Computer-implemented method and condition monitoring device for identifying mechanical damage of a machine

Abstract

A computer-implemented method for identifying mechanical damage of a machine based on performance indicators present at the rotational speed of the machine, comprising the steps of: - receiving, during a calibration period, a plurality of calibration data points, each comprising a rotational speed of the machine and a performance characteristic value assigned to the rotational speed, and determined from sensor data of the machine operating at one or more working points; - calibrating an AI-based Gaussian process regressor by inputting the calibration data points received during the calibration period, and outputting an estimated performance characteristic curve and a range of uncertainty assigned to each value thereof for all rotational speed values within a specified rotational speed range indicative of a normal state of the machine; - receiving a data point to be evaluated after the calibration period; and - comparing it to the estimated performance characteristic curve; and - outputting an anomaly message if it lies outside at least one of the specified limit values of the range of uncertainty depending on the performance characteristic value.

Description

Technical Field

[0001] The present invention relates to a computer-implemented method for identifying mechanical damage to a machine based on performance indicators present at the machine's rotational speed, a condition monitoring device for performing the method, and a corresponding computer program product. Background Technology

[0002] Condition monitoring systems are used to monitor the condition of machines and equipment and identify potential problems early. Based on historical sensor data such as temperature or vibration of the machine or equipment, data analysis methods are used to detect deviations from the "normal state" and take action when necessary. By identifying these deviations from the normal state early, maintenance cycles can be optimized and unplanned downtime can be avoided.

[0003] A major challenge here is learning this "normal state" of the machine. Furthermore, rapid debugging is required; that is, the state monitoring system must be able to detect deviations from the machine's normal state after the shortest possible calibration time. However, from a system perspective, this requires sufficient data to learn the "normal state" across the machine's various states.

[0004] Vibration is an important measurement parameter for identifying machine faults such as bearing damage, shaft misalignment, or imbalance when monitoring the condition of machines, especially their industrial applications. In practice, various performance indicators (KPIs) are often derived from high-resolution vibration data. These KPIs are observed over a long period, and trend and anomaly identification algorithms are applied to these performance characteristic curves over time.

[0005] However, electric motors, especially those operating at different speeds with the aid of converters, exhibit significantly different vibration behaviors across these speed ranges, particularly due to resonance effects. This makes it difficult to compare the resulting KPIs. Therefore, the motor's operating point must be considered when analyzing performance indicators derived from vibration data. This requires providing the condition monitoring system with sufficient historical data from different motor operating points during production operations, enabling the identification of potential anomalies within specific speed ranges. This is particularly difficult to guarantee in the initial period shortly after motor commissioning. Summary of the Invention

[0006] Therefore, the objective of this invention is to provide predictions of performance metrics in a reliable manner, after a short calibration period, based on a limited amount of historical data. Another objective is to determine predictions of performance metrics at machine operating points that have not been previously observed or have been barely observed.

[0007] This task is accomplished by the measures described in the independent claims. Advantageous extensions of the invention are presented in the dependent claims.

[0008] A first aspect of the present invention relates to a computer-implemented method for identifying mechanical damage to a machine based on performance indicators related to the machine's rotational speed, the method comprising the following steps:

[0009] - During the calibration period, multiple calibration data points are received, which include the machine's rotational speed and the performance characteristic value assigned to that speed, and these calibration data points are determined based on sensor data of the machine operating at one or more operating points;

[0010] - The AI-based Gaussian process regressor is calibrated by inputting calibration data points received during the calibration period, and the estimated performance characteristic curve and the uncertainty range of all speed values ​​within a specified speed range indicating the normal state of the machine are output, for each value of the performance characteristic curve.

[0011] - The data points to be evaluated are received during the evaluation period following the calibration period; and

[0012] - Compare this data point with the estimated performance characteristic curve; and

[0013] - If the data point to be evaluated is outside at least one specified limit of the uncertainty range that depends on the performance characteristic value, an exception message is output.

