DE102018214704A1 - Method for preparing a processing of a sensor signal and method for evaluating a sensor signal - Google Patents

Abstract

Method for preparing processing of a sensor signal (22) of a sensor (14), wherein an energy harvester (12) detects a harvester signal (24), the sensor signal (22) and the harvester signal (24) correlate with one another, and from the Harvester signal (24) data are obtained which are used to initialize algorithms which are used in a later processing of the sensor signal (22), input variables for the algorithms being determined from the data.

Description

The invention relates to a method for preparing a processing of a sensor signal, a method for evaluating a sensor signal and an arrangement for performing the method. The invention further relates to a computer program and a machine-readable storage medium for carrying out the method.

State of the art

Sensors, which are also referred to as detectors, are components that can qualitatively or quantitatively record physical or chemical properties or the material properties of their surroundings. The detected quantities are converted into a typically electrical signal, which is referred to herein as the sensor signal. This sensor signal thus carries information about the size or sizes that are to be recorded with the sensor. To evaluate the sensor signal, it has to be processed in many cases so that the desired information can be extracted from the sensor signal.

The first step after reading a sensor signal is to subject this sensor signal to signal processing. This step is necessary in order to extract useful information from the sensor signal and thus to obtain such information from the sensor. Today's algorithms for signal processing are mostly based on spectral analyzes, such as DFT (Discrete Fourier Transform and FFT (Fast Fourier Transform). For example, it is known to analyze the spectrum of a vibration sensor in order to measure the damage frequencies of a bearing Analysis is useful, but it is very expensive for an embedded electronic system with limited resources and limited power consumption. This is especially true when the relevant area to be measured is only a narrow band of the entire spectrum of the sensor signal.

Different methods, such as Sparse DFT, were developed to avoid the calculation of areas of no interest. An alternative is time domain analysis, in which the signal is only filtered in the appropriate frequency ranges with an analog or digital filter and features of the filtered signal are calculated.

Another important aspect is the resolution of the DFT / FFT transformation, which represents a quality variable, with the user default setting of which a compromise between power consumption and complexity or resolution quality must always be taken. Thus, an improved resolution is only achieved through an increased computational effort, which increases the energy requirement.

In order to avoid this, algorithms have been developed with which, on the one hand, good resolution can be achieved and, on the other hand, the number of transformation points can be left the same. Algorithms such as B. Zoom-FFT or the Goertzel algorithm, try to focus the spectral analysis on the area desired by the user after previously specifying the user and thus optimize the above-mentioned compromise.

These alternatives to the FFT mentioned allow, depending on the user request, either to reduce the computing effort or to increase the resolution, but both initially require user input with regard to the frequency range to be examined. Such information is unknown for many applications, e.g. B. in wear monitoring of a linear drive, in which the damage frequencies can be in different areas.

The publication DE 10 2013 201 618 A1 describes a method for signal processing which is carried out in connection with a vibration pickup unit, which comprises a vibration pickup and a signal processing circuit. The signal processing circuit derives information from the signals of the vibration sensor with regard to the amount of loss grains detected by the vibration sensor and additionally examines the signals for vibrations which indicate possible damage to moving parts. For this purpose, the signal processing circuit carries out a Fourier analysis.

Disclosure of the invention

Against this background, a method having the features of claim 1 and an arrangement according to claim 9 are presented. A computer program according to claim 12 and a machine-readable storage medium according to claim 13 are also presented. Embodiments result from the dependent claims and the description.

The method described is used to prepare the processing of the sensor signal obtained, this preparation being used to obtain or extract information or data from a harvester signal which, for example, provides an FFT or DFT in the subsequent processing of the sensor signal , can be used. This information or data is used, for example, to initialize the algorithms used in later processing, that is to say to make the data Determines input variables for the algorithms and / or identifies ranges, e.g. frequency ranges, that are of interest during processing, so that a focused procedure can be selected during later processing: This means, for example, that only ranges, in particular frequency ranges ; of importance are examined or processed.

A so-called energy harvester is used in the method and arrangement described. Energy harvesting, which is also referred to as energy harvesting, is to be understood as the extraction of small amounts of electrical energy from sources such as, for example, ambient temperature, vibration or air flow. This harvesting is primarily carried out for devices with low output. There are possibilities of energy harvesting, for example, in energy conversion through the piezoelectric effect, the thermoelectric effect and the photoelectric effect. Arrangements used for this are called energy harvesters.

