Method for predicting remaining service life of sealing device of piston compressor

Abstract

A computer-implemented method for predicting the remaining service life of a sealing device (10) of a piston compressor (11), the method comprising the steps of: a) providing vibration data (1a) or a characteristic (1b) that can be calculated therefrom of a reciprocating part (12) of the piston compressor (11); b) generating an input matrix (3) on the basis of the vibration data (1a) or the features (1b) that can be calculated therefrom; c) performing a simulation comprising a plurality of simulation steps by applying on the input matrix (3) a model (4) capable of describing error propagation in the closed system, where each simulation step generates an output matrix (5), and where the output matrix (5) of the nth simulation step is used as the input matrix (3) of the (n + 1) th simulation step; f) selecting a precise parameter (6c) from the values of the output matrix (5, 5); g) determining the number of simulation steps that must be carried out until the selected parameter (6c) falls below a predefined threshold value (8a) close to zero, and outputting this number in units of time as the result (9) of the carried out simulation; h) repeating the method steps a) to g) at fixed discrete time intervals to obtain a plurality of simulation results (9, 9), and i) curve-fitting the obtained simulation results (9, 9) and calculating the point of the time (7) at which the curve intersects with a predefined lower boundary (8c), the calculated intersection point being considered as an indication of the occurrence of the fault and of the predicted end of the service life of the sealing device (10).

Description

Technical Field

[0001] This invention relates to a method for determining the degree of degradation of a reciprocating compressor's sealing device and predicting its remaining service life, a preventive maintenance system for a reciprocating compressor, a reciprocating compressor including the preventive maintenance system, a computer-readable storage medium, and the use of the method for monitoring the degradation of a reciprocating compressor's sealing device and / or for predicting the remaining service life of a reciprocating compressor's sealing device and / or for adjusting leaks occurring in the sealing device of a reciprocating compressor. Background Technology

[0002] Condition monitoring is the process of monitoring the condition parameters (vibration, temperature, etc.) of a machine to identify significant changes that indicate developing faults. It is a crucial component of predictive maintenance and is also known as "preventive maintenance" or "anticipatory maintenance." Preventive maintenance allows for the planning of maintenance work or other measures to avoid unexpected machine downtime or indirect damage and its consequences. This can improve the effectiveness of production systems while simultaneously reducing system downtime and maintenance costs.

[0003] For the compression of fluids such as gases or aerosols, compressors are typically used. In a reciprocating compressor, the rotational motion of a crankshaft generated by a drive unit is converted by a connecting rod into the reciprocating motion of one or more pistons, which are used to compress the gas or aerosol drawn into the compressor unit. By design, the piston is subjected to forces that arise during this process, accelerating it from zero speed at the end of each stroke to a high speed in the middle of the stroke, and then back to zero speed. For at least a portion of the compression cycle, the compression chamber is sealed by closed cylinder valves before the compressed gas or aerosol is released to the discharge device by opening at least one cylinder valve. These valves are subject to a degree of wear due to their movement. A piston rod sealing system, also known as a seal, is used on the moving piston rod to seal the working chamber on the crank side. The sealing element is subject to continuous wear due to contact with the reciprocating piston rod.

[0004] Undesirable escape of the gas or aerosol to be compressed from the compressor unit, particularly leaks related to wear of the aforementioned sealing elements and / or cylinder valves, are collectively referred to as leaks. To measure leaks and thus assess the functionality of the compressor unit, corresponding sensors can be used, ideally located on each individual compressor unit or each stage of the compressor. However, for cost reasons, these sensors are typically not provided on the compressor, and / or their measured leak values ​​are not always automatically recorded and transmitted.

[0005] A method for monitoring the operation of a reciprocating compressor using pressure and vibration sensors is known from EP2458214A1. Based on the detected pressure within the equipment and the detected vibration of the reciprocating motion of the compressor, the described protection system calculates the stiffness value of the cylinder assembly and is therefore able to assess its structural integrity consistently during operation. The protection system is also capable of shutting down the compressor after determining that its condition deviates from a predefined standard state.

[0006] EP3436877A1 relates to a method and apparatus for vibration diagnostic monitoring and evaluation of individual machine parts. Frequency analysis is performed using a frequency analyzer, and at least one time signal of machine vibration is recorded and evaluated. The maximum frequency value extracted from the spectrum is then assigned to a specific machine part and compared with target value data, model data, and / or reference data for the corresponding machine part. If the extracted maximum frequency value (determined from the currently measured vibration data) deviates from the target value data, model data, and / or reference data of the machine part, evaluation information or warning information is output.