[0014] Advantageously, a Gaussian process regressor not only predicts the performance characteristic curves but also models the uncertainty of each performance characteristic value. This uncertainty range significantly reduces the calibration time for such a system because it eliminates the need to provide sufficient data points for all operating points. Each data point—that is, each calibration data point and each data point to be evaluated—includes the rotational speed and the performance characteristic value determined at that speed. This uncertainty range allows for the estimation of the confidence level of each data point to be evaluated.

[0015] In an advantageous implementation, the Gaussian process regressor includes a statistical model for regression and prediction of data points, the statistical model including a covariance function for modeling the relationship between input and output values, wherein the hyperparameters of the covariance function are set by calibration. The performance characteristic curve is the mean determined by the Gaussian process regressor, and the uncertainty range is the standard deviation determined by the Gaussian process regressor.

[0016] In an advantageous implementation, the distance between the data point to be evaluated and the average value is calculated, and the distance is compared with the standard deviation of the data point on the performance characteristic curve.

[0017] In other words, for each data point to be evaluated—that is, the performance characteristic value determined at that speed—the distance from the estimated performance characteristic value on the estimated performance characteristic curve is determined, and this distance is compared with the uncertainty range. This prevents erroneous abnormal messages for data points to be evaluated at speeds with large uncertainty ranges.

[0018] In an advantageous implementation, the performance characteristic curve includes performance characteristic values ​​for all speed values ​​within a specified speed range.

[0019] Therefore, this performance characteristic curve can be used for all the data points to be evaluated whose rotational speed is within that range.

[0020] In an advantageous implementation, the Gaussian process regressor is recalibrated at specified time intervals using additional data points.

[0021] Simply executing the performance characteristic curve and thus recalibrating the Gaussian process regressor improves accuracy. Therefore, this method is optimized for processor capabilities. In cases where the calibration process is outsourced to a remote server, the evaluation of the data points to be evaluated can be performed using a simple device located near the machine.

[0022] Advantageously, during the calibration period, an increased number of data points are received per unit time relative to the evaluation period.

[0023] Therefore, the calibration period can be shortened or the amount of calibration data used for the recalibration can be increased, and thus the calibration results can be improved, i.e., a more accurate performance characteristic curve can be output.

[0024] Preferably, data points received during previous operation periods of the machine are used as new calibration data points.

[0025] In one advantageous implementation, the rotational speed is derived from magnetic field data, and performance characteristic values ​​are derived from vibration data measured at the machine.

[0026] Therefore, a physically reliable relationship is given between the values ​​at the data points and the values ​​of the machine parameters measured directly at the machine.

[0027] In an advantageous implementation, if the number of abnormal messages for data points received within a specified time period exceeds a specified limit, a deterioration message is output.

[0028] This deterioration message can be used to proactively indicate a decline in the machine's condition. With multiple specified, especially elevated, limits, different critical levels can be implemented for the machine.

[0029] In an advantageous implementation, the estimated performance characteristic curve and the uncertainty range of each value assigned to the performance characteristic curve, along with the data points to be evaluated, are output on a graphical user interface.

[0030] By displaying the machine's status and behavior across the entire speed range and optionally over time, maintenance personnel receive advance and comprehensive instructions for impending hazardous situations or maintenance tasks.

[0031] In an advantageous embodiment, the mechanical damage is bearing damage, machine shaft misalignment, or machine imbalance.

[0032] The most common reasons for machine operation degradation can be identified through exception messages.

[0033] According to a second aspect, the present invention relates to a condition monitoring device for identifying mechanical damage to a machine based on performance indicators present at the machine's rotational speed, the condition monitoring device comprising at least one processor configured to perform the following steps:

[0034] - During the calibration period, multiple calibration data points are received, which include the machine's rotational speed and the performance characteristic value assigned to that speed, and these calibration data points are determined based on sensor data of the machine operating at one or more operating points;

[0035] - The AI-based Gaussian process regressor is calibrated by inputting calibration data points received during the calibration period, and the estimated performance characteristic curve and the uncertainty range of all speed values ​​within a specified speed range indicating the normal state of the machine are output, for each value of the performance characteristic curve.