For this purpose, energy harvesters can be used which, according to their mechanical structure, have the optimum performance for each structural unit with exactly one excitation frequency. Since broadband energy harvesters are required in the actual application in many cases, the construction principle of the harvester array lends itself. In this, a large number of substructures or individual structures are set up parallel to one another. In terms of construction, each substructure is trimmed to a natural frequency that is slightly out of tune with its neighbor.

In the case of the method presented, the user inputs required according to the prior art are thus provided by system-internal signals of an energy harvester or an energy harvester array in a user-autonomous manner and in an embodiment also in an energy-autonomous manner. In addition to its core function, which consists in providing energy, the Energy Harvester is an array of mechanical-electrical bandpass filters due to its structure. Energy supply and signal provision are thus provided for the pre-control of the evaluation algorithms by the Harvester Array.

It must be taken into account that harvester arrays form mechanical bandpass filters, since, depending on their quality, they are only excited in narrow frequency bands in the vicinity of the resonance. Individual harvester structures such as B. Piezo patches, such as. Bending beams, are therefore particularly sensitive to the excitation frequency (component) of the vibration source in frequency ranges near their natural frequency.

After the preparation of the processing presented here, the actual processing or evaluation of the sensor signal can then take place.

The arrangement presented serves to carry out the described method and typically includes a sensor, an energy harvester and a computing unit, in particular a microcontroller.

In an embodiment, the arrangement represents an energy-saving, energy-self-sufficient sensor system for the detection of vibrations, which has at least one structure for generating energy from vibrations, a microcontroller, at least one sensor for the detection of vibrations, and signal processing algorithms for frequency-resolved analysis of the sensor signals. The supply voltage signal generated by the at least one structure for generating the supply energy of the sensor system is passed to at least one signal input of the microcontroller and used to prepare the processing of the sensor signal, which precedes the actual processing of the sensor signal for its evaluation. During this preparation, information is obtained which is used for the initialization of algorithms, in particular standard algorithms, which in turn are used to subsequently process the signal for evaluating it. This serves to reduce the data and energy consumption of the microcontroller. In addition, the sampling rate of the at least one sensor can be adjusted.

The data or information obtained is thus used in the signal analysis in such a way that only the frequency range separated by the energy generating device is subject to signal detection and signal analysis.

In addition to the frequency position, the associated (voltage) amplitude provides information about how much energy is currently available to the sensor system. For example, the frequency band information can be used to adapt the coefficients of the digital filter or the frequency bins or spectral components of the Goertzel algorithm and, at the same time, to match energy supply and energy demand.

The frequency ranges to be distinguished are separated in terms of construction by the eigenmodes of the associated individual harvester structures. The distances between the bandpass filters are defined by the constructive definition of the harvester array substructures.

Certain array substructures, their voltage, are thus excited as a function of frequency represents the input variable for the algorithm for refining the analysis of the corresponding frequency range. In addition to the frequency position, the associated (voltage) amplitude provides information about how much energy is currently available to the sensor system. As already stated, the coefficients of the digital filter or the frequency bins or spectral components of the Goertzel algorithm can be adapted with the frequency band information and at the same time energy supply and energy demand can be coordinated.

The assignment of the generated electrical signals of the individual patches to the microcontroller unit enables the evaluation algorithm to be initialized or precontrolled. The energy harvester itself thus delivers the pilot signal size via its vibration behavior in order to be able to carry out the software evaluation with increased resolution or reduced computing effort at precisely this point. Multiples at higher frequencies or higher orders can also be selected. So there is only a dependency but no direct coupling to exactly this fundamental frequency. The energy harvester therefore sends trigger signals to the consumer without having to use sensors or software.