[0007] However, a drawback of the aforementioned methods and systems is that users are only notified when one of the calculated values ​​deviates from the ideal value. These methods and systems cannot transmit remaining service life or indicate the wear level of compressor sealing elements in real time. Therefore, methods and systems capable of accurately predicting the future condition and failure time of machine components such as sealing elements remain needed. Furthermore, preventative maintenance often involves replacing long-serving components. Such premature replacement leads to unnecessary costs and is not sustainable.

[0008] According to existing technology, there is also the article "Abrupt fault remaining useful life estimation using measurement from a reciprocating compressor valve failure" by Loukopoulos et al. (Mechanical Systems and Signal Processing, 2019, Vol. 121, pp. 359-372). This article compares the predictive performance of different methods for assessing the remaining useful life (RUL) of reciprocating compressors from the perspective of accuracy and precision. The predictive method discussed is a data-driven method, in which the degradation process is modeled using historical information. The data used for calculation comes from industrial reciprocating compressors and is retrieved from a server. This data is therefore not from real-time raw data from sensors attached to or measuring the compressor.

[0009] However, the drawback of this approach is that it reaches its limits due to new erroneous events (i.e., situations where it has not been trained), and the accuracy of predictions often depends on the quantity and quality of available data. Summary of the Invention

[0010] Based on the above-mentioned prior art, the present invention aims to eliminate these problems and other disadvantages in the prior art, and in particular provides a method by which the wear degree of the sealing device of a piston compressor can be determined and its remaining service life can be predicted cost-effectively and reliably.

[0011] According to the independent claim, the problem is solved by: a method, a preventive maintenance system for a reciprocating compressor, a reciprocating compressor, a computer-readable storage medium, and the use of the method for monitoring the wear of the seals of a reciprocating compressor and / or for predicting the remaining service life of the seals of a reciprocating compressor and / or for readjusting leaks occurring in the seals of a reciprocating compressor. Advantageous embodiments and further developments are the subject matter protected by the dependent claims.

[0012] This problem is addressed in particular by a computer-implemented method for determining the degree of wear and predicting the remaining service life of the seals in a reciprocating compressor. The method includes the following steps a) to h):

[0013] a) Provide vibration data for the reciprocating components of the piston compressor. Alternatively, it is conceivable to use features that can be calculated from the vibration data in the method according to the invention. The vibration data or features that can be calculated from it must be traceable to a specific position (segment M) of the movement of the reciprocating components of the piston compressor.

[0014] b) Create an input matrix based on vibration data or features that can be calculated from it. The input matrix has a dimension (M×N) determined by the number of segments M and the number of samples N.

[0015] c) A simulation consisting of multiple simulation steps is performed by applying a model capable of describing error propagation in a closed system to the input matrix. Each simulation step generates an output matrix. The output matrix of the nth simulation step is used as the input matrix for the (n+1)th simulation step.

[0016] d) Optionally, a Laplace matrix is ​​formed for each output matrix, and the eigenvalues ​​of the corresponding Laplace matrix are computed. The Laplace matrix can be obtained, in particular, by multiplying the corresponding output matrix by its transpose.

[0017] e) Optionally, calculate the normalized distance between the output matrix and the output matrix of the previous simulation step.

[0018] f) Select an exact parameter from the values ​​of the output matrix, or from the calculated eigenvalues ​​of the Laplace matrix formed in optional step d), or from the normalized distance calculated in optional step e). The selected parameter represents a measure of the synchronicity of the vibration data traceable to segment M or the characteristics that can be calculated from it.

[0019] g) Determine the number of simulation steps that must be performed until the selected parameter falls below a predefined near-zero threshold. Based on the sampling rate of the vibration data used in the input matrix or features that can be calculated from it, convert the performed simulation results (i.e., the determined number of simulation steps) into time units.

[0020] h) Repeat steps a) to g) of the above method at fixed discrete intervals to obtain a large number of simulation results.

[0021] i) After each new simulation, taking into account past simulation results, particularly by means of local linear regression or another suitable method, the curve is adapted to the obtained simulation results, and the future time point at which the curve intersects with a predefined lower boundary is determined. The lower boundary is specifically the x-axis (the horizontal axis) that passes through the zero point of the Cartesian coordinate system. The calculated time point at which the curve, adapted to the simulation results, intersects with the lower boundary is considered an indication of the occurrence of the failure, particularly a complete failure, and the predicted end point of the sealing device's service life.

[0022] In the context of this invention, the term "sampling time" is understood to refer to the time it takes to measure or calculate a value (particularly vibration data values ​​or characteristic values ​​that can be calculated from them).

[0023] In the context of this invention, the term "sealing device" refers in particular to a piston rod seal that seals the piston rod and / or a cylinder valve that seals the compression chamber.

[0024] The size of the predefined threshold depends on the input data, i.e., vibration data or features that can be calculated from it, and is close to zero because the second minimum eigenvalue being close to zero indicates that the system has high instability.

