MONITORING METHOD FOR THE OPERATION OF A NUMBER OF ONE OR MORE DRIVE DEVICES, IN PARTICULAR FOR THE OPERATION OF AN INTERNAL COMBUSTION ENGINE, AND SYSTEM DESIGNED FOR THE MONITORING METHOD

Description

[0001] The invention relates to a monitoring method according to the preamble of claim 1 for the operation of a number of one or more drive devices, in particular for the operation of an internal combustion engine, wherein a drive device has a control device and an operating parameter of the drive device is monitored on the basis of operating data for the operating parameter acquired by means of the control device, the method comprising the steps: Monitoring the operating parameter of the drive device with a monitor algorithm for the operating data, in particular a monitor algorithm of an observer module of the control device, applying a fault detection algorithm to the acquired operating data of a fault detection to detect a fault relating to the operating parameter of the drive device, sending a fault indication from the fault detection for the drive device based on a result of the fault detection algorithm; receiving the fault indication and checking the result, in particular receiving and checking in a server device of a control instance, activating a control connection to the drive device with regard to the operating parameter and feeding back a monitoring value for the operating parameter to a feedback connection to the monitor algorithm.

[0002] The invention also relates to a system designed for the monitoring method.

[0003] Control and regulation methods for an individual drive system can now be highly complex and feature sensor support, and are now supposed to be able to adapt comprehensively to the individual state of a drive system. For example, US 2019 / 353131 A1 describes a detection method for operating an internal combustion engine that includes a knock control system.

[0004] WO 2019 / 106534 A1 describes a method for operating a powertrain for a vehicle. The method comprises receiving signals from a plurality of sensors linked to the vehicle. Optimal thresholds for classifications of sensor data can be changed with respect to input signals for the powertrain and the reception of feedback signals. Priorities for the sensor signals can be changed based on signal attribution and by confirming current signal priorities. The adjustment of the operation management can be implemented based on look-up tables, allowing the input signals to the powertrain and the measurement of the use of feedback signals. The method can thus be dynamically adapted and modify powertrain data, which primarily occurs in look-up tables during vehicle operation under a second set of conditions.

[0005] It is also known to use a control-engineered observer for these and other control systems. Observers, in the sense of control engineering, reconstruct non-measurable quantities (states) from known input variables (e.g., manipulated variables or measurable disturbance variables) and output variables (measured variables) of an observed reference system. For this purpose, the observed reference system is modeled in the observer, and the measurable state variables, which are therefore comparable to the reference system, are adjusted using a controller. This is intended to prevent a model from generating an error that grows over time, especially in reference systems with integrating behavior. A more accurate term would be a reference-controlled synthesizer. An observer can be designed precisely when the reference system is observable via the available measured variables.However, observability is generally not a necessary condition for observer design. Instead, it is sufficient if the system is detectable. Observers are used, for example, in state controllers to reconstruct unmeasurable state variables, in discrete-time control systems where the measured variable cannot be updated in each cycle, and in metrology as a replacement for measurements that are technically or economically impractical. A consistent theory for linear system models and constant proportional error feedback was developed starting in 1964 by the American control engineer David Luenberger. The method can, in principle, be extended to nonlinear models.

[0006] In a known procedure for controlling and regulating an internal combustion engine, it is also known in particular situations that a false-positive detection in the sense of the detection of an artifact - i.e., a so-called "artificial result" - can occur; this is generally understood to be in contrast to a "true-positive" detection in the sense of the detection of a "real event".

[0007] US 2018 / 0158314 A1 discloses a method for trend analysis and automatic tuning of alarm parameters for a machine, the method comprising obtaining condition-related measurements of the machine and, based thereon, conditionally triggering an alarm and correlating a relationship between triggered alarms and the detected defects, counting the number of true positives, false negatives and false positives from the current measurement data, comparing the number of counted true positives, false negatives and false positives with the specified acceptable limits.

[0008] WO 2018 / 7802 A1 discloses a method for operating an internal combustion engine having a fuel injection system, which in turn includes a high-pressure accumulator for fuel. The high pressure of the injection system is monitored over time as part of a so-called "monitoring" process. For a high-pressure-dependent start time for an injection, a check is made to determine whether continuous injection monitoring should be implemented by checking whether a high-pressure oscillation has occurred within an oscillation time interval before the injection start. In this case, this also includes error detection, i.e., a false-positive detection of a continuous injection is avoided. This avoids the unnecessary activation of an alarm signal and, possibly, even the shutdown of the internal combustion engine without a valid reason.

[0009] In addition, there is now an expanded option for monitoring the operation of a drive unit, especially an internal combustion engine. Beyond such potentially complex control strategies, it has also proven useful, albeit with a completely different approach, to "merely" implement monitoring of drive systems, particularly to diagnose, for example, the health status of an internal combustion engine. Approaches within the framework of so-called "BIG DATA" monitoring have proven to be advanced in this regard. However, this requires considerable effort to handle large amounts of data, and it must also be available for its evaluation. Furthermore, this requires a high level of hardware and computing power for analytics.In order to train such systems, it is necessary that large amounts of data are also available and computationally efficient algorithms must be able to validate the results in the training process.

[0010] For example, US 2018 / 148037 A1 discloses a method for diagnosing an internal fault of a pressure sensor downstream of a fuel pump in a vehicle's fuel system. Measures for operating the fuel system depend on whether the pressure sensor output flattens for at least a threshold time, which may be indicative of a fault. The procedure is improved by a dynamic learning method in which the pressure of a pressure relief valve and the pressure of the pressure sensor are monitored.

[0011] However, these and other known fault detection approaches based on plausible assumptions using threshold analyses still have room for improvement. Furthermore, most of these methods are relatively specifically tailored to individual drive systems, taking into account the characteristics of a drive system or a fleet of such drive systems and their components.

[0012] US 2019 / 0310618 A1 describes an approach for a system and software for integrating a model-based and data-driven diagnostic system, in which differences in a dynamic system are automatically identified and these differences or deviations are used for fault detection and for identifying or isolating them (so-called "FDI") for an automatic fault identification system. By combining a model-based approach and a data-driven approach for generating residual variables or disturbances, it is possible to detect, isolate, and identify faults. In contrast to standard operation, for example, fault modes can be clustered and classified. Such an approach is applicable, for example, to a fleet of network-connected drive systems such as excavators, commercial vehicles, or conveyor systems.Such a monitoring method for the operation of a number of one or more drive devices can also be improved.