• - Die vorgestellte Vorbereitung der Verarbeitung, die auch als Vorsteuerung bezeichnet werden kann, schränkt die softwareseitig zu analysierenden Frequenzbereiche stark ein. So kann bspw. durch Anschalten bzw. Ausschalten der entsprechenden Filter einer Filterbankstruktur oder durch Anpassen der Nullpositionen der Sparse-DFT-Matrix ein Geschwindigkeitsvorteil erzielt werden.
• - Die Vorbereitung der Verarbeitung ermöglicht es, die in der Realität immer limitierte Rechenleistung und Rechengeschwindigkeit auf einen schmalbandigen Bereich zu konzentrieren.
• - Eine Erhöhung der Genauigkeit, wie z. B. durch Anpassen des Bereichs in Zoom FFT oder in Goertzel-Algorithmus, wird ermöglicht.
• - Der Energy-Harvester selbst sorgt durch seinen mechanischen Aufbau und die daraus abgeleiteten Signale zur Vorbereitung der Verarbeitung und der eigentlichen Verarbeitung dafür, dass der an ihn angeschlossene elektrische Verbraucher, typischerweise eine Recheneinheit, wie bspw. ein Mikrocontroller, weniger Leistung verbraucht.
• - Eine intelligente Energiequelle sorgt für Reduzierung des Energieverbrauchs auf Verbraucherseite.
• - Die erforderliche Rechenleistung kann reduziert werden, womit auch Hardware-Kosten eingespart werden können.
• - Die Kombination zwischen einem automatisch-initialisierten bzw. eingestellten Algorithmus, wie bspw. Zoom FFT oder Goertzel, und einem Festkomma-Format ergibt eine bessere Auflösung als bei einem nicht-fokussierten Algorithmus (FFT) mit einem Fließkomma-Format. Festkomma-Prozessoren sind hardwareseitig kostengünstiger als Fließkomma-Prozessoren.
• - Die obligatorische Pufferspeichergröße für „geerntete“ Energie kann reduziert werden.
• - Der Energiebedarf und die erzeugte Datenmenge kann durch eine dynamisch anpassbare Abtastrate gesenkt werden. Der Harvester liefert nur Signale aus einem Frequenzbereich, woraus ein Leerlaufzustand, bei einem elektrischen Leerlaufzustand bedeutet dies keine Spannung, bei einem mechanischen Leerzustand bedeutet dies keine Linearbewegung, der zu überwachenden Maschine bzw. Anlage geschlossen werden kann, so dass folglich für die Verschleißüberwachung eine langsamere Abtastrate notwendig ist, womit der Energiebedarf des Verbrauchers gesenkt wird und weniger Datenvolumen erzeugt wird. Die dynamische Abtastrate kann neben der Frequenzlage, bestimmt durch die jeweilige Substruktur als Bandpassfilter, auch anhand der Signalintensität, bspw. der Spannung, je Verbindungsanschluss bzw. Patch beeinflusst werden. Damit ist der Harvester durch die erfindungsgemäße Verknüpfung zum Mikrocontroller in der Lage, Lastzustände der zu überwachenden Komponente zu ermitteln, deren Charakterisierung sonst im Zeit- und Frequenzbereich erforderlich ist.

The method presented has, at least in some of the versions, a number of advantages:

• - The presented preparation of the processing, which can also be called pre-control, severely limits the frequency ranges to be analyzed by the software. A speed advantage can be achieved, for example, by switching the corresponding filters of a filter bank structure on or off or by adapting the zero positions of the Sparse DFT matrix.
• - The preparation of the processing makes it possible to concentrate the computing power and computing speed, which is always limited in reality, on a narrowband area.
• - An increase in accuracy, such as. B. by adjusting the area in Zoom FFT or in Goertzel algorithm, is made possible.
• - Thanks to its mechanical structure and the signals derived from it to prepare for processing and the actual processing, the Energy Harvester itself ensures that the electrical consumer connected to it, typically a computing unit such as a microcontroller, uses less power.
• - An intelligent energy source ensures a reduction in energy consumption on the consumer side.
• - The required computing power can be reduced, which can also save hardware costs.
• - The combination between an automatically initialized or set algorithm, such as Zoom FFT or Goertzel, and a fixed-point format results in better resolution than a non-focused algorithm (FFT) with a floating-point format. Fixed-point processors are cheaper in terms of hardware than floating-point processors.
• - The mandatory buffer storage size for "harvested" energy can be reduced.
• - The energy requirement and the amount of data generated can be reduced by a dynamically adjustable sampling rate. The harvester only delivers signals from a frequency range, from which an idle state, with an electrical idle state this means no voltage, with a mechanical idle state this means no linear movement of the machine or system to be monitored, so that consequently a slower one for wear monitoring Sampling rate is necessary, which reduces the energy requirements of the consumer and less data volume is generated. In addition to the frequency position, determined by the respective substructure as a bandpass filter, the dynamic sampling rate can also be influenced on the basis of the signal intensity, for example the voltage, for each connection connection or patch. With the linkage according to the invention to the microcontroller, the harvester is thus able to determine load states of the component to be monitored, the characterization of which is otherwise required in the time and frequency domain.

Further advantages and refinements of the invention result from the description and the accompanying drawings.

It goes without saying that the features mentioned above and those yet to be explained below can be used not only in the combination specified in each case, but also in other combinations or on their own without departing from the scope of the present invention.

