DE202018006488U1 - Event detection system - Google Patents

Abstract

System (1) which is designed to recognize an event in a structure, the structure in turn comprising at least one component, wherein an event is any change in the expected behavior of a component, wherein the system (1) comprises:
- At least one sensor (2) which is designed to be arranged on the structure, the sensor (2) being designed to acquire data from the structure,
- a data analysis module which is designed to receive data from the at least one sensor (2), wherein the data analysis module comprises a database that records known events, each known event being associated with a specific change in the expected behavior of a component, and wherein the data analysis module (3) means for preprocessing the recorded data by the at least one sensor (2) using digital signal processing techniques and
- Comprises warning means, wherein the data analysis module is designed to evaluate the recorded data received from the at least one sensor (2), the evaluation being carried out by a comparison with the known events recorded in the database, and an event in accordance with the evaluation carried out, and
wherein the warning means are designed to inform a user about the detected event.

Description

TECHNICAL FIELD OF THE INVENTION

The present invention belongs to the field of structures, and more specifically belongs to the field of monitoring the condition of such a structure or a component thereof in order to characterize its capacity to serve its intended purpose.

In particular, the invention provides a system and an arrangement for recognizing and warning of changes in the expected behavior thereof for consideration when considering maintenance tasks.

GENERAL STATE OF THE ART

One of the most representative plans in technology today is the development of systems, structures and parts that are guaranteed to work safely as long as an agreed maintenance routine is met. Therefore, maintenance seeks to maintain the standard or design conditions over the life of such a system, structure or part.

In a broad sense, maintenance can include tasks such as inspection, renewal, repair or replacement of parts or components that must be performed during their service life. This stems from the inherent degradation in their performances, resulting in a change in their expected behavior caused by a deviation in their standard or design conditions. As a rule, this deviation is set as exceeding a certain predetermined limit. This is sometimes referred to as a failure or a near-failure condition that is protected by a margin of safety.

Traditionally, only compliance with the planned maintenance tasks ensured that these parts or systems would serve their intended purpose for a limited life. Nevertheless, maintenance tasks usually require the system to be stopped or even switched off in order to enable the activities to be carried out, which can result in disadvantages, delays and low production rates.

In addition, many parts or systems that need to be replaced after scheduled maintenance may already be due to undetected malfunctions that quickly escalate, fail, or damage (and have exceeded the predetermined limit well in advance). In contrast, other components may have a residual life, making them unnecessary to replace.

Therefore, planned maintenance requires that maintenance tasks be collected to be performed on components that are expected to cross their predetermined limit simultaneously or very closely together. Therefore, performing unnecessary maintenance on healthy components and wasting others with a residual lifespan leads to inefficient use of resources.

Although maintenance is common in all industries, all of the above is emphasized in air technology because aircraft maintenance is tightly regulated. Therefore, a significant amount of expensive parts and components are purposely wasted in order to meet the scheduled maintenance tasks.

SUMMARY OF THE INVENTION

The present invention provides a solution to the aforementioned problems by a system designed to recognize an event in a structure according to claim 1 and an arrangement according to claim 8. Preferred embodiments of the invention are defined in the dependent claims.

• - mindestens einen Sensor, der ausgelegt ist, an der Struktur angeordnet zu sein, wobei der Sensor ausgelegt ist, Daten von der Struktur zu erfassen,
• - ein Datenanalysemodul, das ausgelegt ist, Daten von dem mindestens einen Sensor zu empfangen, wobei das Datenanalysemodul eine Datenbank umfasst, die bekannte Ereignisse aufnimmt, wobei jedes bekannte Ereignis mit einer bestimmten Änderung am erwarteten Verhalten einer Komponente verbunden ist, und
• - Warnmittel,

wobei das Datenanalysemodul ausgelegt ist, die erfassten Daten auszuwerten, die von dem mindestens einen Sensor empfangen wurden, wobei die Auswertung durch einen Vergleich mit den in der Datenbank aufgenommenen bekannten Ereignissen durchgeführt wird, und ein Ereignis in Übereinstimmung mit der durchgeführten Auswertung zu erkennen, und
wobei die Warnmittel ausgelegt sind, einen Benutzer über das erkannte Ereignis zu informieren.In a first inventive aspect, the invention provides a system which is designed to recognize an event in a structure, the structure in turn comprising at least one component, wherein an event is any change in the expected behavior of a component, the system comprising :

• - At least one sensor which is designed to be arranged on the structure, the sensor being designed to acquire data from the structure,
• a data analysis module configured to receive data from the at least one sensor, the data analysis module comprising a database that records known events, each known event being associated with a specific change in the expected behavior of a component, and
• - warning means,

wherein the data analysis module is designed to evaluate the recorded data received from the at least one sensor, the evaluation being carried out by comparison with the known events recorded in the database, and to recognize an event in accordance with the evaluation carried out, and
wherein the warning means are designed to inform a user about the detected event.

Throughout this document, a structure is understood to include at least one component that can malfunction and therefore suffer a deviation in its behavior that leads to a change in the expected behavior. In the end, this variance can lead to a failure or a near-failure condition. The aforementioned deviation in the behavior of a component is linked to a specific event.

It should be noted that although an event is cited as any change in the expected behavior of a component, detection of the event is performed for the entire structure on which the sensor is placed. This leads to the fact that different components are mutually influenced in their behavior and therefore an event which is recognized for the entire structure can be derived for a specific component.

The at least one sensor present in the system should be arranged on the structure mentioned in order to acquire data from the structure and therefore from the at least one component that is part of the structure. The data recorded by each sensor is output by it as an entry of information that is to be processed by the data analysis module. Preferably the output is in the form of a signal.

The data analysis module can be arranged either together with the monitored structure or separately from it.

Apart from this, the database present in the data analysis module is a set of data which is associated with one or more known events, the data configuring mathematical models which are used in the evaluation carried out by the data analysis module.

A mathematical model is configured by mathematical elements that are linked together using mathematical operators. The combination of such elements with such mathematical operators defines a particular mathematical model that is stored in the database.

Such particular mathematical operators and mathematical elements are also stored in the database and are applied (in the form of a particular mathematical model specified for at least one event) to any entry received by the at least one sensor of the system in the data analysis module.

These mentioned mathematical models, which are contained in the database and are connected to one or more known events, provide a result in accordance with the mentioned entry of information originating from the at least one sensor.