[0013] In particular, it is desirable that not only error detection in such a process is based on a large amount of data, but also that the reliability of the data used is verified, or that the process can be further developed using such reliable data. It should be considered that the monitoring and control systems themselves are already highly complex. It should also be considered that common practices such as redundant sensor implementation or the further expansion of modeling or look-up tables lead to a further increase in hardware or software costs.

[0014] This is where the invention comes in. Its object is to improve a known monitoring method and system, particularly with a view to improving one or more of the aforementioned aspects. In particular, a monitoring method and system of the type mentioned above is to be improved such that the reliability of the data used in the monitoring method and system is improved and / or the monitoring method and system itself remains capable of further development or optimization during operation.

[0015] In particular, it is an object of the invention to further develop a data-based monitoring method and system of the type mentioned above such that, on the one hand, it is capable of being operated with a reasonable amount of data while, on the other hand, still offering a comparatively broad range of applications. Excessive data and analysis effort, as well as an overly customized orientation of the monitoring method, should preferably be avoided.

[0016] The problem is solved by a monitoring method according to claim 1.

[0017] The invention is based on a monitoring method for the operation of a number of one or more drive devices, in particular for the operation of an internal combustion engine, wherein a drive device has a control device and an operating parameter of the drive device is monitored based on operating data for the operating parameter acquired by the control device. The monitoring method comprises the steps: Monitoring the operating parameter of the drive device with a monitor algorithm for the operating data, in particular a monitor algorithm of an observer module of the control device, applying a fault detection algorithm to the acquired operating data of a fault detection to detect a fault relating to the operating parameter of the drive device, sending a fault indication from the fault detection for the drive device based on a result of the fault detection algorithm; receiving the fault indication and checking the result, in particular receiving and checking in a server device of a control instance, activating a control connection to the drive device with regard to the operating parameter and feeding back a monitoring value for the operating parameter to a feedback connection to the monitor algorithm.

[0018] According to the invention, it is further provided that the result is qualified as false positive (artificial) or true positive (real) after testing.

[0019] According to the invention it is further provided that the disturbance detection has a discrimination unit which sends the result, in particular only the result qualified as true-positive (genuine), to the control connection to an actuator and sends the result, in particular only the result qualified as false-positive (artificial), to the feedback connection to the observer module.

[0020] Starting from a monitoring method of the type mentioned above, the monitoring method according to the invention initially provides, in a manner known per se, the steps mentioned above, according to which—represented here in abbreviated form—an operating parameter can be recorded and monitored for a drive device based on operating data acquired by a control device of a drive device, a fault detection algorithm is applied, and a fault indication can be sent as a result, wherein the result of the fault detection algorithm can be verified after receipt of the fault indication. Only then is a control connection to the drive device activated.With regard to the operating parameter, it is also known to feed back a monitoring value for the operating parameter to a feedback connection to the monitor algorithm per se, namely to enable the feeding back of an operating parameter to the monitor algorithm in the sense of a feedback loop.

[0021] The invention is based on the idea that such approaches for verifying the result, or activating a control connection and a feedback loop, can still be improved within the framework of a data-based monitoring process. To improve the verification, the invention provides for the result to be qualified as false positive (artificial) or true positive (genuine) after verification.

[0022] In this case, a false-positive result is to be understood as an artificially generated result. In this sense, a false-positive result is to be understood as a result that has arisen artificially or is contrived. A false-positive result is to be understood as a non-genuine result that should not be taken seriously, in particular ignored, for adjusting the drive device via the actuator. According to the concept of the invention, a false-positive result can nevertheless be meaningfully used. A false-positive result can also advantageously be used as an artificially generated result, namely for learning feedback. This allows, for example, a monitoring algorithm and / or a fault detection algorithm to be adapted to improve it and, in particular, to avoid false-positive results in the future.

[0023] In this case, however, a true-positive qualified result is to be understood as a genuinely generated result. In this sense, a true-positive qualified result is a result that has been reliably generated or has a reason, or is genuine or justified. Thus, the result qualified as true-positive, and thus "genuine," can be passed to the control connection of an actuator. In particular, a corrective measure can be taken on the drive device by means of the actuator depending on the true-positive qualified result.

[0024] The cases can be applied analogously to negatively qualified outcome types. The four cases are also named differently in different contexts. The English terms true positive, false positive (or false negative and true negative ) is common. In the context of signal detection theory, true-positive cases are also called true-positive or hit (False negative cases are also called miss and true-negatives are also called true-negative cases or correct rejection designated).

[0025] In the following, the results obtained are qualified as true-positive (genuine) or false-positive (artificial). In the following and in the claims, the term "genuine" is used for the term true-positive, and the term "artificial" is used for the term false-positive, in parentheses for clarity.

[0026] According to the invention, it is further provided that the fault detection has a discrimination unit that transmits the true-positive (i.e., qualified as genuine) result to the control connection to an actuator. In particular, it is provided that, after testing, only the results that are qualified as true-positive (i.e., qualified as genuine) are transmitted to the control connection to an actuator. This has the advantage that control and regulation, or the initiation of actions, takes place only on the basis of the true-positive results, rather than on the basis of results that are qualified as genuine. Erroneous action transmission or actuator control is thus avoided.

[0027] According to the invention, it is further provided that at least the false-positive result (i.e., the result qualified as artificial) is passed to the feedback loop. In particular, it is provided that the false-positive result, i.e., the result qualified as artificial, and only this one, is passed to the feedback loop. This has the advantage that the feedback loop is capable of improving, especially for the further development of the observer module, or preferably the algorithms running in the observer module of the control device, such as the fault detection algorithm and the monitoring algorithm, to the extent that false-positive results are reduced.

[0028] In short, following the concept of the invention, the data-based monitoring method is improved in that a control connection is activated, if possible only or increasingly on the basis of test results qualified as genuine, while a feedback is generally designed to improve the observer module in particular or only with regard to the results identified as false-positive, so that these are increasingly avoided within the framework of the further development or increasing optimization of the observer module in the data-based monitoring method for the operation of the number of one or more drive devices.

[0029] The monitoring method according to the invention is thus capable of operating with a reasonable amount of data, while still offering a comparatively broad range of applications. In particular, excessive data and analysis effort, as well as an overly customized approach to the monitoring method, are avoided.