[0025] This invention is based on the recognition that leakage affects the vibration measured on a compressor. Specifically, in addition to increased leakage, damage to seals or guide rings can also cause displacement or deviation from the ideal movement of the piston rod and piston, increasing friction and vibration in the corresponding segments of the piston movement. Since the piston's acceleration force is greatest in the middle of the piston movement, the displacement in this region is also greater and more pronounced through increased vibration. As leakage increases, the displacement increases further, accompanied by increased vibration that spreads through multiple adjacent segments of the piston movement to the end of the stroke.

[0026] This is reflected in the simulation phase of the invention, where the propagation speed of the fault in the direction of the stroke end is simulated. Mapping the measured vibrations to the motion segments of the reciprocating components of the piston compressor, particularly specific positions of the piston, is a necessary preprocessing step to ensure that the input matrix is ​​a two-dimensional representation of the vibrations measured over the entire cycle of the reciprocating device. Not wishing to be limited thereto, it is assumed that the vibration modes in the motion of the reciprocating device, particularly the vibration modes in the piston motion, represent the thermodynamic processes within the cylinder. Information from experts performing manual machine diagnostics supports this assumption, who have reported a correlation between potential leaks or increased friction in the segments and a uniform and concurrent increase in vibration peaks in these segments.

[0027] In the context of this invention, vibrations occurring in the corresponding segments of piston motion and their propagation to neighboring segments are considered as interacting particles. Due to the transient nature of the development of vibrations, the dynamical relationships between the particles can be described, for example, using the time-dependent Ginzburg-Landau model (TDGL), a modification of the dynamic Ising model, where discrete variables represent the magnetic spin (-1, +1) states of the particles. The free energy of such a system can be assessed by observing the changes in the order parameters (G. Gaspari, Physical Review B, 1986, 33(5), 3295-3305).

[0028] The method according to the invention is essentially a hybrid digital sensor that uses vibration data to simulate the function or measurement of a leak sensor. In other words, the method according to the invention makes it possible to simulate leaks that cannot be directly measured using another parameter, namely vibration data or features that can be calculated from it, and to predict damage to seals or valves. The chosen method does not require additional pressure or temperature sensors, or calculations of dynamic pressure, and is therefore cost-effective in terms of the necessary sensor technology. Furthermore, the method according to the invention has proven particularly robust in terms of the accuracy of predicting the end of the service life of sealing devices, for example, in response to changes in compressor usage or other fluctuations not caused by wear.

[0029] In a preferred embodiment of the method according to the invention, the second minimum eigenvalue of the output matrix is ​​selected as a parameter. In this case, the predefined threshold is approximately zero, for example, a value of 0.000025.

[0030] In a preferred embodiment of the method according to the invention, step d) is performed and the second smallest eigenvalue of the Laplacian matrix is ​​selected as a parameter. In this case, the predefined threshold is approximately zero, for example, a quantity of 0.000025.

[0031] The second smallest eigenvalue of the Laplacian matrix is ​​called algebraic connectivity or Fiedler eigenvector (M. Fiedler, Czechoslovak Mathematical Journal, 1973, 23(2), 298-305). This metric plays a crucial role in the synchronization of coupled oscillators and the robustness and cascading failures of networks (S. Strogatz, Physica D 2000, 143, 1-20). If the algebraic connectivity of a graph is zero, the graph will be split into more than one connected component, meaning that dynamic processes will never be synchronized (Hernández et al., Journal of Complex Networks 2014, 2(3), 272-287).

[0032] On the one hand, the method according to the invention can be performed remotely, whereby vibration data measured on the compressor is sent to a separate location and evaluated there.

[0033] Alternatively, in a further embodiment of the method according to the invention, the provided vibration data or features that can be calculated therefrom may also be acquired as part of the performed method. For this purpose, at least one vibration time signal x(t) of the reciprocating motion component of the piston compressor is detected by at least one vibration sensor. Optionally, features can be calculated based on the recorded time signal x(t). The detected vibration time signal x(t) or the optionally calculated features are then divided into multiple time segments, which always correspond to the same position of the reciprocating motion device, particularly the piston or piston rod. For each time segment, the average value of the time signal x(t) recorded in the corresponding time segment or the average value of the optionally calculated features is calculated to obtain an average vibration value or an average feature value. Furthermore, the position of the reciprocating motion component of the piston compressor is detected by at least one position sensor, and the detected position data is provided for performing the method. Based on the recorded position data, the motion of the reciprocating motion component is subdivided into multiple segments M, which are preferably of equal size. Finally, the average vibration value or average characteristic value is assigned to the motion segment M such that the corresponding average vibration value or the corresponding average characteristic value always corresponds to the same motion segment M of the reciprocating motion component, and thus reflects the leakage at a certain point in the vibration data accordingly. For example, with a time interval of one minute, a certain number of measurement points are obtained in a certain motion segment M of the reciprocating motion component within one minute, and these measurement points are averaged and provided as the minute average value of the corresponding segment M in the input matrix.