As mentioned, data captured by a sensor is linked to the behavior of a component. Therefore, data associated with a component whose expected behavior is recognized to have changed is used to identify and categorize events as known events. For this reason, mathematical models are to be set up which establish an exact relationship between such known events and their associated data recorded by the at least one sensor.

This step is the so-called training phase, which is carried out with a time delay when the structure is operated and the amount of recorded data increases. This additional amount of data that is used for the training phase is hereinafter referred to as data collected in training. As a result, the mentioned mathematical elements or the mathematical model itself can be tuned to better predict the known event at hand.

Therefore, in the training phase, mathematical elements of the mathematical model are automatically adjusted using the data analysis model, with entries (especially the data captured by the sensor) being incorporated into the data analysis model, which provides confirmed outputs and allows the definition of additional entries that the corresponding mathematical model with each validate precise event.

Furthermore, as the amount of data collected increases, new events can be recognized and therefore new mathematical models can be formed from their associated data. These data are subsequently also recorded in the database of the data analysis module to ensure their implementation.

The data analysis model is then ready to carry out an evaluation of the recorded data by comparing the mentioned entries from the at least one sensor with the mathematical elements and operators of the mathematical model that are recorded in the database. The result of the mentioned comparison of the recorded information and the available data present in the database in accordance with the corresponding mathematical model provide a result. The result is evaluated, which provides an output which corresponds to the eventual detection of an event.

wobei y das vom Modell bereitgestellte Ergebnis ist, x die einbezogenen Einträge, die Teil der vom Sensor erfassten Daten sind, und ,W, b' bestimmte mathematische Elemente jedes Modells sind und ,·, +' mathematische Operatoren eines derartigen Modells sind, die eine Beziehung für die Elemente ,W, b' des Modells bereitstellen. Die vorliegende Kombination der Elemente ,W, b' mit den Operatoren ,·, +' bildet das gezeigte lineare mathematische Modell.It should be noted that the mathematical models mentioned include any type of model that enables an event to be obtained on the basis of the entries provided by the at least one sensor present in the system. The events taken into account are preferably recognized using a linear mathematical model, for example: $$y=\mathrm{W.}\cdot x+b,$$

where y is the result provided by the model, x is the entries involved which are part of the data acquired by the sensor, and 'W, b' are certain mathematical elements of each model and '·, +' are mathematical operators of such a model which are a Provide relationship for elements, W, b 'of the model. The present combination of the elements' W, b 'with the operators, ·, +' forms the linear mathematical model shown.

Each type of event can be associated with the same or a different mathematical model, whereas the same type of event can be associated with the same mathematical model even though it is performed using different mathematical elements or mathematical operators.

In addition, a mathematical model can be further developed with additional data acquired during training. This means that the mathematical elements or mathematical operators that are taken into account for a particular mathematical model can evolve in order to provide more precise mathematical elements which, based on the amount of entries or recorded data from the at least one sensor of the system, are better transferred to the appropriate event are adjusted.

Preferably, each mathematical model or mathematical elements thereof corresponds or correspond to different types of known events which are taken into account for the detection by the present system, and therefore the required data is acquired from the structure with the aid of a specific sensor.

Advantageously, the detection of an event can be achieved by acquiring data from the structure using any type of sensor, taking into account that the sensor is influenced by the event that is to be detected in order to provide accurate data to the mathematical model.

• • leicht erkannt werden, bevor sie zu einem großen Problem werden, und
• • durch geringfügige Anpassungen korrigiert werden.

Since the presence of a known event is inherently undesirable, the system according to the invention provides continuous tracking of the state of the structure in order to quickly identify and recognize such known events, the aforementioned recognition therefore being as accurate as necessary. In addition, minor problems can be avoided by continuously tracking the condition of the structure

• • Easily spotted before they become a major problem, and
• • be corrected by making minor adjustments.

This has a direct impact on the overall maintenance costs, which is dramatically reduced compared to the costs of standard normal maintenance that does not take into account the health of the particular component and uses a "generic recipe". In addition, this standard normal maintenance overlooks residual life of components of the structure.

With the aid of the system according to the invention, personalized maintenance is achieved for each particular structure in accordance with the particular behavior of the components thereof. Therefore, planned breaks in the operation of the structure for performing maintenance tasks can be unnecessary, or at least be greatly reduced, since the condition of the components is already known and the mentioned breaks can clearly be avoided by the output of the present system.

• - ein Kontaktsensor, wie ein Beschleunigungsmesser,
• - ein optischer Sensor,
• - ein Temperatursensor,
• - ein Luftfeuchtigkeitssensor,
• - ein Schalldrucksensor,
• - ein elektrischer Sensor,
• - ein magnetischer Sensor oder
• - ein IR-

Sensor.The at least one sensor of the system is preferably one of the following types:

• - a contact sensor, such as an accelerometer,
• - an optical sensor,
• - a temperature sensor,
• - a humidity sensor,
• - a sound pressure sensor,
• - an electrical sensor,
• - a magnetic sensor or
• - an IR

Sensor.

Advantageously, this allows the system to collect data from different fields or categories, which enables the system to identify known events of different types using the same data analysis module and one or more associated mathematical models.

This is particularly applicable in those structures with components that are contained in mechanisms or for components that are exposed to extreme working environments.

In a particular embodiment, at least one sensor is an accelerometer configured to collect frequency and amplitude data in accordance with the vibrations of the structure.

The condition of the component (s) is advantageously based on frequency patterns within the vibration spectrum. Apart from that, this provides information about the vibration patterns of the components. This is particularly applicable to components such as motor rotors, bearings or supports thereof.

This can enable the detection of events such as overshoots or imbalance in motor rotors.

Therefore, this particular embodiment is of particular interest for auxiliary engines or engines of the aircraft.

In a specific embodiment, the system further comprises recording means which are designed to store the data recorded by the at least one sensor. This enables the data analysis module to provide detection of an event based on data previously captured or to reevaluate possible events based on the data.

Advantageously, if the structure collapses for any reason, stored backup data may be of particular interest in identifying a new event associated with a failure that was previously unrelated.

In a preferred embodiment of the invention, the recording medium is a computer-readable storage medium or a data carrier.

In a specific embodiment, the data analysis module comprises means for preprocessing the recorded data by the at least one sensor using digital signal processing techniques. The techniques are, for example, a Fourier transformation, a wavelet transformation or noise removal.