[0030] Overall, the data-based monitoring method thus takes advantage of the well-known "BIG DATA" approaches for monitoring methods already explained in principle at the beginning. However, the discrimination of results provided for by the invention during the test allows for a "false positive" or "true positive" qualification. This makes it possible, on the one hand, to improve the control connections and, on the other hand, to improve the learning feedback; thus, ultimately, to get by with less data effort. This, in turn, advantageously reduces the large amounts of data required within the framework of the well-known "BIG DATA" concepts or limits them to a manageable level. Likewise, the effort required for statistical analysis of the data volumes is reduced or limited to a manageable level.Furthermore, according to the concept of the invention, this is achieved with the advantage of a --in short, titled -- "catch-all" approach, in which individual and customized algorithms can be increasingly dispensed with.

[0031] To date, identifying an actual condition requires either manual evaluation by specialist personnel or the creation of highly customized models, as explained above, that are trained for a specific fault symptom. For example, to monitor an internal combustion engine, an algorithm for detecting valve clearance can be developed, trained, and maintained. This requires significant development and maintenance effort. However, as soon as such conceptually specific algorithms and training methods are to be applied to a different internal combustion engine, these customized algorithms must be adapted and verified.

[0032] The concept of the present invention, in contrast, is based on a comparatively broadly applicable, compatible approach that provides false / true positive detection using the discrimination unit and can be implemented in any drive unit for an actuator and feedback connection, regardless of its individual design. The invention itself, in its concept, primarily utilizes the false positive qualified results within a feedback loop to improve the observer's identification and thus to train the monitoring function based on the feedback; that is, primarily the reduction of false positive results. This then has an increasingly positive effect in combination with the true positive results used for the actuator connection, in that the monitoring method can be trained comparatively nonspecifically and with little data expenditure.

[0033] Over time, this synergistic combination results in the learning function minimizing false positive results and the control link increasingly optimizing the use of only true positive results. The inventive concept is thus internally consistent and can be used for the data-based monitoring and / or control process regardless of the individual drive unit type.

[0034] The concept of the invention advantageously leads to a comparatively universal algorithm capable of detecting impending malfunctions of a drive unit with respect to an operating parameter at an early stage. This makes it possible to utilize service margins and capacities more effectively, avoiding unplanned downtimes in favor of planned service deployments. In principle, service intervals can also be extended and overall operating approaches optimized, for example, with respect to operating parameters such as fuel consumption and resource expenditure.

[0035] Advantageous further developments of the invention can be found in the dependent claims and specify in detail advantageous possibilities for realizing the concept explained above within the scope of the task and with regard to further advantages.

[0036] Advantageously, the monitoring value and the result, in particular only the result qualified as a false positive (artificial), are passed to the feedback connection to the observer module. For example, long-term trend detection can be implemented. In particular, a so-called "step change" detection with automatic message generation can be established additionally or alternatively. This advantageously reduces the computational effort while still providing very effective advantages in error detection.

[0037] In particular, the feedback connection from the actuator to the observer module can be designed as a learning feedback.

[0038] By means of such or a similar "feedback" loop, it is advantageously possible, particularly when optimized in a training process, to reduce the number of false-positive (artificial) results for the observer particularly efficiently.

[0039] In particular, it is planned: Adapting the monitoring algorithm and / or the fault detection algorithm with the false-positive (artificial) qualified result, and / or initiating a corrective measure on the drive device by means of the actuator depending on the true-positive (real) qualified result.

[0040] In particular, if the control instance denies the existence of a fault, the fault detection algorithm can be adapted in such a way that the specificity of the fault detection algorithm for detecting this fault is increased.

[0041] In particular, if the control authority confirms the existence of the fault, the control authority can issue a rectification indication signal to initiate rectification of the fault.

[0042] Preferably, the true-positive (genuine) qualified result implies or confirms the presence of the disorder, and / or the false-positive (artificial) qualified result implies or confirms the absence of the disorder. In other words, the test result advantageously either confirms the presence of the disorder or denies the presence of the disorder.

[0043] It is preferably provided that the monitoring of the operating parameters comprises monitoring the operating and measured variables using an observer system, in particular comprising a physical model of the drive device with the operating variables for reconstructing states of the drive device taking into account the measured variables.

[0044] In particular, the fault detection algorithm can comprise applying a set of criteria with one or more fault identification criteria to the at least one detected operating parameter. Preferably, it can be provided that a result of applying a fault identification criterion to a detected operating parameter indicates whether a fault in the drive device is present.

[0045] Preferably, the operating parameters are recorded as a discrete-time operating data sequence, with the disturbance being detected as a disturbance curve of an operating parameter for the discrete-time operating data sequence. In particular, the further development of the method within the framework of monitoring functions for discrete-time data can be used for further development. This can advantageously be used for further development within the framework of a long-term trend detection or step-change detection with automatic message generation.

[0046] Preferably, it is provided that a technical fault area of ​​the drive device is assigned to the fault and / or displayed, wherein the fault display comprises a fault history and a fault area.

[0047] In particular, each fault identification criterion is assigned fault origin information that contains indications of the origin of the fault.

[0048] It is advantageous that the fault indication signal indicates the fault origin information which is associated with the fault identification criterion on the basis of which the fault indication signal was sent.

[0049] Preferably, the method is used to operate a single drive device and the application of the fault detection algorithm includes a comparison of a current value of an operating parameter with at least part of the course of the operating parameter recorded over the operating period.

[0050] Advantageously, for example, general monitoring functions can be applied to data from an individual drive device. In particular, it has proven advantageous to compare current measured values ​​from an individual drive device with existing measured values ​​from the individual drive device, in particular from the individual drive device's own past. For example, within the scope of the general monitoring functions for data from an individual drive device, an existing measured value from a past period, for example, singular or averaged over the past 3 months, can be used in comparison to a current measured value relating to an individual drive device.

[0051] Preferably, it is additionally or alternatively provided that the method is used for operating a plurality of drive devices, at least one operating parameter is detected for each of the drive devices, and the application of the fault detection algorithm includes a comparison of a value of an operating parameter of a drive device with values ​​of the operating parameter of the remaining drive devices of the plurality of drive devices.

[0052] General monitoring functions can thus be applied particularly to data from entire fleets. For example, it has proven advantageous to compare measured values ​​from an entire fleet to identify so-called "outliers"; that is, data or data clusters related to an individual drive device that deviate significantly or persistently over a significant period of time when compared to measured values ​​from the entire fleet.