[0034] In a preferred embodiment of the method according to the invention, the time when the selected parameter falls below a predefined threshold is predicted only after the parameter reaches a predetermined critical value. The critical value marks the beginning of an irreversible degradation process in the sealing device.

[0035] This significantly reduces the fluctuation range of simulation results and thus improves the accuracy of predicting the actual occurrence time of sealing device failures. Furthermore, the predicted start point shifts towards the direction of actual failures, thereby improving prediction accuracy and reducing the risk of false positive predictions.

[0036] In a preferred embodiment of the method according to the invention, the model applied to the input matrix is ​​a time-dependent Ginzburg-Landau model (TDGL).

[0037] In the context of this invention, the time-dependent Ginzburg-Landau model (TGDL) is found to be particularly suitable as a generalization of the n-vector and two-dimensional Ising model (Gaspari et al., Physical Review B 1986, 33(5), 3295-3305).

[0038] The use of reciprocating compressors typically occurs under conditions where there are external influences on the vibration of the reciprocating moving parts of the compressor. For example, when using reciprocating compressors on ocean-going vessels, the dominant wave patterns and associated ship motions themselves will cause vibrations that may overlap with the vibrations detected on the reciprocating compressor.

[0039] To reduce this impact, additional noise reduction processing can be performed on the provided vibration data or features.

[0040] In a preferred embodiment of the method according to the invention, the provided and optional denoised vibration data or features are normalized between -1 and +1, which is advantageous depending on the selected model and the distribution of the vibration data.

[0041] In a preferred embodiment, the method according to the invention further includes a data transmission step for transmitting vibration data or characteristics to at least one ground station. Alternatively or additionally, the method according to the invention may also include a data transmission step for transmitting the predicted remaining service life of the sealing device to the user interface of the reciprocating compressor. In both cases, at least the simulation step and the prediction of the time when the selected parameters are below a predefined threshold are performed wholly or partially on the ground using a ground station.

[0042] Vibration data measured on the reciprocating compressor is transmitted to a ground station, enabling remote monitoring of the compressor's condition without requiring the full computing power necessary to execute the method according to the invention on or near the compressor. Furthermore, transmitting the predicted remaining service life of the sealing device to the compressor's user interface, such as to the control panel or indicator lights, prevents users from being surprised by unforeseen compressor failures. This is particularly advantageous when monitoring reciprocating compressors used on ocean-going vessels, as unforeseen failures outside port pose significant safety risks.

[0043] In a preferred embodiment, the method according to the invention further includes transmitting the identified time and / or the predicted end point of the sealing device's service life to the user. This transmission is preferably real-time. Alternatively or additionally, according to the invention, it is also conceivable to assess the degree of wear of the sealing device based on the behavior of parameters in a certain number of simulation steps performed, and to transmit the resulting degree of wear to the user. This transmission is also preferably real-time.

[0044] This allows users of reciprocating compressors to know the condition of the compressor and, if necessary, take timely preventative measures for any necessary maintenance interruptions.

[0045] In a preferred embodiment of the method according to the invention, the sealing device is a piston rod seal that seals with the piston rod, or a cylinder for defining the compression chamber. Piston rod seals and cylinder valves are components of a reciprocating compressor; damage or even malfunction of these components severely affects the compression process and thus reduces the performance of the reciprocating compressor.

[0046] In a preferred embodiment of the method according to the invention, the execution of the method is initiated automatically, that is, at a time selectable by the user or at fixed time intervals.

[0047] The automation of this process significantly reduces the risk of unpredictable failures in reciprocating compressors. Furthermore, it allows for monitoring of the execution of maintenance tasks specified by the manufacturer and / or user.

[0048] In principle, all of the above embodiments can be combined with each other.