The preprocessing enables the comparison and evaluation carried out by the data analysis module to be simpler, since the data to be evaluated are already taken into account in the most precise manner in order to be included as an entry in the corresponding mathematical model.

It must be observed that the preprocessing means are preferably additional mathematical operators and mathematical elements which are stored in the database. Therefore, additional calculations must be carried out in the mathematical model.

• - ein Schallsignal,
• - ein visuelles Signal oder
• - ein elektrisches Signal für ein entferntes System.

In a specific embodiment, the warning means are designed to generate at least one of the following warning signals:

• - a sound signal,
• - a visual signal or
• - an electrical signal for a remote system.

The warning means enable the system to warn a user operating the structure that any component is not working as expected, in order to take appropriate action if necessary. In principle, this is achieved through sound and visual signals that are easier to identify by users, such as the crew or operators.

Otherwise, in electronic systems, the user is a controller that can also suitably be notified of information about the behavior of any component, updated by electrical signals, in order to harmonize and optimize the overall operation while maximizing efficiency. Nevertheless, electrical signals can also be reported to the crew or remote operators of the system and subsequently shown on a display, for example.

In a particular embodiment, the present system is designed for use in an aircraft that includes an aerospace structure.

It goes without saying that the aircraft comprises the aerospace structure and the at least one sensor arranged thereon. The data analysis module and / or the warning means can, however, also be placed either on board the aircraft or remotely.

Advantageously in view of the economic disadvantage emphasized in the prior art when replacing components with a remaining service life in order to adhere to the maintenance schedule, this system enables better control of the condition of the components. It therefore enables the definition of a “dynamic” maintenance routine that takes into account the actual condition or the remaining service life of components until they reach their predefined limits.

The actual health or actual remaining life of a component is counted as the actual change in its expected behavior.

In a specific embodiment, the evaluation of the recorded data and the detection of the event are carried out in real time; i.e., the recorded data being performed as live data.

Therefore, mathematical elements and / or mathematical operators of the mathematical model are automatically adjusted by the data analysis model in the training phase, with entries in the data analysis model being included that provide confirmed results and enable the definition of exact elements that validate the corresponding mathematical model with each precise event .

The system can advantageously be updated continuously, as a result of which the database with the corresponding behavior of the component is improved at every moment.

Apart from this, since the occurrence of any event is continuously checked by the evaluation process, close control of the health component is achieved in order to immediately send a warning to the user.

• - eine Struktur, die wiederum mindestens eine Komponente umfasst, und
• - ein System nach einer der Ausführungsformen des ersten erfinderischen Gesichtspunkts.

In a second inventive aspect the invention provides an arrangement comprising:

• - A structure, which in turn comprises at least one component, and
• a system according to one of the embodiments of the first inventive aspect.

In a specific embodiment, the structure is an aviation structure, preferably an aviation structure on board an aircraft.

This enables better control of the condition or remaining service life of the components in the structure in order to carry out maintenance tasks in an optimized manner. It also allows an aircraft to avoid unnecessary breaks, which drastically reduces its exploitation. As an example, pauses can be any type of start prohibition, such as overhaul, long term, or short term maintenance tasks.

In a certain embodiment, the warning means are located in an aircraft, in particular in the pilot's cockpit.

This advantageously enables an event to be identified more quickly by the pilot - user - in order to take the necessary measures if necessary. Ultimately, this can reduce the spread of possible component failures by identifying the source from the very beginning.

In a specific embodiment, the data analysis module is arranged in an aircraft, and it is additionally designed to receive data from the at least one sensor with the aid of wiring.

In a specific embodiment, the data analysis module is arranged on the ground and designed to receive data from the at least one sensor by means of wireless communication.

Due to dimensional issues, it is often not possible to get additional equipment on board (which should be further certified by standard rules). Therefore this solution avoids the need for a physical connection between them, which mainly concerns weight and space problems.

In addition, it is likely that sophisticated equipment will be used on the ground rather than on-board equipment as there is no barrier to certification. As an additional benefit, a double-checking security measure can be achieved as the system is monitored both on the ground and on board.

In a specific embodiment, the data analysis module is designed to notify the warning means of the detection of an event by means of wireless communication.

This enables an alert to be distributed to a group of users. In particular, this can make it possible to inform various intended users in accordance with the detected event. In this sense, the data analysis module acts as a communication source sending the correct information to one or more intended recipients - users.

In a specific embodiment, the recording means are located in an aircraft.

Advantageously, this enables the data to be stored while they are being acquired, so that the data analysis module does not have to access the data immediately. This also enables the recorded data to be backed up.

All of the information in this specification (including the claims, description and the Drawings) described features and / or all of the steps can be combined in any combination, with the exception of combinations of such mutually exclusive features and / or steps.

Figure list

• 1a-1c Diese Figuren zeigen verschiedene Architekturen nach verschiedenen Ausführungsformen des Systems nach der Erfindung.
• 2 Diese Figur zeigt eine Abweichung des Verhaltens einer Komponente, was als Überschreiten einer vorab festgelegten Grenze festgesetzt ist, nach dem Stand der Technik.
• 3a-3b Diese Figuren zeigen Daten, die vom Sensor eines Systems nach der Erfindung erfasst wurden, wobei die Erfassung roh durchgeführt wird.
• 4a-4b Diese Figuren zeigen die erfassten, in 3a gezeigten Daten, nachdem sie mit digitalen Signalverarbeitungstechniken vorverarbeitet sind.
• 5 Diese Figur zeigt ein Beispiel einer Trainingsphase in Übereinstimmung mit den wie in 3a gezeigten erfassten Daten.
• 6 Diese Figur zeigt ein Ablaufdiagramm, das eine Ausführungsform zum Erkennen eines Ereignisses in einer Struktur beschreibt.

These and other features and advantages of the invention will be clearly understood in view of the detailed description of the invention resulting from a preferred embodiment of the invention given by way of example only and not limited thereto, with reference to the drawings.

• 1a-1c These figures show different architectures according to different embodiments of the system according to the invention.
• 2 This figure shows a deviation in the behavior of a component, which is defined as exceeding a predetermined limit, according to the prior art.
• 3a-3b These figures show data acquired by the sensor of a system according to the invention, the acquisition being carried out raw.
• 4a-4b These figures show the captured, in 3a after being preprocessed with digital signal processing techniques.
• 5 This figure shows an example of a training phase in accordance with those as in FIG 3a captured data shown.
• 6th This figure shows a flow diagram that describes an embodiment for recognizing an event in a structure.

