment based on solving the multi-objective optimization function; if the current monitoring moment is not the first monitoring moment within the current monitoring period, then obtain the weight coefficients of each clustering algorithm at the current monitoring moment based on the weight coefficients of each clustering algorithm at the previous monitoring moment.

6. The method for monitoring the state of a hydrogen compressor based on vibration parameters according to claim 5, wherein, Based on solving the multi-objective optimization function to obtain the weight coefficients of each clustering algorithm at the current monitoring moment, the following content is also included: Construct the multi-objective optimization function F and then solve it to obtain the values of the weight coefficients of each clustering algorithm at the current monitoring moment: ， ， The constraints are as follows: , Among them, is the second constant parameter; represents the weight coefficient corresponding to the i-th clustering algorithm; represents the sum of squared errors of the i-th clustering algorithm; represents the first weight matrix, which is a column vector composed of weight coefficients; represents the second weight matrix, which is the first weight matrix transposed, and is a row vector; represents the first weight matrix and the correlation matrix quadratic form; is the regularization parameter.

7. A method for monitoring the state of a hydrogen compressor based on vibration parameters according to claim 5, characterized in that, Based on the weight coefficients of each clustering algorithm at the previous monitoring moment, to obtain the weight coefficients of each clustering algorithm at the current monitoring moment, the following content is also included: Denote the current monitoring time as , the previous monitoring time as , where z≥2 and z is a positive integer. Then, the weight coefficient of the i-th clustering algorithm at the current monitoring time is : ， Among them, represents the weight coefficient of the i-th clustering algorithm at the previous monitoring moment ; represents the weight coefficient of the j-th clustering algorithm at the previous monitoring moment , where 1 ≤ j ≤ n and j is a positive integer; represents the learning rate; represents the global distribution at the previous monitoring moment; represents the distribution of the i-th clustering algorithm at the previous monitoring moment; represents the distribution of the j-th clustering algorithm at the previous monitoring moment; ) represents the JS divergence; represents the distribution and the global distribution of the JS divergence; represents the distribution and the global distribution of the JS divergence.

8. A method for monitoring the state of a hydrogen compressor based on vibration parameters according to claim 1, characterized in that, After S4, it further includes S5: S5. If the cluster result obtained by the clustering algorithm with the ability to autonomously generate cluster results does not become the cluster result with the highest score, then the cluster result obtained by the clustering algorithm with the ability to autonomously generate cluster results is recorded as a set to be verified. After technicians regularly verify and analyze the set to be verified, if the cluster result obtained by a certain clustering algorithm in the set to be verified is considered correct by the technicians for y consecutive times and is different from the cluster result set by the technicians, then the technicians increase the weight coefficient of the corresponding clustering algorithm.

Images