[0053] Furthermore, the object is achieved by the invention with a system of claim 11. The system is designed according to the invention for a monitoring method according to the concept of the invention or one of the developments for the operation of a number of one or more drive devices, wherein a drive device has a control device and an operating parameter of the drive device is monitored based on operating data for the operating parameter acquired by the control device. According to the invention, the system comprises: the number of one or more drive devices, in particular an internal combustion engine, wherein the control device is signal-connected to at least one drive device, and a server device of a control instance is signal-connected to the control device. Preferably, the server device of the control instance comprises: a receiving module for receiving the fault indication; a checking module for checking the result of the detected fault by the control instance after receiving the fault indication; a discrimination unit that transmits the true-positive (genuine) qualified result to the control connection to an actuator and the false-positive (artificial) qualified result to the feedback connection to the observer module.

[0054] It is preferably provided that the control device of the drive device comprises: a detection module configured to detect and monitor the operating parameters of the drive device using a monitor algorithm of an observer module of the control device, an execution module configured to apply a fault detection algorithm of a fault detection to detect a fault, a transmission module configured to transmit a fault indication from the fault detection for the drive device based on a result of the fault detection algorithm.

[0055] Preferably, one or more adaptation modules and / or remedy indication modules are provided and designed to Adapting the monitoring algorithm and / or the fault detection algorithm with the false-positive (artificial) qualified result, and / or initiating a corrective measure on the drive device by means of the actuator depending on the true-positive (real) qualified result.

[0056] In particular, the fault detection algorithm can be adapted depending on the result of the check, wherein a rectification indication module which is designed to output a rectification indication signal from the control instance to initiate rectification of the fault if the presence of the fault has been confirmed by the control instance.

[0057] Embodiments of the invention are now described below with reference to the drawings in comparison to the prior art, some of which is also shown. These are not necessarily intended to represent the embodiments to scale; rather, where useful for explanation, the drawings are schematic and / or slightly distorted. With regard to additions to the teachings immediately apparent from the drawings, reference is made to the relevant prior art. The general idea of ​​the invention is not limited to the exact form or detail of the preferred embodiment shown and described below, or to an object that would be limited compared to the object claimed in the claims. For specified dimensioning ranges, values ​​within the stated limits are also intended to be disclosed as limit values ​​and can be used and claimed as desired.Further advantages, features and details of the invention will become apparent from the following description of the preferred embodiments and from the drawing, which shows: . Fig. 1 is a diagram of a basic concept of a data-based monitoring method for the operation of a number of drive devices with a control device and an observer, comprising a fault display and feedback to the observer; Fig. 2 is a schematic representation of the concept according to the invention, expanded within the scope of a particularly preferred embodiment, wherein, according to the concept of the invention, a discrimination unit is provided between the observer on the one hand and the control connection and feedback on the other; Fig. 3 is a schematic representation of the system for implementing the data-based monitoring method with the drive device, the control device, an observer, and a control instance, wherein in this embodiment the discrimination unit is provided in the control instance;Fig. 4A shows a first example of a data-based monitoring method according to a first modified preferred embodiment for the operation of a drive device in the form of an internal combustion engine and having a control device, an observer, and a monitoring instance; Fig. 4B shows a second example of a data-based monitoring method according to a second modified embodiment for the operation of a plurality of drive devices, each of which has a control device, wherein, according to the second modification, a common observer and monitoring instance are provided.

[0058] Fig. 1 shows a structure of a data-based monitoring method 1000 for the operation of a single or a number of drive devices 1, one of which is shown here as an internal combustion engine by way of example, and which is Fig. 2 is explained in more detail within the framework of a particularly preferred embodiment. Each of the internal combustion engines has a control unit (ECU) and a data logger 10 provided therein, which is capable of recording operating data 200 of operating parameters BP of the drive device 1; in the present case, the operating data 200 are recorded as a time-discrete operating data sequence transiently over the course of operation of the drive device 1 as a function of time.

[0059] The drive device 1 has a transmitting unit 11, which transmits the corresponding operating data 200 assigned to the operating parameters BP in a transmitting signal 101 to an observer module 20. The control unit ECU, the data logger 10, the transmitting unit 11, and the observer module 20 can be parts of an integrated or systemic control device.

[0060] The observer module 20 is capable of checking or monitoring the data 200 of the operating parameters BP transmitted with the transmission signal 101, i.e., the time-discrete operating data sequence of the operating data 200, over time. The observer, i.e., the observer module 20, has observer software 300 comprising several modules and data structures 300, which are not shown in detail here. Part of the modules and data structures 300 is at least one observer software 310 with a corresponding monitor algorithm 310A. The monitor algorithm 310A, as part of the observer software 310, is primarily designed to monitor the time-discrete operating data sequence of the operating data 200 with regard to certain aspects of an operating parameter BP represented therein; for example, here permanently and continuously (as symbolically represented as "24-hour" monitoring).

[0061] Such observers are also used, for example, in embodiments not shown here with a state controller for the reconstruction of non-measurable state variables or, as with the present embodiment, in time-discrete controls in which the measured variable cannot be updated in each cycle or in measurement technology as a replacement for measurements that are not technically or economically possible.

[0062] Using an observer of the observer module 20 provides for the reconstruction of the non-measurable variables (states)—here, operating parameters BP—from known input variables (e.g., manipulated variables or measurable disturbance variables) and output variables (measured variables) of an observed reference system. For this purpose, the observed reference system is emulated as a model M in the modules and data structures 300 (not shown in detail here), and the measurable state variables, which are therefore comparable to the reference system, can be tracked using a controller R or—as in the present case—simply monitored.

[0063] Part of the software structure 300 of the observer module 20 is also a Fig. 2 shown in more detail fault detection 320 with a fault detection algorithm 320A for detecting a fault relating to the drive device 1. The application of the fault detection algorithm 320A for the acquired operating data 200 in the fault detection 320 enables the detection of a fault relating to the operating parameter BP of the drive device 1 due to the specified aspect; this is described with reference to Fig. 2 and then the Fig. 4A and Fig.4B explained in more detail.

[0064] Based on the fault detection 320, as part of an automatic signaling 400, the sending of a fault indication 410 is provided, which can indicate a fault of the drive device 1 with respect to an operating parameter BP based on a result E of the fault detection algorithm 320A.

[0065] The fault indication 410 can be received directly or via a control of a control instance K in an actuator 30 and, after checking the result E of the fault indication 410, for example in a server device, can lead to the actuator 30 being activated to the drive device 1 via the control connection 500. The control connection 500, which is only symbolically shown here, can therefore actuate a travel path for the direct control and regulation of the drive unit 1 and / or trigger an action that is received as a signal from an external location and executes the action - in this respect, Fig. 1 The control connection 500 and, in this case, also the actuator 30 are shown abstractly. Within the framework of the observer structure, it is also fundamentally possible to initiate a feedback loop 600, also symbolically shown here, in the sense of feeding back an operating parameter BP, e.g., an actual value thereof and / or an error ERR, to the observer 20.