[0049] A preventative maintenance system for use with a reciprocating compressor further addresses this problem. The preventative maintenance system includes a computing device with at least one processor. The computing device is configured to receive position data of the reciprocating motion components of the reciprocating compressor, receive vibration data of the reciprocating motion components of the reciprocating compressor, or features that can be calculated from these vibration data or features, and assign these vibration data or features to motion segments M of the reciprocating motion components of the reciprocating compressor. The computing device is further configured to create an input matrix from the received vibration data or features and perform multiple simulations, each simulation comprising multiple simulation steps, using a model on the input matrix that can be used to describe error propagation in a closed system. Each simulation produces an output matrix, where the output matrix of the nth simulation step is used as the input matrix for the (n+1)th simulation step. The interval between two simulations corresponds to a defined time unit in each case. In particular, the interval between two simulations can be between one minute and six hours, preferably between one minute and four hours, and particularly preferably between one minute and sixty minutes. The smaller the time interval between the selected two simulations, the smoother the results, but the time required to obtain the results will increase. The computing device is optionally further configured to form a Laplace matrix of the corresponding output matrix and to calculate the eigenvalues ​​of the corresponding Laplace matrix or to calculate the normalized distance between the output matrix of the (n+1)th simulation step and the output matrix of the nth simulation step. The computing device is also configured to select a parameter from the values ​​of the output matrix, or from the calculated eigenvalues ​​of the optionally formed Laplace matrix, or from the optionally calculated normalized distance. The selected parameter represents a measure of the synchronicity of vibration data traceable to segment M or features that can be calculated therefrom. The computing device is configured to determine the number of simulation steps that must be performed until the selected parameter falls below a predefined near-zero threshold, as a result of simulation in each case, and output this number in time units. The time unit, or simulation time, is calculated from the time interval of the (average) values ​​of the input matrix, which is generated from the vibration data. The computing device is also configured to adjust the curve based on the obtained simulation results and to calculate the future time when the curve intersects a predefined lower boundary (specifically the x-axis) using an appropriate method (e.g., local linear regression), and thus obtain the failure time as a date. The calculated intersection point is considered an indicator of the predicted endpoint of failure occurrence and the service life of the sealing device. Finally, the calculation device is also configured to calculate based on the remaining service life of the sealing device of its output reciprocating compressor.

[0050] The advantages of the preventive maintenance system according to the present invention mainly derive from the advantages of the method according to the present invention already described.

[0051] In an advantageous embodiment, the preventive maintenance system according to the invention further includes at least one position sensor designed to detect the position of the reciprocating motion component of the piston compressor and communicatively coupled to a computing device. In this embodiment, the preventive maintenance system also includes at least one vibration sensor designed to detect the time signal x(t) of the vibration of the reciprocating motion component of the piston compressor and communicatively coupled to the computing device. The computing device is configured to divide the detected vibration time signal x(t) into multiple time segments and form an average vibration value for each time segment. Furthermore, the computing device is also configured to divide the motion of the reciprocating motion component into multiple segments M (preferably of equal size) based on a determined position and assign the average vibration value to these motion segments M.

[0052] This allows for closer monitoring of the condition of the piston compressor's sealing components and more accurate prediction of its remaining service life.

[0053] In a preferred embodiment of the preventive maintenance system according to the invention, the system includes at least one position sensor and at least one vibration sensor disposed on the cylinder head or crosshead of a piston compressor, the cylinder head or crosshead being associated with the reciprocating motion components of the piston compressor.

[0054] It has been found that mounting vibration sensors on the cylinder head allows for particularly reliable information regarding the condition and remaining service life of cylinder valves, especially the exhaust valve. Furthermore, it has been found that mounting vibration sensors on a crosshead provides particularly accurate information about the condition and remaining service life of the piston rod seals.

[0055] The reciprocating compressor, which includes the aforementioned preventative maintenance system, further addresses this issue.

[0056] In a preferred embodiment, the reciprocating compressor is designed as a reciprocating compressor with vertical piston movement. Vertical piston movement ensures low wear on the seals and guide elements. This type of reciprocating compressor is commonly used on ocean-going vessels.

[0057] Computer-readable storage media further solve this problem. A computer-readable storage medium contains a computer program including computer-readable program code. When executed by at least one processor of a computer, this program code is adapted to cause the processor to perform the methods described herein.

[0058] In the context of this invention, the term "storage medium" also includes, in particular, cloud storage, flash memory, and embedded memory.

[0059] This problem was ultimately solved by monitoring the degradation of the reciprocating compressor's seals using the methods described herein. Alternatively or additionally, the problem can be solved by using the methods described herein to predict the remaining service life of the reciprocating compressor's seals, particularly predicting remaining operating time or replacement date. As another option or supplement to the above uses, the problem can be solved by using the methods described herein to adjust for leaks occurring in the reciprocating compressor's seals. Attached Figure Description

[0060] The various embodiments of the present invention will now be explained in more detail with reference to the accompanying drawings, wherein the same or corresponding elements are generally provided with the same reference numerals. The figures show:

[0061] Figure 1 A flowchart illustrating the process sequence is shown according to the present invention;

[0062] Figure 2 Graphical representation of the output matrix 5,5' after n simulation steps;

[0063] Figure 3 The change of the real part Re of the second minimum eigenvalue 6d in n simulation steps;

[0064] Figure 4a Simulation results of all initial simulations (9, 9');

[0065] Figure 4b Figure 4a The local linear regression results in simulation results 9, 9' are shown.