DETAILED DESCRIPTION OF THE INVENTION

1a-1c show a group of different architectures according to different embodiments of the system ( 1 ) according to the invention. In particular shows 1a a first embodiment of the present system ( 1 ) on board an aircraft ( 10 ).

The system ( 1 ) is designed to detect an event in at least one structure of the aircraft ( 10 ) that is mounted in it. Such a structure comprises at least one component, for example an engine or a valve, whose behavior changed by an expectation is regarded as an event. While the system ( 1 ) on an airplane ( 1 ) is arranged, one or more sensors ( 2 ) arranged together with such a structure to collect data therefrom.

The data analysis module ( 3 ) is on board with the sensor ( 2 ), which in turn is arranged on the aircraft structure of the aircraft in order to receive the recorded data. After that, such recorded data is processed by the data analysis module ( 3 ) processed, and if an event is detected, warning means ( 4th ), such as a display or a snooze alarm, also on board, designed to inform a user afterwards about such a recognized event.

Preferably the warning means ( 4th ) to warn the user in the cockpit of the aircraft ( 10 ) suitable. This warning is configured by any warning signal, such as sound, visual, or a combination thereof. There is also provision for electrical signals to switch on certain reserve equipment.

Preferably the warning means ( 4th ) designed to notify the user only when an event is detected.

In addition, the system ( 1 ), which is completely on board, recording means ( 5 ) that enable the saved data to be saved as backup data.

In this entire first embodiment of the system ( 1 ), which is completely on board the aircraft ( 10 ), all connections are preferably wire connections.

1b shows a second embodiment of the present system ( 1 ), which is partly on board an aircraft ( 10 ) is arranged and is partially intended to be arranged on the ground. In the simplest example, the sensor ( 2 ) on board while the data analysis module ( 3 ) and the warning means ( 4th ) are arranged on the ground to provide information about an event.

Preferably the system comprises ( 1 ) Recording media ( 5 ) on the plane ( 10 ) are arranged on board, which store the recorded data from the structure.

• • über drahtlose Kommunikation an ein Datenanalysemodul (3) gesendet werden, das in einer Basisstation (20) am Boden angesiedelt ist, und/oder
• • in einem physischen Aufzeichnungsmittel (5) gespeichert und später auf verzögerte Weise ausgewertet werden.

Therefore, the recorded data

• • via wireless communication to a data analysis module ( 3 ) sent in a base station ( 20th ) is settled on the ground, and / or
• • in a physical recording medium ( 5 ) can be saved and later evaluated in a delayed manner.

• • an das Datenanalysemodul (3) des Bodenstandorts (20) gesendet werden sowie
• • vom Aufzeichnungsmittel (5) behalten werden und/oder
• • nur gespeichert und physisch zum Datenanalysemodul (3) des Bodenstandorts (20) transportiert werden.

The at least one sensor ( 2 ) and the recording medium ( 5 ), both of which are attached to the structure are arranged are connected by wiring. Therefore, the recorded data

• • to the data analysis module ( 3 ) of the soil location ( 20th ) are sent as well
• • from the recording medium ( 5 ) be kept and / or
• • only stored and physically to the data analysis module ( 3 ) of the soil location ( 20th ) be transported.

As soon as the data analysis module ( 3 ) carries out the evaluation with the received recorded data and finally recognizes an event, the warning means inform ( 4th ) accordingly. Communication between the two elements is preferably achieved by wiring on the floor. In addition, the intended user can be anyone who is responsible for the operation of the structure.

1c shows a third embodiment of the present system ( 1 ), which is partly on board an aircraft ( 10 ) and partially on the ground, as another example of the system described in the second embodiment ( 1 ).

As can be seen, the sensor ( 2 ) on the structure and therefore on board the aircraft ( 10 ) arranged. The data analysis module ( 3 ) on board the aircraft and can be connected to recording means ( 5 ), which provide a backup memory. The system also includes ( 1 ) at the air-conveyed end warning means ( 4th ) to inform the user of eventual detection of an event.

At the bottom end the system includes ( 1 ) an additional data analysis module ( 3 ) and additional warning devices ( 4th ), which evaluate the data received from the airborne end or finally inform you of a recognized event.

Therefore, the data recorded by the structure can either be entirely taken from the airborne data analysis module ( 3 ) or from the additional data analysis module ( 3 ) can be processed and evaluated on the ground or a combination thereof. In addition, both data analysis modules ( 3 ) evaluate the recorded data in real time and / or with a delay.

In particular, in this embodiment one of the data analysis modules ( 3 ) Means for preprocessing the recorded data by the sensor ( 2 ) with digital signal processing techniques, while the other evaluates the preprocessed data to detect an event.

Both data analysis modules ( 3 ) can communicate with each other using wireless communication. Therefore the data analysis module ( 3 ) laid out on the ground, from the data analysis module ( 3 ) on board the aircraft ( 10 ) to receive sent data.

2 shows how the state of the art defines a deviation in the behavior of a component. This example shows a graph showing amplitude versus time for a particular operating condition of a structure. These dates ( 30th ) were from a sensor ( 2 ), particularly detected by an accelerometer.

In the following, an accelerometer should be understood as a device that measures an actual acceleration of a structure. Because an accelerometer has several broad applications across many different technical fields, it was used to illustrate the prior art. Accelerometer applications can include aircraft and missile navigation systems, lathes, tablet computers and digital cameras so that images are always displayed upright on screens, flight stabilization of drones, etc.

Further, accelerometers measure vibration in the horizontal, vertical and axial directions on both the inboard and outboard motor parts. Aside from typical mechanical problems such as misaligned clutches and imbalance, vibration analysis can also identify electrical problems that cause mechanical vibrations. Some common electrical faults include air gap fluctuation, broken rotor bars, and corrugation on bearings.

Therefore it should be in 2 The diagram shown should be regarded as a typical example of the prior art. Vibration amplitude is the feature that quantifies the strength of the vibration, in this case captured data ( 30th ) in the form of vibrations in units of G (where “G” is equal to the acceleration of gravity, ie 9.8 m / s ² ) with time (t).