[0066] Fig. 2 shows a preferred embodiment of an observer structure for implementing a data-based monitoring method 1000 according to the concept of the invention, which is improved in relation to the aforementioned observer structure of the monitoring method 1000.

[0067] According to the provisions relating to Fig. 1 The data-based monitoring method 1000 is explained here with reference to a drive device 1 with a control device comprising a controller ECU, a data logger 10, which provides operating parameters BP comprising operating data 200 to an observer module 20. The observer module 20 comprises, as part of its modules and data structures 300, an observer software 310 or other monitor and a fault detection 320.

[0068] As part of a signaling 400, a fault indication 410 can also be automatically fed to a control instance K with a result E of the fault detection algorithm 320A to check the result E of the fault indication 410, in order to then activate a control connection 500 and / or a feedback 600 after checking the result E. The result E of the fault detection algorithm 320A can in this case be an automated message that is transmitted.

[0069] According to the concept of the invention, in the present embodiment, the Fig. 2 However, the control instance K is particularly advantageously configured with a discrimination unit DE that receives the result E from the observer module 20. This discrimination unit DE can, for example, be located in a central office or a central server, or be connected to another suitable server. In this respect, sufficient computing capacity and data background can be available for this discrimination unit DE and the decision regarding the result E. The discrimination unit DE makes a decision based on the appropriate examination of the result E with sufficient computing capacity and data background.

[0070] The result E is checked in the discrimination unit DE and classified according to a criterion or category for a false-positive result EF or a true-positive result EP . After checking for a false-positive result EF -- that is, whether the result E is recognized as artificial -- or for a true-positive result EP -- that is, whether the result E is recognized as genuine -- the result E qualifies as a false-positive result EF for further processing in the lower branch or a true-positive result EP in the upper branch, as described in Fig. 2 is shown.

[0071] As part of the two-strand automatic signaling 400, a corresponding qualification signal SE P or SE F is provided, which is also used in different ways in the present case.

[0072] The true-positive signal SE P based on the genuine - i.e. true-positive - result EP, namely in this embodiment only the true-positive signal SE P based on the true-positive signal SE P qualified as genuine - is made available in a first signaling line 420.1 of the control connection 500 in order to then provide an actuator 30.

[0073] The false-positive signal SE F based on the artificial --that is, false-positive-- result EF is made available in a second signaling strand 420.2 of the feedback 600 to the observer module 20; namely, in this embodiment, only the false-positive signal SE F -

[0074] In the present embodiment, the actuator 30 has a service and / or verification module 30.1. The service and / or verification module 30.1 automatically defines a service or verification action and forwards it as part of the automatic signaling 400 in a first signaling line 420.1 from the first module, in this respect, as a service and / or verification module 30.1. The service or verification action 540 can be influenced by or with a personnel deployment. The service or verification action 540 can also, but not only, be sent to service personnel. The service or verification action 540 is forwarded, in particular, as a service signal 430 to an execution module 30.2 of the service unit in the sense of an actuator 30. The actuator therefore comprises a second module, an execution module 30.2. The execution module 30.2 ensures the execution of the service or verification action 540.

[0075] The execution module 30.2 transmits a feedback signal SR after performing a defined, possibly predefined, service or verification action 540. The feedback signal SR is combined with the aforementioned false-positive signal SE F as part of the signaling 400.

[0076] The feedback signal SR is transmitted as part of the signaling 400 with the aforementioned false-positive signal SE F from the second signaling strand 420.2 as part of the feedback 600 to the observer module 20. The observer module 20 thus receives "feedback" not only on the executed service or verification action 540, but also on the aforementioned false-positive signal SE F from the second signaling strand 420.2.

[0077] The observer module 600 therefore receives not only the implementation status in the sense of an actual value of the operating parameter BP from the execution module 30.2, but also information about the artificial -- i.e. false-positive -- results EF that have been "filtered out" in the discrimination unit DE.

[0078] As a result, the design of the Fig. 2 a data-based monitoring method 1000 for the operation of a number of one or more drive devices 1, in particular for the operation of an internal combustion engine, wherein a drive device 1 has a controller and an operating parameter of the drive device is monitored on the basis of operating data for the operating parameter acquired by means of the control device, comprising the steps: Monitoring the operating parameter of the drive device with a monitor algorithm for the operating data, in particular a monitor algorithm 310A of an observer module of the control device; applying a fault detection algorithm to the acquired operating data of a fault detection to detect a fault relating to the operating parameter of the drive device; sending a fault indication from the fault detection for the drive device based on a result of the fault detection algorithm 320A; receiving the fault indication and checking the result, in particular receiving and checking in a server device of a control instance; activating a control connection to the drive device with respect to the operating parameter; and feeding back a monitoring value for the operating parameter to a feedback connection to the monitor algorithm.

[0079] According to the concept of the invention, it is provided that the result is qualified as false positive (artificial) or true positive (real) after testing, and the disturbance detection has a discrimination unit which sends the result, in particular only the result qualified as true-positive (genuine), to the control connection to an actuator and sends the result, in particular only the result qualified as false-positive (artificial), to the feedback connection to the observer module.

[0080] The monitoring value and, in particular, only the result qualified as false positive (artificial) is passed to the feedback connection to the observer module.

[0081] In the present case, the feedback connection from the actuator to the observer module is also designed as a learning feedback, wherein the method further comprises the steps: Adapting the monitoring algorithm and / or the fault detection algorithm with the false-positive (artificially) qualified result, and / or initiating a corrective measure on the drive device by means of the actuator depending on the true-positive (genuine) qualified result. Specifically, if the monitoring entity denies the presence of a fault, the fault detection algorithm can be adapted in such a way that the specificity of the fault detection algorithm for detecting this fault is increased. If the monitoring entity confirms the presence of the fault, the monitoring entity can output a rectification indication signal to initiate rectification of the fault. In other words, the presence of the fault is implied or confirmed for the true-positive (genuine) qualified result and / or the non-existence of the fault is implied or confirmed for the false-positive (artificially) qualified result. I.e.The result of the test will either confirm the existence of the fault or deny the existence of the fault.

[0082] Based on this, the observer module 600 is enabled to implement a learning algorithm, which is implemented as an "improved filter" via the discrimination unit DE provided in the control instance for the occurrence of false-positive results EF. The concept of the invention is realized in this embodiment in that operating data 200 for operating parameters BP are continuously monitored as a time-discrete sequence, and any disturbance or irregularity is detected. This is done in the discrimination unit DE using a model-based filter model that enables a plausibility check of the results E, as explained below.