[0066] Figure 5 The change of the critical index CI over time;

[0067] Figure 6 experience Figure 5 The actual remaining service life and the predicted remaining service life after the irreversible degradation process. Detailed Implementation

[0068] Figure 1A flowchart illustrating the sequence of a possible embodiment of a method according to the invention for determining wear and predicting the remaining service life of a reciprocating compressor's sealing device is shown. In step a), vibration data 1a, or features 1b calculated therefrom, by at least one vibration sensor belonging to the reciprocating compressor, is provided, and then in step b), an input matrix 3 is generated. This includes, for example, generating a data structure based on the measured vibration data 1a or the calculated features 1b, such that columns in the input matrix 3 represent vibration data 1a or the calculated features 1b at specific locations on the piston, and rows in the input matrix 3 represent vibration data 1a or features 1b varying over time. This produces an N×N matrix 3 or an N×M matrix 3. At different locations of the vibration sensor on the reciprocating compressor, conclusions can be drawn regarding the critical state and type of degradation of the sealing device. Following step b), a simulation comprising multiple simulation steps begins in step c). In step c.1), a state-field-based vibration dynamics simulation is performed, in this embodiment using the time-dependent Ginzburg-Landau model 4. If correctly applied, the results of a large number of existing state-field models 4 used to describe the propagation of faults in enclosed systems are equivalent. To solve the partial differential equations, a discrete version of the Laplace operator Δ is recursively applied to the output matrix 5 of the previous simulation step in each simulation step. Figure 1 In the example shown, the output matrix 5 of the first simulation step c.1) is therefore used as the input matrix 3ʹ of the second simulation step c.2), which in turn produces the output matrix 5ʹ. Figure 2 The output matrices 5 and 5' are shown as heatmaps after different numbers of simulation steps. The graph for time n=1 shows multiple white and black dots, representing the values ​​(states) of output matrix 5 after the first simulation step. After several simulation steps, the output matrix changes only slightly. Figure 1 In the embodiment of the method according to the invention shown, after each step c.1) to c. cn), a Laplace matrix 6a, 6aʹ, 6a'' is formed for the obtained output matrices 5, 5ʹ, 5''. This can be achieved, for example, by multiplying the corresponding output matrices 5, 5ʹ, 5'' with their transposes (not shown). The eigenvalues ​​6b, 6bʹ, 6b'' of the corresponding Laplace matrices 6a, 6aʹ, 6a'' in the simulation step are then calculated, which in Figure 1 The flowchart steps d.1), d.2), and dn) are explicitly shown. As shown in the flowchart steps f.1), f.2), and fn), for each simulation step, an exact parameter 6c is now selected from the calculated eigenvalues ​​6b, 6bʹ, and 6b''. In this example, the selected parameter 6c is the second smallest eigenvalue for each simulation step. Figure 3An example is shown showing the variation of the real part Re of the second minimum eigenvalue 6d over n simulation steps. In method step g), the simulation step where the second minimum eigenvalue first falls below a predefined near-zero threshold 8a is identified. This simulation step returns the simulation time as the simulation result 9, specifically in hours, minutes, or seconds, and can be a future time, particularly a calendar date. The simulation time is calculated from the distance between measurements, i.e., the number of sampling time points N in the input matrix 3, which is generated from the vibration data 1a or the feature 1b from which it can be calculated. For example, when using minute averages, the portion below the predefined threshold 8a after 750 simulation steps n can be converted into approximately half a day in time units by dividing 750 by 1440 minutes (24 hours × 60 minutes). Subsequently, further simulation is performed according to method step h), by repeating method steps b), generating input matrix 3, c), executing simulation step 4 and generating Laplace matrix 6a, d), calculating eigenvalues ​​6b, f), selecting and storing the second minimum eigenvalue 6c, g), identifying the number of simulation steps until this value falls below a threshold close to zero, and returning the value of that simulation time. Finally, in process step i), a suitable method is used to fit the curve with the obtained simulation results 9, 9', 9'', and the time 7 when the curve intersects the predefined lower boundary 8c is calculated. Figure 4a In this process, all initial simulation results (x-axis) correspond to the number of simulation steps until the curve falls below a predefined threshold 8a (y-axis). Figure 4b It shows the Figure 4a The examples shown are adjustments to the simulation results 9, 9', 9'' obtained using local linear regression with Loess smoothing. The failure time or remaining service life is calculated based on the intersection of the regression curve and a predefined lower boundary 8c, which in this example is the x-axis.

[0069] Figure 5 The curve shows the change of the critical index CI over time. This curve can be used to determine the point at which the sealing device begins the irreversible degradation process and to begin calculating its remaining service life. This point in time can be determined by the statistical distribution of simulation results, especially when a certain percentage of the simulation results are below the limit value. Figure 5 The limit value is 1, and it is marked with a white line after approximately 92 days (when the damage begins).