[0036] - The data points to be evaluated are received during the evaluation period following the calibration period; and

[0037] - Compare this data point with the estimated performance characteristic curve; and

[0038] - If the data point to be evaluated is outside at least one specified limit of the uncertainty range that depends on the performance characteristic value, an exception message is output.

[0039] The described condition monitoring device is capable of robust anomaly detection using relatively few data points and can model the dependence of machine damage performance indicators (KPIs) on rotational speed with sufficient accuracy. Even if a rotating machine, especially a motor, is operated in a different operating state at a later point in time after the calibration phase, it can indicate the extent to which this is abnormal.

[0040] According to a third aspect, the present invention relates to a computer program product comprising a non-volatile computer-readable medium that can be directly loaded into the memory of at least one digital computer, the non-volatile computer-readable medium comprising program code portions that, when executed by the at least one digital computer, cause the at least one digital computer to perform the steps of the method.

[0041] Computer program products, such as computer program devices, may be provided or supplied as computer-readable storage media, in the form of memory cards, USB memory sticks, CD-ROMs, DVDs, or as downloadable files from servers on a network. Attached Figure Description

[0042] Embodiments of the method and apparatus according to the invention are illustrated by way of example in the accompanying drawings and will be described in more detail in the following description.

[0043] Figure 1 An exemplary illustration of data points recorded on a machine with two working points is shown in the diagram.

[0044] Figure 2 An embodiment of the method for identifying mechanical damage to a machine according to the present invention is shown in flowchart form;

[0045] Figure 3 An exemplary embodiment of an estimated performance characteristic curve having a range of uncertainty, determined by the method according to the present invention, is shown;

[0046] Figure 4 An exemplary process according to the method of the present invention during the calibration and evaluation periods is illustrated schematically; and

[0047] Figure 5 An embodiment of the condition monitoring device according to the present invention is illustrated in block diagram.

[0048] In the embodiments and accompanying drawings, the same or equivalent elements may be equipped with the same reference numerals. In principle, the elements shown and their dimensional relationships should not be considered to scale; rather, the elements may be shown to scale at a larger size for better presentation and / or better understanding.

[0049] Unless otherwise stated in the following description, the terms “receive,” “calibrate,” “compare,” “output,” etc., preferably refer to operations and / or processes and / or processing steps that alter and / or generate data and / or transform such data into other data, wherein such data may in particular be presented or exist as physical quantities, such as electrical pulses.

[0050] The condition monitoring device and its components, such as at least one data interface, a return unit, an evaluation unit, or an output interface, may include one or more processors. The processor may be, in particular, a central processing unit (CPU), a microprocessor, or a microcontroller, such as an application-specific integrated circuit (ASIC) or a digital signal processor, which may be combined with a memory unit for storing program instructions. Detailed Implementation

[0051] Previous condition monitoring systems either made the performance metrics (hereinafter also referred to as KPIs) derived from vibration or motor current data completely unrelated to the machine's current operating point, or, for example, divided the possible bandwidth of the rotational speed s into multiple fixed operating points BP1, BP2, etc. Figure 1 As shown in the figure. However, this approach has drawbacks. If the resulting KPIs, for example, which might measure the severity of motor damage, are not set to correlate with operating points, it is difficult to identify clear trends or anomalies in these KPIs, as they can fluctuate very drastically with operating points.

[0052] Divide into separate, strictly isolated working points BP1 and BP2, such as Figure 1 The diagram only partially addresses this problem. For example, if outlier detection, also known as anomaly detection, is performed based on the distributions V1 and V2 of previous data points at the corresponding operating points BP1 and BP2, and thresholds k1, k2 or k2, k3 are defined for data points outside a specific confidence interval, the following problems may arise:

[0053] Although the performance metric KPI exhibits an approximately linear dependence on the machine's rotational speed s and therefore shows no anomalous behavior, the resulting two distributions V1 and V2 differ significantly. For example, the data point DP1, which is immediately to the right of the boundary between the two operating points BP1 and BP2, is within the normal range for operating point BP1, but would be considered anomalous relative to operating point BP2 because this data point is far outside distribution V2, which is below the threshold k2.