[0083] Fig. 3 shows in a slightly modified manner --but fundamentally similar-- the structure of an observer system for the monitoring method 1000 in the structure, whereby reference is made here to the previously mentioned explanations using the same reference symbols.

[0084] In detail, the discrimination unit DE of the control entity K is shown here, which is spatially remote from the observer module 20 with or in a computing unit KR designed as a server with memory KS of the control entity K. The signaling 400 is thus primarily wireless between the control entity K and the observer module 20. The signaling is realized as part of the data logger 10 and the observer module 20, namely as a connection it can be implemented to or in a suitable control device; this can also be with the control ECU of the internal combustion engine 1 or include this, as symbolically shown in Fig. 3 is shown.

[0085] In a variation of the Fig. 2 In the structure shown, the true-positive result EP is transmitted to an actuator by means of the vehicle control unit ECU as part of a true-positive signaling SE P of the control connection 500. In this case, the actuator 30 is thus designed as an actuator on the vehicle by means of the control unit ECU in order to control the vehicle by changing the operating parameters via the actuator 30 in accordance with the true-positive result EP. The false-positive result EF is made available to the observer module 20 as part of the previously explained false-positive signaling SE F, in this case primarily to the observer software or other monitor 310 and / or the fault detection 320, namely for improvement within the framework of a learning loop of the monitor algorithm 310A or fault detection algorithm 320A stored there.

[0086] In the following, an example is explained -- building on the previously explained concept -- how an implementation of basic rules for reducing false positive results EF can be carried out within the framework of the algorithms 310A, 320A. It is also explained by way of example how these can be trained in order to form a corresponding filter F shown here, which is increasingly improved. This can be implemented within the framework of or for the monitor algorithm 310A and / or within the framework of or for the fault detection algorithm 320A. Overall, the concept leads to the processes in the embodiments according to Fig. 2 and Fig. 3 to enable automatic detection of service activities, such as Fig. 2 As a result, the filter F -- with time and increasing training -- will increasingly allow fewer false-positive results EF to pass through, so that the transmission of the results E by means of the previously explained fault indicator 410 already has a high quality.

[0087] For this purpose, a model M of the observer module 20, symbolically represented here, is provided, which enables a plausibility check of the results E. The plausibility check provides for valid basic rules in the sense of functional relationships or correlations to be checked between the available operating parameters or the transient, time-discrete sequence of the operating parameters. Thus, within the framework of the observer module 20, the operating parameters BP, such as the operating and measured variables, can be monitored. This can be done using an observer system of the monitoring method 1000 comprising a physical model M of the drive device with the operating parameters BP for reconstructing states of the drive device while taking the operating parameters BP into account.

[0088] Examples are given below with reference to Fig. 4A and Fig. 4B illustrated and explained by way of example. In this respect, reference is made to the previously mentioned reference symbols for the following description of the figures, whereby the same reference symbols are used for identical or similar parts or parts with the same or similar function.

[0089] One in Fig. 4A The first example of a model filter in the model M shown includes, for example, testing a correlation between the operating parameters BP for an "engine start" and an "oil temperature" T OIL . Fig. 4A For this purpose, a first exemplary possibility of implementation with respect to one of the aforementioned embodiments of a monitoring method 1000 is explained according to the inventive concept with respect to an individual drive unit 1, such as an internal combustion engine, with reference to the aforementioned reference numerals using a concrete example of a data set from the data logger 10, which is in Fig. 4A are shown in more detail.

[0090] In Fig. 4A The fault display 410 specifically comprises a time-discrete data sequence for operating parameters oil temperature T OIL , boost pressure PL , exhaust gas temperature T AG and idle speed VN for the drive unit 1.

[0091] These operating parameters are qualified within the observer module 20 using a model M. In this example, the monitoring method 1000 thus provides that the method is used to operate a single drive device 1; and the application of the fault detection algorithm includes a comparison of a current value of an operating parameter with at least part of the course of the operating parameter recorded over the operating period.

[0092] A previously explained discrimination unit DE can first produce and verify a result E. To do so, one or more questions must be verified or changes identified based on data and computing capacity, as indicated, for example, in the reference symbol for the discrimination unit DE: Are there any historical changes in the data for an individual drive unit? Has a significant step change been detected? Has a trend or trend change been identified?

[0093] The discrimination unit DE can also classify the detection results according to a criterion or into a category for a false positive result EF or a true positive result EP .

[0094] Specifically, in this case Fig. 4A In addition, it is indicated (symbolically with a check mark) how, according to the previously explained model filter M, some parameters are recognized or classified as normal or within the nominal range. This applies in this case to the boost pressure PL , the exhaust gas temperature T AG and the idle speeds (nominal speed VN ). However, the operating parameter for the oil temperature T OIL seems to show an anomaly, as can be seen in the enlarged illustration for T OIL in Fig. 4A is shown. The Discrimination Unit DE has sufficient computing capacity and data background for these findings and decisions based on an appropriate examination of a result E.

[0095] Such an inherently positive result E can be supplied as part of an automated fault indication 410 together with the operating state of the internal combustion engine (in the form of a drive unit 1 shown here). In contrast to the aforementioned operating state of a start-up process, a discrimination unit DE in the control instance K would recognize as result E that the internal combustion engine of the drive unit 1 is obviously in working mode and yet a critical gradient development in the oil temperature exists, albeit still within the limiting threshold ranges Limit LO and Limit HI.

[0096] The discrimination unit DE would therefore characterize this result as true-positive, i.e. real and not artificial.

[0097] A corresponding true-positive signal SE P can, for example, contain the message that the internal combustion engine is experiencing a critical gradient in the oil temperature T OIL. It can contain the corresponding data packet shown in DE around the operating time. It can contain the instruction to check the operating state of the internal combustion engine.

[0098] The result would thus be made available within the scope of a signaling 400 with a fault indication 410 with the true-positive signal SE P as a real - i.e., true-positive - result EP of the control connection 500 with a service signal 430, whereupon in this case an actuator 30 takes place in the sense of an automated service signaling, as previously explained. This is shown in Fig. 4A also only the path of the true-positive signaling 420.1 with the true-positive result EP .