[0070] Figure 6 This shows the variation in the accuracy of the prediction error (in calendar days) over six weeks prior to the end of the actual service life of the sealing device (i.e., the occurrence or complete failure of the sealing device). From Figure 6As can be seen, the difference between the predicted end of service life and the actual occurrence of the failure event is very small, especially on the last day before the actual failure of the sealing device, which shows that the indication of the predicted end of service life is very accurate.

Claims

1. A computer-implemented method for determining the degree of wear of a sealing device (10) of a piston compressor (11) and predicting its remaining service life, the method comprising the following steps: a) providing vibration data (1a) of a reciprocating component (12) of the piston compressor (11) or features (1b) that can be calculated therefrom, which data or features can be traced back to a movement segment M of the reciprocating component (12) of the piston compressor (11); b) generating an input matrix (3) from the vibration data (1a) or features (1b) that can be calculated therefrom, wherein the input matrix (3) has a dimension (M×N) determined by the number of segments M and the number of sampling times N; c) performing a simulation comprising a plurality of simulation steps by applying a model (4) capable of describing error propagation in a closed system to the input matrix (3), wherein each simulation step generates an output matrix (5, 5ʹ), and wherein the output matrix (5) of the n-th simulation step is used as the input matrix (3ʹ) of the n+1-th simulation step; d) optionally, forming a Laplacian matrix (6a, 6aʹ) for each output matrix (5, 5ʹ), in particular by multiplying the corresponding output matrix (5, 5ʹ) by its transposed matrix (5a, 5aʹ), and calculating the eigenvalues ​​(6b, 6bʹ) of the corresponding Laplacian matrix (6a, 6aʹ); e) optionally, calculating the normalized distance between the output matrix (5ʹ) of the n+1th simulation step and the output matrix (5) of the nth simulation step; f) selecting exactly one parameter (6c) from the values ​​of the output matrix (5, 5ʹ), or from the calculated eigenvalues ​​(6b, 6bʹ) of the optionally formed Laplacian matrix (6a, 6aʹ), or from the optionally calculated normalized distance; g) determining as a result of the simulation performed (9) the number of simulation steps that must be performed until the selected parameter (6c) falls below a predefined near-zero threshold value (8a) and outputting this number in time units; h) repeating method steps a) to g) at fixed discrete time intervals to obtain a plurality of simulation results (9, 9'); as well as i) curve fitting the obtained simulation results (9, 9'), in particular by means of local linear regression, and calculating the time (7) at which the curve intersects a previously defined lower boundary (8c); wherein the selected parameter (6c) represents a measure of the synchronicity of the vibration data (1a) traceable to the segment M or of the characteristic (1b) which can be calculated therefrom, and wherein the calculated intersection point is regarded as an indication of the occurrence of a failure, in particular a complete failure, and of the predicted end of the service life of the sealing device (10).

2. The method according to claim 1, wherein the second smallest eigenvalue (6d) of the Laplacian matrix (6a) is selected as parameter (6c), and wherein the predefined threshold (8a) is approximately zero, in particular the threshold (8a) has a magnitude of 0.000025.

3. The method according to claim 1 or 2, wherein the vibration data (1a) or the features (1b) that can be calculated therefrom are obtained by the following sub-steps: - detecting at least one time signal x(t) of the vibration of a reciprocating component (12) of the piston compressor (11) by means of at least one vibration sensor (13); - optionally, computing features (1b) from the recorded time signal x(t); - dividing the detected vibration time signal x(t) or the optionally calculated characteristic (1b) into a plurality of time segments (1d) and forming an average value of the detected time signal x(t) or the optionally calculated characteristic (1b) in the corresponding time segments (1d) to obtain an average vibration value (2a) or an average characteristic value (2b); - detecting the position of a reciprocating component (12) of a piston compressor (11) by means of at least one position sensor (14) and providing detected position data (1c); - based on the detected position data (1c), dividing the movement of the reciprocating member (12) into a plurality of segments M, preferably of equal size; and - assigning the mean vibration value (2a) or the mean characteristic value (2b) to the movement segment M in such a way that the corresponding mean vibration value (2a) or the corresponding mean characteristic value (2b) always corresponds to the same movement segment M of the reciprocating component (12).

4. The method according to claim 1 , wherein the prediction of the time (7) for which the selected parameter (6c) is below a predefined threshold value (8a) is performed only after the parameter (6c) reaches a predetermined critical value (8b) which marks the beginning of an irreversible degradation process in the sealing device (10).

5. The method according to any of the preceding claims, wherein the model (4) applied to the input matrix (3) is a time-dependent Ginzburg-Landau model (TDGL).

6. The method according to claim 1 , further comprising a data transmission step for transmitting the vibration data (1a) or the characteristics (1b) to at least one ground station (15) and / or for transmitting the predicted remaining service life of the sealing device (10) to a user interface (11a) of the piston compressor (11), wherein at least the simulation step and the prediction of the time (7) for which the selected parameter (6c) is below a predefined threshold value (8a) are performed entirely or partially on the ground via the ground station (15).