[0054] Therefore, the operating points BP1 and BP2 cannot be arbitrarily divided, otherwise false alarms may occur in some cases. Since the specific vibration behavior within the speed range may vary greatly depending on the motor and cannot be fully understood in practice, it is impossible to reasonably select extreme values.

[0055] Therefore, when analyzing performance indicators derived from vibration data, the motor's operating point must be considered. However, this requires providing sufficient historical data from different motor operating points to a data-driven, machine learning-based condition monitoring system during production operations, enabling the identification of potential anomalies within specific speed ranges. This is particularly difficult to guarantee in the short period following motor commissioning.

[0056] The proposed computer-implemented method for identifying mechanical damage, such as that in rotating machines, solves this task. Figure 2 An exemplary flow of the method is illustrated as a flowchart.

[0057] The basic idea of ​​this method is to model performance characteristics using a Gaussian process regressor. These characteristics, based on vibration data, serve as indicators of motor mechanical damage and are strongly correlated with rotational speed. Specifically, the following steps are performed:

[0058] During the calibration period, multiple calibration data points kDP are received, as described in step S1. These calibration data points include the machine's rotational speed and a performance characteristic value assigned to that speed. Here, the rotational speed is determined based on sensor data from the machine operating at one or more operating points. The rotational speed is derived from magnetic field data, and the performance characteristic value is derived from vibration data measured at the machine.

[0059] In step S2, the AI-based Gaussian process regressor GPR is calibrated by inputting the calibration data points kDP received during the calibration period. The calibrated GPR provides the estimated performance characteristic curve gKPI and the uncertainty range assigned to each value of this performance characteristic curve for all speed values ​​within a specified speed range as output. The estimated performance characteristic curve gKPI indicates the normal state of the machine.

[0060] This completes the calibration of the Gaussian process regressor (GPR), and the estimated performance characteristic curve gKPI can be used to evaluate data points recorded during machine operation. Correspondingly, during the evaluation period following the calibration period, the data point bDP to be evaluated is received (see S3). This data point bDP is compared with the corresponding performance index of the estimated performance characteristic curve gKPI at the rotational speed of data point bDP (see S4). If the data point bDP to be evaluated is outside at least one specified limit value, an exception message AM is output, where the at least one limit value depends on the uncertainty range of the performance characteristic value (see S5).

[0061] Preferably, if the evaluated data point bDP is outside the uncertainty range, i.e., if the performance characteristic value of the data point bDP is greater than the upper limit or maximum performance characteristic value of the uncertainty range, or if the performance characteristic value of the data point is less than the lower limit or minimum performance characteristic value of the uncertainty range, then an exception message AM is output. This limit value can also be specified as being within and / or outside the uncertainty range.

[0062] If the data point bDP to be evaluated is within the specified limit, no exception message is output, see step S6. Optionally, in this case, a "normal state message" can be output, confirming that the machine is in a normal state within the specified limit. This can be represented, for example, by outputting the evaluated data point in a performance characteristic curve graph, such as... Figure 3 As shown in the image.

[0063] The Gaussian Process Regressor (GDP) comprises a statistical model for regression and prediction of data points. This GDP uses a covariance function to model the relationship between input and output values. During calibration, the hyperparameters of this covariance function are estimated based on the received calibration data points kDP to fit the regressor. Using this model, predictions for new input values ​​are calculated by generating probability distributions for possible output values. The GDP also provides an estimate of the uncertainty in its predictions.

[0064] In specific cases, the model primarily provides a continuous function, which, as a result, describes the dependence of these performance characteristic values ​​on rotational speeds within a specified range and thus represents the estimated performance characteristic curve. Furthermore, the model provides an explanation of the uncertainties within that rotational speed range.

[0065] Figure 3 An example of an estimated performance characteristic curve 11 over a specified speed range is shown, which characterizes normal operation of the machine, and is output by a Gaussian process regressor based on approximately ten received calibration data points 12.