In addition, upper ( 31 ) and lower limits ( 32 ) can be observed with respect to certain levels of Gs in the accelerometer. In particular, these limits act as a predetermined limit on the peak-to-peak value of the vibration. Nevertheless, those skilled in the art can use any other value, such as a peak per se, an average, or rms values.

The peak-to-peak value indicates the maximum deviation of the shaft, a useful amount if, for example, the vibratory displacement of a structure or a component thereof is essential for a maximum load or considerations relating to mechanical clearance, and is therefore a representative value for is the health of a structure or component.

Therefore, in the prior art, a deviation is specified for the expected behavior of a component at the point where the vibration values are outside a predetermined range ( 31 , 32 ) are located. Examples of such exceeding ( 33 ) in the vibration values are in 2 shown.

Possible slight deviations that exceed the limits set in advance ( 31 , 32 ) do not exceed, but include an operation of the component that deviates from the expected, but since no exceedance ( 33 ) is noticed, it is assumed that the component continues to exhibit acceptable behavior in accordance with the predetermined limits mentioned ( 31 , 32 ) has a set threshold. This possible failure will only be noticed if the behavior of the component is influenced to exceed the pre-established limits ( 31 , 32 ) exceeds, while it remains unnoticed until then.

Hereinafter the same situation of data acquisition from an accelerometer arranged on an aerospace structure is shown in order to better represent the improvements of the present invention over the prior art.

3a shows data received from a sensor ( 2 ) captured in a raw way ( 40 ) are, according to the present embodiment of the invention. In particular, those from the sensor ( 2 ) recorded ( 40 ) Raw data is shown as a spectrogram performed via an accelerometer measurement, i.e. a 3-D diagram with the abscissa axis identified with time ("s"), the ordinate axis identified with frequency ("Hz") and the z- Axis is identified with the amplitude of the vibration. In addition, the minor to large amplitude values are identified on the z-axis with a gradient from the lightest to the darkest hues.

It should be noted that two vibration patterns in 3a are shown, that is, two strips of the spectrogram mentioned, which are separated by vertical boxes ( 41 , 42 ) are surrounded. These different vibration patterns or boxes ( 41 , 42 ) stipulate that at least two different operating conditions occurred at the time when the sensor ( 2 ) such data ( 40 ) has recorded.

In particular, the vertical boxes ( 41 , 42 ) surrounding stripes are part of their respective vibration patterns.

3a show the sensor ( 2 ), i.e. the accelerometer, collected data ( 40 ) as an entry of information that the data analysis module ( 3 ) is to be processed. Nevertheless, these can from the sensor ( 2 ) recorded data are preprocessed with digital signal processing techniques, such as a Fourier transformation, wavelet transformation, noise removal or the like.

It is highly recommended that the collected data ( 40 ) preprocessing in short time slices, such as one (1) second, to streamline computing resources. This is of particular applicability in the case where the data module analysis is on board in the aircraft. 3b shows a time slice ( 43 ) one ( 1 ) Second obtained from the data of 3a was extracted.

Nevertheless, it should be clear to those skilled in the art that the size of the disk depends on what is being sought (e.g. to recognize an event associated with an oscillation that takes place every 10 seconds, the disk should at least be larger than that).

It should therefore be clear that before the recorded raw data ( 40 ) a step to split this into several time slice samples ( 43 ) is carried out.

Show as an example 4a and 4b the recorded ( 43 ) Data as in 3b shown, already preprocessed with a Fourier transformation technique ( 44 , 45 ). Show accordingly 4a and 4b Amplitude or phase ("rad", radians) versus frequency ("Hz") as the result of such a Fourier transformation.

Such collected data, either preprocessed ( 44 , 45 ) or raw ( 40 , 43 ) are the entries made by the data analysis module ( 3 ) are to be received in order to be processed afterwards.

ist der Eintrag, der vom Sensor (2) gemessen ist und in das Datenanalysemodul (3) einbezogen wird, der in dieser Ausführungsform in Form eines 2x1-Vektors ist: $$\overrightarrow{X}=\left[\begin{array}{c}{x}_1\\ x_2\end{array}\right]$$

$\left(\overrightarrow{X}\right)$

is the entry that the sensor ( 2 ) is measured and in the data analysis module ( 3 ), which in this embodiment is in the form of a 2x1 vector: $$\overrightarrow{X}=\left[\begin{array}{c}{x}_1\\ x_2\end{array}\right]$$

wird vom Datenanalysemodul (3) ausgewertet, um ein Ereignis zu erkennen, das mit der überwachten Struktur oder einer Komponente davon verbunden ist. Das Datenanalysemodul (3) umfasst mathematische Modelle zum mathematischen Verarbeiten eines derartigen Eintrags.The entry $\left(\overrightarrow{X}\right)$

is used by the data analysis module ( 3 ) is evaluated in order to detect an event that is related to the monitored structure or a Component of which is connected. The data analysis module ( 3 ) includes mathematical models for mathematically processing such an entry.

in Form eines 2x1-Vektors ist, lauten die mathematischen Elemente in dieser Ausführungsform folgendermaßen: $$\begin{array}{cc}\overline{W}=\left[\begin{array}{cc}{w}_{11}& w_{12}\\ w_{21}& w_{22}\end{array}\right]& \overrightarrow{b}=\left[\begin{array}{c}{b}_1\\ b_2\end{array}\right]\end{array}$$

For example, considering a linear mathematical model such as y = W * x + b, 'W, b' are considered mathematical elements and '·, +' are considered mathematical operators. As the entry $\left(\overrightarrow{X}\right)$

is in the form of a 2x1 vector, the mathematical elements in this embodiment are as follows: $$\begin{array}{cc}\overline{\mathrm{W.}}=\left[\begin{array}{cc}{w}_{11}& w_{12}\\ w_{\mathrm{21st}}& w_{\mathrm{22nd}}\end{array}\right]& \overrightarrow{b}=\left[\begin{array}{c}{b}_1\\ b_2\end{array}\right]\end{array}$$

These particular mathematical operators and mathematical elements are stored in the database and are applied (in the form of this particular mathematical model specified for at least one event) to the entry in order to carry out a comparison.