7. The method according to one of the preceding claims, wherein the sealing device (10) is a piston rod seal (10a) sealing the piston rod or a cylinder valve (10b) defining a compression chamber.

8. The method according to one of the preceding claims, wherein the execution of the method is started automatically at a time selectable by a user or at fixed time intervals.

9. The method according to any one of the preceding claims, further comprising the following sub-steps: - transmitting the identified time (7) and / or the predicted end of the useful life of the sealing device (10) to a user, in particular in real time; and / or - evaluating the degree of wear of the sealing device (10) based on the behavior of the parameter (6c) during the number of simulation steps performed, and transmitting the degree of wear thus obtained to a user, in particular in real time.

10. A preventive maintenance system (20) for use with a piston compressor (11), the preventive maintenance system (20) having a computing device (21) including at least one processor (22), the computing device (21) being configured to: - receiving position data (1c) of a reciprocating component (12) of a piston compressor (11); - receiving vibration data (1a) of a reciprocating component (12) of a piston compressor (11) or characteristics (1b) that can be calculated therefrom; - assigning vibration data (1a) or characteristics (1b) to a movement segment M of a reciprocating component (12) of a piston compressor (11); - generating an input matrix (3) from the received vibration data (1a) or features (1b); - performing a plurality of simulations, each simulation comprising a plurality of simulation steps, by applying a model (4) capable of describing error propagation in a closed system on the input matrix (3), wherein each simulation step generates an output matrix (5), and wherein the output matrix (5) of the n-th simulation step is used as the input matrix (3ʹ) of the n+1-th simulation step; - Optionally, forming a Laplacian matrix (6a) of the corresponding output matrix (5) and calculating eigenvalues ​​(6b) of the corresponding Laplacian matrix (6a); - Optionally, calculating the normalized distance between the output matrix (5ʹ) of the n+1th simulation step and the output matrix (5) of the nth simulation step; - selecting a parameter (6c) from the values ​​of the output matrix (5, 5') or from the calculated eigenvalues ​​(6b, 6b') of the optionally formed Laplacian matrix (6a, 6a') or from the optionally calculated normalized distances, wherein the selected parameter (6c) represents a measure of the synchronicity of the vibration data (1a) that can be traced back to the segment M or of the features (1b) that can be calculated therefrom; - as a result of the simulation (9, 9') in each case, determining the number of simulation steps that must be carried out until the selected parameter (6c) falls below a predefined near-zero threshold (8a) and outputting this number in time units; - curve fitting is performed on the obtained simulation results (9, 9'), in particular by means of local linear regression, and the time (7) at which the curve intersects a previously defined lower boundary (8c) is calculated, the calculated intersection point being considered as an indication of the occurrence of a failure and the predicted end of the useful life of the sealing device; and - The remaining service life of the sealing device (10) of the output piston compressor (11).

11. The preventive maintenance system (20) according to claim 10, further comprising at least one position sensor (14) designed to detect the position of the reciprocating component (12) of the piston compressor (11) and communicatively coupled to the computing device (21); and at least one vibration sensor (13) designed to detect the time signal x(t) of the vibration of the reciprocating component (12) of the piston compressor (11) and communicatively coupled to the computing device (21), the computing device (21) being configured to divide the detected vibration time signal x(t) into a plurality of time segments (1d) and form an average vibration value (2a) of the corresponding time segment (1d), divide the movement of the reciprocating component (12) into a plurality of segments M of preferably equal size based on the determined position, and assign the average vibration value (2a) to these movement segments M so that the corresponding average vibration value (2a) always corresponds to the same movement segment M of the reciprocating component (12).

12. The preventive maintenance system (20) according to claim 11, wherein at least one vibration sensor (13) is arranged on a cylinder head (16) or a crosshead (17) of a piston compressor (11) associated with the reciprocating component (12).

13. A piston compressor (11) comprising a preventive maintenance system (20) according to any one of claims 10 to 12.

14. The piston compressor (11) according to claim 14, which is designed as a piston compressor (11) with vertical piston movement.

15. A computer-readable storage medium (30) containing a computer program (31), the computer program (31) comprising computer-readable program code (32) configured to, when executed by at least one processor (22) of a computer, cause the processor (22) to perform a method according to any one of claims 1 to 9.

16. Use of the method according to any one of claims 1 to 9 for monitoring the degradation of a seal (10) of a piston compressor (11) and / or for predicting the remaining service life of a seal (10) of a piston compressor (11), in particular predicting the remaining operating time or a replacement date and / or for correcting leakage occurring at a seal (10) of a piston compressor (11).

Images