• 1 zeigt in einer schematischen Darstellung eine Ausführung der beschriebenen Anordnung.
• 2 zeigt in einem Graphen eine Bandpass-Struktur.
• 3 zeigt in einer vereinfachten Darstellung ein Energy-Harvester-Array.
• 4 zeigt Harvester-Signale eines Energy-Harvester-Arrays.
• 5 zeigt in einem Flussdiagramm eine Signalverarbeitung nach dem Stand der Technik.
• 6 zeigt in einem Flussdiagramm eine Signalverarbeitung entsprechend dem vorgestellten Verfahren.

Brief description of the drawings

• 1 shows a schematic representation of an embodiment of the arrangement described.
• 2 shows a bandpass structure in a graph.
• 3 shows a simplified representation of an energy harvester array.
• 4 shows harvester signals from an energy harvester array.
• 5 shows a signal processing according to the prior art in a flowchart.
• 6 shows in a flowchart a signal processing according to the presented method.

Embodiments of the invention

The invention is illustrated schematically in the drawings using embodiments and is described in detail below with reference to the drawings.

1 shows a schematic representation of an embodiment of an arrangement for performing the method described, the total with the reference number 10 is designated. The order 10 is designed as an energy-autonomous or energy-self-sufficient and user-autonomous or user-self-sufficient unit. The order 10 exhibits an energy harvester 12 , a sensor 14 , in this case a motion sensor for recording or detecting vibrations or vibrations, a buffer memory 16 , a microcontroller 18th as a computing unit and a transmitting unit 20th on. The order 10 is thus designed as a vibration sensor system.

These vibrations are from the sensor 14 into a sensor signal 22 , typically an electrical signal carrying information about the vibrations is converted to the microcontroller 18th is forwarded for evaluation. The vibrations are also from the energy harvester 12 , which for this purpose comprises a suitable detector or sensor, or to which one is assigned. The energy harvester 12 also converts the vibrations into a signal, the so-called harvester signal 24 that also goes to the microcontroller 18th and in this version also to the buffer tank 16 is passed on. It should be noted that the harvester signal 24 and the sensor signal 22 correlate or are related to each other. This means that the two signals 22 and 24 are related or interdependent. This is clear in this version, since both the sensor 14 as well as the energy harvester 12 absorb the same vibrations or at least vibrations that are related to each other.

The harvester signal 24 carries or transmits an energy or a time profile of a power, the energy being used to the microcontroller 18th to operate. The energy transfer can be direct and / or buffered via the buffer memory 16 respectively. This energy is used in the microcontroller 18th the preparation of the processing of the sensor signal 22 according to the method presented and / or the further processing of this sensor signal 22 taking into account the presented preparation for processing the sensor signal 22 perform. It should be noted that this further processing can also take place externally. It is important that the arrangement 10 through the supply via the energy harvester 12 can be done autonomously. In addition, this can also be done autonomously since user input is no longer required. This is done from the information that the harvester signal 24 carries, data obtained during the later processing of the sensor signal 22 be taken into account by preparing this processing accordingly.

It is therefore between an initial preparation for processing the sensor signal 22 and the subsequent processing of the sensor signal 22 distinguished. When preparing the processing of the sensor signal 22 are those from the harvester signal 24 The data obtained is used for the sensor signal 22 to be prepared in such a way that it can be suitably processed or analyzed in the subsequent processing, which typically includes a frequency analysis, such as, for example, FFT or DFT. The preparation of the processing thus serves, for example, to obtain input variables or values for the initialization of algorithms used in the processing and to carry out this initialization accordingly. From the from the harvester signal 24 The data obtained or taking this data into account are used to determine or determine input variables for the algorithms used in the processing. This can mean, for example, that only a certain frequency range of the sensor signal is left in the later processing 22 is looked at. The sensor signal 22 is thus prepared for the subsequent processing in the preparation of the processing presented here, or the algorithms for the subsequent processing are set up or initialized accordingly.

The sending unit 20th serves to give the information obtained to the outside world.

2 shows in a graph 50 , on the abscissa 52 the frequency [Hz] and at its ordinate 54 a size [dB] is plotted, a bandpass structure which has, for example, the harvester signal. Different courses for different filters are shown, namely a first course 60 for a first filter, a second course 62 for a second filter, a third course 64 for a third filter, a fourth gradient 66 for a fourth filter, a fifth gradient 68 for a fifth filter, a sixth gradient 70 for a sixth filter, a seventh course 72 for a seventh filter and an eighth gradient 74 for an eighth filter.

This bandpass structure can be used to obtain information or data which can be used to initialize algorithms for processing a sensor signal. For example, it can be determined, in particular by means of the amplitudes, in which spectral range there is increased energy.