in das Datenanalysemodul ein bestimmtes Ergebnis $\left(\overrightarrow{y}\right)$

vom mathematischen Modell bereitgestellt. Ein derartiges Ergebnis $\left(\overrightarrow{y}\right)$

wird danach unterschieden, um ein Ereingnis als ein individuelles Ergebnis $\left(\overrightarrow{y}\right)$

vom Erwarteten zu erkennen oder zu entdecken.As a result, will be in accordance with a particular entry $\left(\overrightarrow{X}\right)$

a certain result in the data analysis module $\left(\overrightarrow{y}\right)$

provided by the mathematical model. Such a result $\left(\overrightarrow{y}\right)$

is then differentiated to an event rather than an individual outcome $\left(\overrightarrow{y}\right)$

to recognize or discover from the expected.

identifiziert, die vom mathematischen Modell bereitgestellt werden.This means that the database of the data analysis module ( 3 ) records known events, each associated with a specific change in the expected behavior of a component, these known events will produce specific results $\left(\overrightarrow{y}\right)$

identified that are provided by the mathematical model.

in das mathematische Modell eingegeben ist, verschiedene Ergebnisse $\left(\overrightarrow{y_1},\overrightarrow{y_2}\right)$

im Vergleich bereit. Derartige Ergebnisse $\left(\overrightarrow{y_1},\overrightarrow{y_2}\right)$

werden im Auswertungsprozess als Ausgaben diskretisiert, um das Ergebnis mit einem bekannten Ereignis zu korrelieren. Falls irgendwelche der derartigen Ergebnisse bereits mit einem bekannten Ereignis verbunden sind, wird im Auswertungsprozess ein derartiges Ergebnis auch mit der Tatsache verbunden, dass das gleiche Ereignis erkannt wurde. Deshalb soll der Auswertungsprozess als ein Verbinden des Ergebnisses mit der Erkennung eines Ereignisses verstanden werden. Auf diese Weise sind die Vorgänge des Vergleichens und Auswertens, die im Datenanalysemodul (3) durchgeführt werden, mit Ergebnissen und Ausgaben des Models bzw. Systems (1) verbunden.For example, in the diagram of 3a , each vibration pattern represents each one as different entries $\left(\overrightarrow{{\mathrm{<figure-callout\; id="1"\; label="X"\; filenames="DE202018006488U1\_0027.png,DE202018006488U1\_0028.png"\; state="\{\{state\}\}">X</figure-callout>}}_1},\overrightarrow{{\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="DE202018006488U1\_0027.png,DE202018006488U1\_0028.png"\; state="\{\{state\}\}">X</figure-callout>}}_2}\right)$

entered into the mathematical model gives different results $\left(\overrightarrow{{\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="DE202018006488U1\_0027.png,DE202018006488U1\_0028.png"\; state="\{\{state\}\}">y</figure-callout>}}_1},\overrightarrow{{\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="DE202018006488U1\_0027.png,DE202018006488U1\_0028.png"\; state="\{\{state\}\}">y</figure-callout>}}_2}\right)$

ready in comparison. Such results $\left(\overrightarrow{{\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="DE202018006488U1\_0027.png,DE202018006488U1\_0028.png"\; state="\{\{state\}\}">y</figure-callout>}}_1},\overrightarrow{{\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="DE202018006488U1\_0027.png,DE202018006488U1\_0028.png"\; state="\{\{state\}\}">y</figure-callout>}}_2}\right)$

are discretized as outputs in the evaluation process in order to correlate the result with a known event. If any of these results are already associated with a known event, such a result is also associated with the fact that the same event was recognized in the evaluation process. Therefore the evaluation process should be understood as a connection of the result with the recognition of an event. In this way, the processes of comparing and evaluating, which are carried out in the data analysis module ( 3 ), with results and outputs from the model or system ( 1 ) connected.

As can easily be seen, the present system is ( 1 ) trained to recognize events from different patterns during the acquisition by a correlation with known events or operating conditions of structures.

As an example, if the expected behavior of a valve - a component - is to work fully open (100% flow capacity), an event should be considered when the valve (under the same working conditions) is partially closed (e.g. 93% Flow capacity) works, which therefore leads to a change in its expected behavior.

The expected behavior of the valve is in 3a identified as the first vibration pattern (100% flow capacity), while the second vibration pattern provides a result in the data analysis module associated with an output with the detection of an event, ie the valve operates with a flow capacity of 93%.

Therefore the system ( 1 ) previously trained to recognize events, such as partially closed (e.g. 93% flow capacity) operation of the valve, in accordance with collected data associated with a component whose expected behavior is recognized as changed. In this training phase, the data associated with components whose expected behavior is recognized as changed is used to identify and categorize events as known events.

This system advantageously enables ( 1 ) the detection of problems (identified as events) that the current state of the art would leave unnoticed. As already discussed, this can lead to preventive maintenance measures (or actions by the crew if the detection is carried out on board) that take into account the actual condition of the component. Identifying potential problems before they happen saves time and money because a component is no longer replaced before it fails and causes further damage.

In a further development of the system ( 1 ) the output generated by the data analysis module ( 3 ) is provided in the evaluation step, accompanied by a percentage of the certainty of the detected event, if one occurs.

5 FIG. 11 shows an example of a diagram which is generated in the training phase in accordance with the in 3a data shown is used. As already noted, the mathematical model is adapted by the data analysis module in the training phase in order to better recognize events or to be able to recognize other events.

First it is assumed that the vibration diagram ( 40 ) includes known events that occurred during the data acquisition by the sensor ( 2 ) were identified. For example, it can be known that a certain event took place at a certain point in time of the structure operation (or a component operation) and therefore this time slice of the recorded data should be used for correlation with a known event.

• • Das erste (100 %) (47) und das zweite (93 %) (48) Vibrationsmuster können verwendet werden, um das mathematische Modell fein abzustimmen, wie es mit den gleichen bekannten Ereignissen im Beispiel von 3a verbunden ist, und
• • das dritte Vibrationsmuster (49) kann verwendet werden, um ein mathematisches Modell zu konfigurieren, das die Erkennung eines neuen Ereignisses berücksichtigt.

At this point in time, the collected data ( 40 ) can either be used to tune the mathematical model in order to better predict the known events already taken into account and recorded, or used to identify new events that previously went unnoticed (according to the methods already mentioned in the prior art). In the in 5 embodiment shown:

• • The first (100%) ( 47 ) and the second (93%) ( 48 ) Vibration patterns can be used to fine tune the mathematical model, as it did with the same known events in the example of 3a connected, and
• • the third vibration pattern ( 49 ) can be used to configure a mathematical model that takes into account the detection of a new event.