[0099] Nevertheless, following the service action, an actual state will be determined in the sense of a signal feedback SR. A false-positive signal SE F with respect to a false-positive signal EF and a corresponding signal 420.2 and a feedback 600 will not occur in this case, since the verification in the discrimination unit has just performed a true-positive, qualified as genuine, test.

[0100] In this respect, the Fig. 4A The result E shown is different from the one in the example explained below for a vehicle start. For the correlation of the operating parameters BP "engine start" and "oil temperature" T OIL, a temporally increasing development of the oil temperature is initially recognized as positive but not qualified as a genuine - i.e., true-positive - result EP; i.e., in the case of a start-up process and the temporal development of the oil temperature as increasing, this is classified as normal operation. In terms of a signaling 400, this would be unsuitable for the first signaling branch 420.1, and would therefore be qualified as an artificial - i.e., false-positive - result EF. A result E that contains a noticeable increase in oil temperature would therefore NOT lead to the forwarding of a fault indication 410 for the first signaling branch 420.1 in the case of an engine start - rather, in the sense of a signaling 400, it would be fed to a second signaling branch 420.2 for feedback 600.

[0101] If, however, an engine start does not occur and a noticeable oil temperature increase is reported as part of the result E, this result E would be calculated as follows: Fig. 4A explained--then also qualified as a positive - but as a true-positive result EP and forwarded to an actuator 30 in the course of the fault indication 410. This example shows how the number of cases of artificial - i.e. false-positive - results EF in the control connection 500 can be reduced and at the same time the modules and data structures 300 such as observer software or other monitor 310 and fault detection 320 can be trained.

[0102] According to a second example of a model filter in model M, the detection of an artificial—i.e., false positive—result EF may, for example, involve the availability of a combination of information regarding the location of drive unit 1 (e.g., available via GPS) and the operating parameter BP. Information may be available that drive unit 1 is not running and information that a slight step increase in oil pressure is reported between the states of "engine stop" and "engine start." This could signal an oil filter change.

[0103] In this case, according to a model M of the filter, as previously explained, this CANNOT trigger a fault indication 410. A step increase in oil pressure P OIL would initially be recognized as positive, but not qualified as a genuine—i.e., true-positive—result EP. This means that between the states of an "engine stop" and an "engine start," the temporal development of a step increase in oil pressure P OIL would be classified as a normal oil filter change. In terms of signaling 400, this would be unsuitable for the first signaling line 420.1, and would therefore be qualified as an artificial—i.e., false-positive—result EF.

[0104] However, if one of the aforementioned conditions is not met, a corresponding fault indication 410 for the first signaling line 420.1 can trigger a filter change action via the actuator 30.

[0105] The process can improve the monitoring and fault detection algorithms 310A, 320A by means of algorithm training following the previously explained signaling 400 concerning a false-positive signal SE F and a true-positive signal SE P. A corresponding training algorithm T310, T320 can be provided to increase the detection rate for false-positive results and thus reduce the number of false-positive results.

[0106] In the sense of Fig. 3 For example, the model filter module can issue recommendations and also cause the actuator 30 to perform actions.

[0107] Such a model filter can be implemented in the controller or observer module 20, since it can be realized with comparatively little computational effort. However, in this embodiment, more complex checks of the results E for artificial or genuine are arranged as part of the discrimination unit DE in the control instance K.

[0108] However, as explained, a discrimination unit may also be part of the observer module 20 and be arranged downstream of a model filter M or directly downstream of the filter module F (comprising the monitor and disturbance detection algorithm 310A, 320A).

[0109] A third example may concern the combination of operating parameters BP for a situation explained below: An inlet pressure shows a noticeable pressure loss, while upstream of the turbocharger there is a constant pressure and a constant inlet air temperature.

[0110] In this case, this would trigger a fault indication 410, since this combination would be identified as positive and also qualified as a true—i.e., true-positive—result EP. A corresponding fault indication 410 can be generated for the first signaling line 420.1, and the service and / or verification module 30.1 can then recognize that there is a 70% probability that a turbocharger problem exists.

[0111] This may be based on the assumption that the turbocharger is obviously sucking in intake air at a constant pressure and at a constant intake temperature, but is not producing the pressure that it should produce for an intake pressure on the engine.

[0112] The actuator 30, in the sense of an actuator of the actuating unit, can thus, depending on the operating situation of the drive unit 1, provide a throttle position for the intake air at the inlet manifold in order to maintain the intake pressure or to operate the turbocharger at higher speeds. At the same time, the execution module 30.2 can issue a recommendation to check the turbocharger.

[0113] Such operating parameters BP and others other than those previously mentioned as examples can be recorded as a time-discrete operating data sequence, whereby the disturbance is recognized as a disturbance profile of an operating parameter BP for the time-discrete operating data sequence; for example, by means of a threshold value analysis of the operating data for an operating parameter BP.

[0114] For this purpose, a technical fault area of ​​the drive system can also be assigned to the fault and / or displayed, with the fault display comprising the fault history and the fault area. For example, each fault identification criterion can be assigned fault origin information that contains clues to the origin of the fault; and the fault indication signal indicates the fault origin information that is assigned to the fault identification criterion based on whose result the fault indication signal was sent.

[0115] Based on Fig. 4B A further fourth example of a model filter in the model M is explained. Fig. 4B shows a further exemplary implementation possibility with respect to one of the aforementioned advantageous embodiments of a monitoring method 1000; in this case with respect to a fleet with a plurality of drive units.

[0116] Fig. 4B shows, in a variant, the processes in a discrimination unit DE of a control authority K with regard to a fleet of a plurality of drive units 1.

[0117] In this case, a result E is transmitted as fault indication 410, in which a correlation is plotted with respect to a first specific operating parameter value BP1 and a second specific operating parameter value BP2; namely, for each of the drive units 1 N . In this respect, the corresponding operating parameters BP are referred to herein as BP n .

[0118] It can be seen that the operating parameter BP x for the drive unit X of the fleet, which is identified as different from the drive units N, lies significantly outside the ensemble of operating parameters BP n. This means that the method for operating a plurality of drive units N is used, and for each of the drive units N, at least one operating parameter BP n is recorded. The application of the fault detection algorithm comprises a comparison of a value of an operating parameter BP x of a drive device X with values ​​of the operating parameter BP n of the remaining plurality of drive devices N.

[0119] The qualification of the operating parameter BP x in contrast to the operating parameters BP n of the ensemble of operating parameters can be carried out via a distance measure A, which is shown here symbolically and merely as an example as part of a result E.

[0120] For example, a distance A to the average D of the ensemble of operating parameters BP n can be defined with respect to each of the operating parameters BP n .