3 shows an embodiment of an energy harvester array, the total with the reference number 100 is designated and a first substructure 102 , a second substructure 104 and includes a third substructure. The substructures 102 . 104 . 106 each have a weight 112 . 114 respectively. 116 that have a different mass and each have a bending strip 122 . 124 respectively. 126 with piezoelectric properties with a conductive base 130 are connected. Vibrations cause the bending strips to vibrate 122 . 124 . 126 , this creates an electrical voltage, like this with arrows 132 . 134 and 136 is made clear. These tensions 132 . 134 . 136 are different and all carry information about the triggering vibrations.

4 shows in a graph 150 on the abscissa 152 the frequency and its ordinate 154 an achievement P is plotted, signal curves of the outputs of an energy harvester array with three substructures, namely a first curve 160 for a first substructure, a second course 162 for a second substructure and a third course 164 for a third substructure. It should be noted that each substructure has its own transfer function so that the respective bending strip is excited depending on the excitation frequency. The courses 160 . 162 and 164 are evaluated in order to obtain data that are taken into account when processing a sensor signal.

5 shows a conventional signal processing in a flowchart. A distinction is made between a user-autonomous area 200 and a user-specific area 202 , First, an energy supply is provided. In a first step 210 then a measurement is carried out, there is a conventional energy supply (block 212 ) in front. Then takes place in step 214 the processing of the measurement results, e.g. using Sparse DFT, FFT, Görtzel etc. This is, for example, a manual user input (block 216 ) based. Finally in step 218 output and / or further processing of the results.

6 shows a flowchart of a procedure that is energy-autonomous and user-autonomous. First, an energy supply is provided. In a first step 250 the measurement then takes place. An energy harvester array 252 provides the energy for this. Then takes place in step 254 the use of the analysis algorithms, such as, for example, Sparse DFT, FFT, Görtzel, etc., data or information obtained by the energy harvester array being taken into account for the initialization of the algorithms. Finally in step 256 the output and / or further processing of the results obtained.

Regardless of the embodiment shown, the data that are obtained by the energy harvester or from the harvester signal, for. B. acceleration signals, for example vibration data. The data is generated on the energy harvester by its operating principle, which is based on a mechanical-electrical conversion effect, such as piezoelectric, triboelectric, electrodynamic (inductive) or electrostatic (capacitive).

In the process, the data are obtained from the harvester signal or transmitted with the harvester signal.

QUOTES INCLUDE IN THE DESCRIPTION

This list of documents listed by the applicant has been generated automatically and is only included for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent literature cited

• DE 102013201618 A1 [0008]

Claims (14)

Method for preparing a processing of a sensor signal (22) of a sensor (14), wherein an energy harvester (12) detects a harvester signal (24), the sensor signal (22) and the harvester signal (24) correlate with one another, and The Harvester signal (24) is used to obtain data which are used to initialize algorithms which are used when the sensor signal (22) is subsequently processed, the input variables for the algorithms being determined from the data. Procedure according to Claim 1 , in which an energy which is transmitted with the harvester signal (24) is used to prepare the processing of the sensor signal (22). Procedure according to Claim 1 or 2 , in which the sensor signal (22) after the preparation is processed in accordance with the preparation, the energy of the harvester signal (24) and the data obtained from the harvester signal (24) being used for this purpose. Procedure according to one of the Claims 1 to 3 , in which the preparation is used to select a spectral range from the sensor signal (22). Procedure according to one of the Claims 1 to 4 , in which the sampling rate of the sensor (14) is adjusted taking into account the data obtained. Procedure according to one of the Claims 1 to 5 , in which the sensor signal (22) is processed after the preparation. Procedure according to one of the Claims 1 to 6 , in which the sensor (14) picks up the same size that is detected by the energy harvester (12) for mechanical-electrical conversion. Procedure according to Claim 7 , in which vibrations are recorded with the sensor (14), these vibrations being at least partially detected by the energy harvester. Arrangement for preparing a processing of a sensor signal (22) which is used to carry out a method according to one of the Claims 1 to 8th is set up. Arrangement after Claim 9 which comprises a sensor (14), an energy harvester (12) and a computing unit. Arrangement after Claim 9 or 10 , in which the energy harvester (12) is designed as an energy harvester array (100). Method for evaluating a sensor signal, characterized in that the evaluation of the sensor signal is the method for preparing the sensor signal according to one of the Claims 1 to 8th goes ahead. Computer program with program code means, which is set up to implement a method according to one of the Claims 1 to 8th to be executed when the computer program on a computing unit, in particular a computing unit in an arrangement (10) according to Claim 9 , is performed. Machine-readable storage medium with a computer program stored on it Claim 13 ,

Images