Therefore, the recorded data is correctly classified beforehand depending on the events to be recognized.

The data within each vibration pattern ( 47 , 48 , 49 ) are divided into subordinate time slices. For each vibration pattern ( 47 , 48 , 49 ) a number of subordinate time slices configures the training group ( 47.1 , 48.1 , 49.1 ), while a different number of slices make up the control group ( 47.2 , 48.2 , 49.2 ) configured. This process is called the “labeling phase”. It is recommended that the vibration pattern remain stable throughout its entirety.

Such a control group ( 47.2 , 48.2 , 49.2 ) assesses how good the detection of the respective event is for which the data analysis module ( 3 ) is trained. In the example of 5 For each vibration pattern, odd subordinate time slices are created for the training group ( 47.1 , 48.1 , 49.1 ) is used, and subordinate time slices are used for the control group ( 47.2 , 48.2 , 49.2 ) is used. Nonetheless, those skilled in the art can see other ways of labeling the collected data in relation to the particular event for which the data analysis module ( 3 ) is trained.

For the mathematical example of 3a returning, the training phase can be explained as follows. With the third vibration pattern ( 49 ) collected data, it is correlated with a known event of the valve operating at 80% of its flow capacity. Therefore, a result that is provided by the mathematical model is to be evaluated with such an event of "Valve works with 80% of the flow capacity".

besser zu unterscheiden, wird ein bestimmtes Ergebnis bereitgestellt, wie zum Beispiel: $$\overrightarrow{y}=\left[\begin{array}{c}{y}_1\\ y_2\\ y_3\end{array}\right]$$

wobei y₁, y₂ und y₃ bereits gegeben und mit einem Ereignis von „Ventil arbeitet mit 80 % der Durchflusskapazität“ korreliert sind. Falls deshalb das Ergebnis $\left(\overrightarrow{y}\right)$

das vom mathematischen Modell bereitgestellt wird, vom gegebenen y₁, y₂ und y₃ abweicht, kann zumindest gefolgert werden, dass ein derartiges bestimmtes Ereignis noch nicht erkannt worden ist.To a future result $\left(\overrightarrow{y}\right)$

to distinguish better, a specific result is provided, such as: $$\overrightarrow{y}=\left[\begin{array}{c}{y}_1\\ y_2\\ y_3\end{array}\right]$$

where y ₁ , y ₂ and y _{3 are} already given and are correlated with an event of "Valve works with 80% of the flow capacity". If so the result $\left(\overrightarrow{y}\right)$

that is provided by the mathematical model deviates from the given y ₁ , y ₂ and y ₃ , it can at least be concluded that such a specific event has not yet been recognized.

In order for the other known events to be recognized, another given result of y ₁ , y ₂ and y _{3 must be} specified.

Therefore, the mathematical model stored in the database should be updated so that it can recognize any new event like the one mentioned.

die mit jeder untergeordneten Zeitscheibe der Trainingsphase des aktuellen Vibrationsmusters verbunden sind, werden in das Datenanalysemodul (3) eingeführt, das wiederum das frühere gegebene Ergebnis $\left(\overrightarrow{y}\right)$

bereitstellt.The entries $\left(\overrightarrow{X}\right)$

that are associated with each subordinate time slice of the training phase of the current vibration pattern are entered into the data analysis module ( 3 ) introduced, which in turn is the result given earlier $\left(\overrightarrow{y}\right)$

provides.

einem bestimmten bekannten Ereignis und dem gegebenen Ergebnis $\left(\overrightarrow{y}\right)$

entsprechen, wird eine Kombination von mathematischen Elementen mit den mathematischen Operatoren konfiguriert. Dies führt zu dem abgestimmten mathematischen Modell, das in der Datenbank des Datenanalysemoduls (3) aufgenommen ist.Therefore, with the limitations that the entries $\left(\overrightarrow{X}\right)$

a certain known event and the given outcome $\left(\overrightarrow{y}\right)$

will correspond to a combination of math Elements configured with the math operators. This leads to the coordinated mathematical model that is stored in the database of the data analysis module ( 3 ) is recorded.

It should be noted that the matched mathematical model is strongly influenced by the basic mathematical model to be used. Even so, depending on the complexity and nature of the data being collected, those skilled in the art can use different basic mathematical models.

For example, with the linear math model used above (i.e. y = W x + b), the fitted math elements are: $$\begin{array}{cc}\overline{\mathrm{W.}}=\left[\begin{array}{cc}{w}_{11}& w_{12}\\ w_{\mathrm{21st}}& w_{\mathrm{22nd}}\\ w_{31}& w_{32}\end{array}\right]& \overrightarrow{b}=\left[\begin{array}{c}{b}_1\\ b_2\\ b_3\end{array}\right]\end{array}$$

nicht ändern soll (d. h., sie bleibt 2x1 in diesem bestimmten Beispiel), muss angemerkt werden, dass die frühere Dimension des Elements W (d. h. 2×2) zweifellos auf die Dimension 3×2 erhöht wird, da ein neues Ereignis erkannt werden kann. Gleichermaßen wird die Dimension des Elements $\overrightarrow{\text{b}}$

auch von 2×1 auf 3×1 erhöht, da das neue Ereignis erkannt werden kann.Since the dimension of the entries $\left(\overrightarrow{X}\right)$

should not change (i.e., it remains 2x1 in this particular example), it must be noted that the element's earlier dimension W. (ie 2 × 2) is undoubtedly increased to the dimension 3 × 2, since a new event can be recognized. Similarly, the dimension of the element $\overrightarrow{\text{b}}$

also increased from 2 × 1 to 3 × 1, since the new event can be detected.

Once the fitted math model is configured, the child time slices that make up the training group are used as additional entries to validate the model with the event at hand. Optionally, the validity of the adapted mathematical model can be considered to have been reached when a certain success rate is obtained.

Furthermore, it is also possible for the training phase to use preprocessed data for the identification phase.

As noted above, the mathematical model can be configured to provide a result associated with an output that incorporates a percent certainty.