[0066] Figure 10 illustrates the performance characteristic curve 11, which represents the dependency between the performance characteristic value and the rotational speed *s* modeled using a Gaussian process regressor. The shaded area indicates confidence intervals, such as a 90% confidence interval, which indicates the uncertainty range 13. This means that uncertainty range 13 has a 90% probability of containing the true value of the performance indicator. It can be seen that the dependency is not linear. While a linear increase is observed in the rotational speed range between 2600 and 2700 rpm, the performance indicator decreases again with increasing speed in the range around 2800 rpm. Due to the small divergence here, the uncertainty of this method is very low and therefore sensitive to outliers. However, in the ranges below 2400 rpm and above 2900 rpm, the uncertainty increases sharply because there are no calibration data points from which a "normal state" could be reliably derived. Nevertheless, a rough explanation can still be made because, based on the previous calibration data points 12, a continuous function representing the performance characteristic curve 11 was learned, which describes the performance characteristic curve as well as possible and takes into account the uncertainties of the method.

[0067] Since this uncertainty can be modeled using a Gaussian process, the calibration time for this method and the state monitoring device executing it can be significantly reduced because sufficient calibration data points are not absolutely required for all operating points of the machine. Furthermore, this method is robust to false alarm anomaly detection and corresponding anomaly messages.

[0068] Preferably, the Gaussian process regressor is recalibrated at specified time intervals using additional data points recorded on the machine at later time points. This improves the accuracy of the resulting, newly estimated performance characteristic curves and the range of uncertainty.

[0069] Figure 4 An embodiment of the method is shown, which includes details regarding the recalibration of the Gaussian process regressor and the resulting recalibrated estimated performance characteristic curves.

[0070] By using sensors on the machine to be monitored, brief, high-resolution snapshots of vibration data (VD) and magnetic field data (FD) are recorded at calibration intervals, along with optional additional data used to calculate the machine's rotational speed (see M1, M2). Within these high-resolution snapshots, a large amount of vibration data (VD) and magnetic field data (FD) is measured. These intervals can be cyclical or predefined according to any other scheme, or can be triggered based on one or more events.

[0071] Then, from these magnetic field data FD, the current rotational speed s of the machine at the time of recording is calculated (see M3), and this current rotational speed is stored (see M5). Using various signal processing methods, performance characteristic values ​​KPI are calculated from the vibration data VD (see M4), and similarly, these performance characteristic values ​​are stored, for example, in a database (see M5).

[0072] The two values ​​are coupled into a data point kDP, i.e., the rotational speed s and the associated performance characteristic value KPI, which is used as input for calibration of a Gaussian process regressor or regression model (see M6). This Gaussian process regressor or regression model calculates a Gaussian distribution for a given rotational speed s, which describes the machine's "normal behavior" within that rotational speed range by its mean and standard deviation. The calibrated Gaussian process regressor GPR now provides this performance characteristic curve for condition monitoring (see M7).

[0073] Here, an initial number of N data points is used to fundamentally calibrate the regression model. Then, data points recorded at later time points are used for retraining. Preferably, data points from a range of values ​​where only a small number of data points are available are used for retraining / recalibration. For example, during operation, data from a very high speed range is added, where only a small number of data points are available, and therefore, only poor predictions of the machine's behavior can be made so far for that speed range. The resulting performance characteristic curve includes performance characteristic values ​​for all speed values ​​within the specified speed range.

[0074] Then, during machine condition monitoring (this is called the evaluation period), referring to M8, the new data point bDP to be evaluated is compared with the average value within that range. To do this, the distance between the new data point bDP to be evaluated and the average value is calculated, and this distance is compared with the uncertainty range of that point on the performance characteristic curve. If this distance is greater than a specified limit, such as n times the standard deviation, the new data point to be evaluated is marked as abnormal. An abnormality message is output. Preferably, during this calibration period, an increased number of data points are received per unit time relative to the evaluation period.

[0075] The data points to be evaluated can also be used as calibration data points.