[0121] For the operating parameter BP x it would be determined that over time t this distance A(t) exceeds a threshold value S.

[0122] In this case, a true-positive result is output as part of a true-positive signaling SE P with corresponding signaling 420.1 from a service and / or verification module 30.1 to an execution module 30.2.

[0123] Here too, this is linked to a recommendation in a service signal 430 that the drive unit X shows abnormal operating behavior with regard to the operating parameter BP x and needs to be checked with regard to the operating parameter BP x .

[0124] This procedure can be used in addition to the use of the Fig. 4B shown distance with respect to a threshold value-- a number of other statistical evaluation methods to give a result E. The distance can --as in Fig. 4B shown--with a substantially fixed distance dimension A. Additionally or alternatively, the distance --as shown in Fig. 4B shown—can also be given with a distance measure A(t) as a function of time. The aforementioned distance measure A, A(t) can be used to determine the distance. It can also be used to group error situations in a statistical manner, which can be defined more generally and / or can be determined independently of the nature of a specific operating parameter BP n . LIST OF REFERENCE SYMBOLS

[0125] 1Drive equipment 10Data logger 11Transmitting unit 30Actuator 30.1Service and / or verification module 30.2Execution module 101Transmitting signal 200Operating data 300Modules and data structures 310Observer software or other monitor 310AMonitoring algorithm 320Fault detection 320AFault detection algorithm 400Automatic signaling 410Fault indication 420.1, 420.2First, second signaling line 430Service signal 500Control connection 540Service or check action 600Feedback 1000Monitoring procedure A, A(t)Distance measure DEDiscrimination unit BP BP X BP N Operating parameter DAverage EEResult ECUControl EF Artificial --i.e. false-positive-- result EP Real --i.e. true-positive-- result ERRError FFilter KRRecomputing unit KSMemory KControl instance Limit LO and Limit HI Threshold ranges MModel PL Boost pressure RController SSteps SRFeedback signal SE P True-positive signal T310, T320Training algorithm T OIL Oil temperature T AG Exhaust gas temperature VN Idle speed.

Claims

1. A monitoring method for operating a number of one or more drive devices, in particular for operating an internal combustion engine, wherein a drive device has a control device, and an operating parameter of the drive device is monitored on the basis of operating data for the operating parameter, acquired by means of the control device, comprising the following steps: - monitoring the operating parameters of the drive device with a monitoring algorithm for the operating data, - applying a fault detection algorithm to the acquired operating data of a fault detection to detect a fault concerning the operating parameter of the drive device, - sending a fault indication from the fault detection for the drive device, based on a result of the fault detection algorithm; - receiving the fault indication and checking the result, - activating a control connection to the drive device with respect to the operating parameter and feeding back a monitoring value for the operating parameter to a feedback connection to the monitoring algorithm, characterised in that the result is qualified as false-positive or true-positive after testing, and - the interference detection has a discrimination unit which - passes the result, qualified as a true-positive, to the control connection to a control element, and - passes the result, qualified as a false-positive, to the feedback connection to the observer module.

2. The method according to claim 1, characterised in that the discrimination unit concerning the qualified result - only passes the result, qualified as true-positive, to the control connection to a control element, and - only passes the result, qualified as false-positive, to the feedback connection to the observer module.

3. The method according to claim 1 or 2, characterised in that the monitoring value and the result, in particular only the result, qualified as false-positive, are passed to the feedback connection to the observer module.

4. The method according to any one of the preceding claims, characterised in that the feedback connection from the control element to the observer module is configured as a learning feedback, and the method further comprises the following steps: - adapting the monitoring algorithm and / or the fault detection algorithm with the false-positive qualified result, and / or - initiating a corrective action on the drive device by means of the control element depending on the true-positive qualified result.

5. The method according to any one of the preceding claims, characterised in that, for the true-positive qualified result, a presence of the fault is implied or confirmed and / or for the false-positive qualified result, a non-presence of the fault is implied or confirmed.

6. The method according to any one of the preceding claims, characterised in that the monitoring of the operating parameters comprises monitoring the operating and measured variables applying an observer system, comprising a physical model of the drive device with the operating variables for reconstruction of states of the drive device taking into account the measured variables.

7. The method according to any one of the preceding claims, characterised in that the operating parameters are acquired as a time-discrete operating data sequence, wherein the fault is recognized as a fault curve of an operating parameter for the time-discrete operating data sequence.

8. The method according to any one of the preceding claims, characterised in that a technical fault area of the drive device is attributed to the fault and / or displayed, wherein the fault display comprises the fault history and the fault area.

9. The method according to any one of the preceding claims, wherein - the method is used to operate a single drive device; and - applying the fault detection algorithm involves a comparison of an actual value of an operating parameter with at least part of the history of the operating parameter, acquired over the operating period.

10. The method according to any one of the preceding claims, wherein - the method to operate a plurality of drive devices is used; - at least one operating parameter is acquired for each of the drive devices; and - applying the fault detection algorithm includes comparing a value of an operating parameter of a drive device with values of the operating parameter of the remaining ones of the plurality of drive devices.

11. A system, configured to carry out a monitoring method according to any one of claims 1 to 10 for operating a number of one or more drive devices, wherein a drive device has a control device and an operating parameter of the drive device is monitored on the basis of operating data for the operating parameter, acquired by means of the control device, wherein the system comprises: - the number of one or more drive devices, in particular an internal combustion engine, wherein - the control device is signal-connected to at least one drive device, and - a server device of a control authority is signal-connected to the control device.

12. The system according to claim 11, characterised in that the server device of the control authority has: - a receiver module for receiving the fault indication; - a verification module for verifying the result of the detected fault by the control authority after receiving the fault notification, - a discrimination unit that sends the true-positive qualified result to the control connection to an control element and sends the false-positive qualified result to the feedback connection to the observer module.

13. The system according to claim 11 or 12, characterised in that the control device of the drive device comprises: - an acquisition module, that is configured to acquire and monitor the operating parameters of the drive device with a monitoring algorithm of an observer module of the control device, - an execution module, that is configured to apply a fault detection algorithm of a fault detection to detect a fault, - a sending module, that is configured to send a fault indication from the fault detection for the drive device, based on a result of the fault detection algorithm.

14. The system according to any one of claims 11 to 13, characterised by one or more adaptation modules and / or remedy indication modules, configured for - adapting the monitoring algorithm and / or the fault detection algorithm with the false-positive qualified result, and / or - initiating a corrective action on the drive device by means of the control element depending on the true-positive qualified result.