• - Bereitstellen einer Struktur und eines Systems, das umfasst:
  • mindestens einen Sensor, der ausgelegt ist, an der Struktur angeordnet zu sein, um Daten von dieser zu erfassen,
  • ein Datenanalysemodul mit einer Datenbank, wobei das Datenanalysemodul Daten von dem mindestens einen Sensor empfängt, und
  • Warnmittel,
• - Anordnen des mindestens einen Sensors in der zu überwachenden Struktur,
• - Erfassen von Daten des Verhaltens der Struktur und Bereitstellen derartiger Daten $\left(\overrightarrow{X}\right)$
  
  an das Datenanalysemodul,
• - Vergleichen der erfassten Daten mit bekannten Ereignissen, die in der Datenbank des Datenanalysemoduls aufgenommen sind,
• - Bereitstellen eines Ergebnisses $\left(\overrightarrow{y}\right)$
  
  des erwähnten Vergleichs,
• - Auswerten eines derartigen Ergebnisses durch Unterscheidung, wodurch eine Ausgabe bereitgestellt wird,
• - Erkennen eines Ereignisses in Übereinstimmung mit einer derartigen Ausgabe und
• - Informieren des Benutzers über das erkannte Ereignis mithilfe des Warnmittels.

6th shows the steps for detecting an event in a monitored structure. In short, the steps are shown as follows:

• - Providing a structure and a system that includes:
  • at least one sensor which is designed to be arranged on the structure in order to acquire data from it,
  • a data analysis module having a database, the data analysis module receiving data from the at least one sensor, and
  • Warning means,
• - Arranging the at least one sensor in the structure to be monitored,
• - Gathering data on the behavior of the structure and providing such data $\left(\overrightarrow{X}\right)$
  
  to the data analysis module,
• - Compare the recorded data with known events that are recorded in the database of the data analysis module,
• - Providing a result $\left(\overrightarrow{y}\right)$
  
  the mentioned comparison,
• - Evaluation of such a result by differentiation, whereby an output is provided,
• - detecting an event in accordance with such an output and
• - Informing the user of the detected event using the warning means.

As already noted, a training phase can take place parallel to the acquisition of data from the structure in order to adapt the comparison between the acquired data and the known events that are recorded in the database of the data analysis module.

1. i. Aufteilen des Intervalls von erfassten Daten, in dem ein bekanntes Ereignis eintritt, in Datenscheiben,
2. ii. Kennzeichnen der Scheiben in eine Trainingsgruppe und eine Kontrollgruppe,
3. iii. Abstimmen der Datenbank des Datenanalysemoduls, die bekannte Ereignisse aufnimmt, um einen genaueren Vergleich zwischen den Daten und den bekannten Ereignissen darin zu erhalten,
4. iv. Validieren des abgestimmten Vergleichs durch die Datenscheiben der Kontrollgruppe und
5. v. Aktualisieren der Datenbank des Datenanalysemoduls, die bekannte Ereignisse aufnimmt,

vorgesehen.Therefore, in this training phase, between the steps of collecting data and comparing the collected data, there are also steps to:

1. i. Dividing the interval of recorded data in which a known event occurs into data slices,
2. ii. Marking the targets in a training group and a control group,
3. iii. Tuning the database of the data analysis module, which records known events, in order to obtain a more precise comparison between the data and the known events in it,
4. iv. Validation of the agreed comparison by the data slices of the control group and
5. v. Updating the data analysis module's database that records known events,

intended.

6th also shows with dashed arrows that an iteration takes place between the steps of reconciling the database and its subsequent validation until a certain stopping condition is reached.

The recorded data used in the training phase can either be used to adapt the database of the data analysis module in order to either better predict known events that have already been recorded or to identify new events.

Claims (14)

System (1) which is designed to recognize an event in a structure, the structure in turn comprising at least one component, wherein an event is any change in the expected behavior of a component, wherein the system (1) comprises: - At least one sensor (2) which is designed to be arranged on the structure, the sensor (2) being designed to acquire data from the structure, - a data analysis module which is designed to receive data from the at least one sensor (2), wherein the data analysis module comprises a database that records known events, each known event being associated with a specific change in the expected behavior of a component, and wherein the data analysis module (3) means for preprocessing the recorded data by the at least one sensor (2) using digital signal processing techniques and - Comprises warning means, wherein the data analysis module is designed to evaluate the recorded data received from the at least one sensor (2), the evaluation being carried out by a comparison with the known events recorded in the database, and an event in accordance with the evaluation carried out, and wherein the warning means are designed to inform a user about the detected event. System (1) according to Claim 1 wherein the system (1) further comprises recording means which are designed to store the data detected by the at least one sensor (2). System (1) according to one of the preceding claims, wherein the at least one sensor (2) is one of the following types: - a contact sensor (2), such as an accelerometer, - an optical sensor (2), - a temperature sensor (2), - a humidity sensor (2), - a sound pressure sensor (2), - an electrical sensor (2), - a magnetic sensor (2) or - an IR sensor (2). System (1) according to one of the preceding claims, wherein the warning means are designed to generate at least one of the following warning signals: - a sound signal, - a visual signal or - an electrical signal for a remote system (1). The system (1) according to any one of the preceding claims, wherein at least one sensor (2) is an accelerometer which is designed to acquire frequency and amplitude data in accordance with the vibrations of the structure. System (1) according to one of the preceding claims, for use in an aircraft comprising an aerospace structure. System (1) according to one of the preceding claims, wherein the evaluation of the recorded data and the detection of the event are carried out in real time. Arrangement comprising: - a structure which in turn comprises at least one component, and - a system (1) according to one of the Claims 1 to 7th . Arrangement according to the preceding claim, wherein the structure is an aerospace structure, preferably an aerospace structure on board an aircraft. Arrangement according to one of the Claims 8 or 9 , wherein the warning means are arranged in an aircraft, in particular in the cockpit. Arrangement according to one of the Claims 8 to 10 , wherein the data analysis module (3) is arranged in an aircraft, and wherein it is additionally designed to receive data from the at least one sensor (2) with the aid of wiring. Arrangement according to one of the Claims 8 to 11 , wherein the data analysis module (3) is arranged on the ground and is designed to receive data from the at least one sensor (2) by means of wireless communication. Arrangement according to one of the Claims 8 to 12 , wherein the data analysis module (3) is designed to notify the warning means of the detection of an event by means of wireless communication. Arrangement according to one of the Claims 8 to 13th wherein the recording means are arranged in an aircraft.

Images