[0076] As mechanical damage progresses, an upward trend in performance eigenvalues ​​can be expected for the data points being evaluated. That is, new data points will drift towards higher performance eigenvalues. Since the model parameters of the Gaussian process regressor do not change for a certain period after recalibration, the distance between the new performance eigenvalues ​​and the learned mean will become increasingly larger, eventually exceeding the previously defined limit of n times the standard deviation. This trend can be detected by the increase in detected anomalies and corresponding anomalous messages, thereby detecting the deterioration of the mechanical fault. In one variant, in this case, the method triggers a warning and outputs a deterioration message.

[0077] Figure 5 A condition monitoring device 30 is shown for identifying mechanical damage to machine 20 based on performance indicators present at machine speeds. The condition monitoring device 30 includes: at least one processor configured to; a calibration data interface 31; a regressor unit 32; a measurement data interface 33; an evaluation unit 34; and an output interface 35.

[0078] Machine 20 is a rotating machine, such as a motor, especially an electric motor. Sensors arranged on machine 20 detect the field strength and vibration of the machine. From the sensor data, the machine's rotational speed and one or more performance indicators are derived, which are indications of mechanical damage. Mechanical damage that can be identified by performance indicators includes, for example, bearing damage, machine shaft misalignment, and / or machine imbalance.

[0079] The calibration data interface 31 is designed to receive multiple calibration data points during the calibration period. These calibration data points include the rotational speed of the machine 20 and the performance characteristic value assigned to that rotational speed, and these calibration data points are determined based on sensor data of the machine operating at one or more operating points. Optionally, the rotational speed and the performance indicators present at that rotational speed can be derived from the sensor data within the condition monitoring device 30.

[0080] The regressor unit 32 is designed to: calibrate an AI-based Gaussian process regressor by inputting calibration data points received during the calibration period; and output the estimated performance characteristic curve and the uncertainty range of all speed values ​​within a specified speed range indicating the normal state of the machine, assigned to each value of the performance characteristic curve.

[0081] The measurement data interface 33 is designed to evaluate the data points to be evaluated during the evaluation period following the calibration period.

[0082] Evaluation unit 34 is designed to compare the data points to be evaluated with the estimated performance characteristic curves.

[0083] Output interface 35 is designed to output an exception message if the data point to be evaluated is outside at least one specified limit of the uncertainty range depending on the performance characteristic value. Preferably, output interface 35 is designed as a graphical user interface. On this graphical user interface, the estimated performance characteristic curve, the uncertainty range assigned to each value of the performance characteristic curve, and the data point to be evaluated are output.

[0084] A computer program product includes a non-volatile computer-readable medium that can be directly loaded into the memory of at least one digital computer, and the non-volatile computer-readable medium includes program code portions that, when executed by the at least one digital computer, cause the at least one digital computer to perform the steps of the method described above and to function as a state monitoring device.

[0085] Compared to previous solutions, the described condition monitoring device 30 can perform robust anomaly detection using relatively few data points and model the dependence of machine damage KPIs on rotational speed with sufficient accuracy. Even if the motor is operated in a different operating state at a later time point after the calibration phase, it can indicate the extent to which this is an anomaly. As the number of data points increases, the behavior of machine 20 can be modeled in increasing detail over time, as the Gaussian process regressor is always retrained periodically. This method significantly reduces the calibration time of the condition monitoring line without significantly increasing the risk of false detections. This significantly improves the performance capabilities of the condition monitoring system and thereby significantly increases its acceptance.

[0086] The purpose of this specification is solely to provide illustrative examples and demonstrate further advantages and special features of the invention. Therefore, this specification should not be construed as limiting the scope of application of the invention or the patent rights claimed in the claims. In particular, the features disclosed in conjunction with the methods described herein can be reasonably used to extend the methods described herein, and vice versa.

Claims

1. A computer-implemented method for identifying mechanical damage to a machine (20) based on performance indicators present at a rotational speed of the machine (20), the method comprising the following steps: Sensors on the machine are used to monitor sensor data, including magnetic field data and vibration data. During the calibration period, multiple calibration data points (kDP) are received (S1), each including a value of the rotational speed of the machine (20) and a performance characteristic value assigned to the rotational speed, and the calibration data points are determined based on the sensor data of the machine (20) operating at one or more operating points. The AI-based Gaussian process regressor (GPR) is calibrated by inputting calibration data points (kDP, 12) received during the calibration period, and the estimated performance characteristic curve (gKPI, 11) and the uncertainty range (13) of each value of the performance characteristic curve (gKPI, 11) for all speed values ​​within a specified speed range indicating the normal state of the machine (20) are output. The data points (bDP) to be evaluated are received (S3) during the evaluation period following the calibration period; and Compare the data points to be evaluated (bDP) with the estimated performance characteristic curves (gKPI, 11) (S4); and If the data point to be evaluated (bDP) is outside at least one specified limit of the uncertainty range (13) depending on the performance characteristic value, an exception message (AM) is output (S5).

2. The method according to claim 1, wherein, The Gaussian process regressor (GPR) includes a statistical model for regression and prediction of data points. The statistical model is a covariance function used to model the relationship between input and output values, wherein the hyperparameters of the covariance function are set by the calibration.

3. The method according to claim 1, wherein, The performance characteristic curves (gKPI, 11) are the average values ​​determined by the Gaussian process regressor (GPR), and the uncertainty range is the standard deviation determined by the Gaussian process regressor.

4. The method according to claim 3, wherein, Calculate the distance between the data point (bDP) to be evaluated and the mean, and compare the distance with the standard deviation of the data point (bDP) on the estimated performance characteristic curve (gKPI, 11).

5. The method according to claim 1, wherein, The performance characteristic curves (gKPI, 11) include the performance characteristic values ​​for all speed values ​​within the specified speed range.

6. The method according to claim 1, wherein, The Gaussian process regressor (GPR) is recalibrated at specified time intervals using additional calibration data points (kDP).

7. The method according to claim 1, wherein, During the calibration period, an increased number of calibration data points (kDPs) are received per unit time relative to the evaluation period.

8. The method according to claim 1, wherein, The rotational speed is derived from the magnetic field data, and the performance characteristic value is derived from the vibration data measured at the machine (20).

9. The method according to claim 1, wherein, If the number of aberration messages (AM) for data points (bDPs) received within a specified time period exceeds a specified limit, a deterioration message is output.

10. The method according to claim 1, wherein, On the graphical user interface, the estimated performance characteristic curve (gKPI, 11) and the uncertainty range (13) of each value assigned to the performance characteristic curve (gKPI, 11) are output, along with the data point to be evaluated (bDP).

11. The method according to claim 1, wherein, The mechanical damage is bearing damage, shaft misalignment of the machine (20), or imbalance of the machine (20).

12. A condition monitoring device (30) for identifying mechanical damage to a machine (20) based on performance indicators present at a rotational speed of the machine (20), the condition monitoring device comprising at least one processor configured to perform the following steps: Sensors on the machine are used to monitor sensor data, including magnetic field data and vibration data. During the calibration period, multiple calibration data points (kDP) are received (S1), each including a value of the rotational speed of the machine (20) and a performance characteristic value assigned to the rotational speed, and the calibration data points are determined based on the sensor data of the machine (20) operating at one or more operating points. The AI-based Gaussian process regressor (GPR) is calibrated by inputting calibration data points (kDP, 12) received during the calibration period, and the estimated performance characteristic curve (gKPI, 11) and the uncertainty range (13) of each value of the performance characteristic curve (gKPI, 11) for all speed values ​​within a specified speed range indicating the normal state of the machine (20) are output. The data points (bDP) to be evaluated are received (S3) during the evaluation period following the calibration period; and Compare the data points to be evaluated (bDP) with the estimated performance characteristic curves (gKPI, 11) (S4); and If the data point to be evaluated (bDP) is outside at least one specified limit of the uncertainty range (13) depending on the performance characteristic value, an exception message (AM) is output (S5).

13. A computer program product comprising a non-volatile computer-readable medium that can be directly loaded into the memory of at least one digital computer, the non-volatile computer-readable medium including a program code portion that, when executed by the at least one digital computer, causes the at least one digital computer to perform the method according to claim 1.

Images