METHOD FOR CONDITION EXAMINATION OF A PUMPING UNIT AS WELL AS SOFTWARE APPLICATION, STORAGE MEDIUM AND EXAMINATION DEVICE FOR EXECUTING THE METHOD

Description

[0001] The invention relates to a method for examining the condition of at least one component of a pump unit, in particular for determining the speed of the pump unit or with regard to a fault by analyzing the airborne sound emitted by the pump unit, comprising recording a first audio signal during operation of the pump unit at a first geometric location with a first distance to the pump unit and recording a second audio signal during operation of the pump unit at a second geometric location with a second distance to the pump unit that is less than the first distance.

[0002] The spectral analysis of noise emitted by pump units during normal operation or in the event of a fault is well known. For example, German patent application DE102009022107 A1 proposes, among other things, measuring the airborne noise and determining the rotational sound frequency using signal analysis, such as frequency analysis, such as a fast Fourier transform (FFT) or autocorrelation, from which the drive speed is then determined.

[0003] For acoustic measurements, knowing the exact distance between the sound source and the microphone is necessary, as the sound level depends on distance and location. Only then can measured absolute sound levels (in decibels) be compared.

[0004] A distance of 1 m is generally recommended for acoustic measurements. Noise levels from objects at other distances can be calculated analytically from this. To measure the exact distance, commonly available physical measuring devices are used that are different from the acoustic measuring device, such as a folding rule, tape measure, or laser rangefinder. While the acoustic measuring device, which includes the microphone, is mounted on a tripod, the distance is measured manually, possibly using two hands, and adjusted accordingly. It goes without saying that this method is cumbersome, time-consuming, and often error-prone.

[0005] To analyze the faults of a pump unit, characteristic vibration frequencies within its spectrum are examined. However, acoustic measurements are controversial among experts. Instead, vibration sensors are preferred. These are mounted on the pump unit and their sensor signals are evaluated, as described, for example, in EP 1972793 A1 or EP 3 563 062 B1.

[0006] However, international application WO 2015 / 197141 A1 describes the installation of a vibration sensor as disadvantageous and complex, and proposes, as an alternative, the measurement of a sound signal emanating from the pump unit using a microphone in or on a communication handset. This signal is then processed and used to identify a sound-specific condition or error. Identification is performed by comparing it with reference values ​​stored for the specific pump unit, whereby a model must be loaded into the testing software at the beginning of the testing procedure. WO 2015 / 197141 A1 proposes showing the user on a display the positions at which the handset or microphone should be positioned. Furthermore, the user should move the handset relative to the pump unit, i.e.bring it closer or further away from it, pan it to the right, left, up or down, while the examination software evaluates the received acoustic signal in real time in terms of signal quality and amplitude and provides feedback to the user when an optimal position is reached. Positioning takes between 10 seconds and one minute and the distance is between 5 cm and one meter. If the examination software reports the distance as appropriate, the actual noise measurement takes place in a second step. The disadvantage of this is that there is no guarantee that the distance identified as appropriate is identical to the distance previously shown on the display and to be expected. The described method also reaches its limitations if the pump is particularly loud due to a fault.This results in the user being suggested an incorrect distance for the audio recording, thus failing to detect any potential pump unit errors. Furthermore, there is no guarantee that a previously identified appropriate position will be maintained during the measurement. This means that a different distance may inadvertently be present during the subsequent measurement, particularly if the handset is held in the hand, especially since it has previously been swung back and forth. Finally, the actual distance of the handset or microphone from the pump unit is not further considered in the evaluation in WO 2015 / 197141 A1, nor is it saved.

[0007] In EP 3563062 B1, the method from WO 2015 / 197141 A1 is described as problematic because the sound signals must be recorded in the same way so that the data can be processed and usable results are obtained. In practice, however, it is not possible to influence the position of the handheld device and the ambient noise with sufficient accuracy to apply the method to detect both engine mount faults and cavitation. Even if the user attempts to take multiple measurements independently, it would be very difficult, if not impossible, to detect engine mount faults and cavitation. Therefore, EP 3563062 B1 again proposes the use of a vibration sensor.

[0008] WO 2016 / 059112 A also describes a method for acoustically determining the condition of a pump unit, for example, a heating circulation pump. A standard smartphone can be used for recording, which processes the acoustic data using signal processing and examines it for any anomalies or errors that may occur. For example, bearing damage or cavitation within the pump can be detected. The advantage of this acoustic fault analysis during operation of the pump unit is that a service technician can examine the pump unit for faults during operation without special sensors or a complex test setup. However, EP 3242036 A1 describes this method as disadvantageous because it cannot be carried out in noisy environments, or at least not with the required reliability.EP 3242036 A1 therefore proposes a completely different method, namely a video analysis in which the speeds of changes between pixels and pixel groups of successive images of a video sequence are determined and the state of the pump unit is determined on the basis of the determined speeds of changes.

[0009] It is therefore an unsolved problem to determine the condition and / or any fault of a pump unit by evaluating the airborne sound it emits with sufficient accuracy when, on the one hand, ambient noise is present and, on the other hand, an inaccurate positioning of the microphone relative to the pump unit is possible.

[0010] It is therefore an object of the present invention to provide a method that overcomes the aforementioned disadvantages and enables a correct result when examining the airborne noise emitted by the pump unit with regard to a possible fault in the pump unit, despite the presence of noise from a secondary noise source. Furthermore, it is an object of the invention to provide a software application, a computer-readable storage medium, and a mobile examination device for carrying out the method. In addition, the positioning of the examination device relative to the pump unit is to be facilitated, and the examination result is to be made insensitive to deviations of the actual measurement position from a predetermined measurement position, in particular, independent of a specification of a measurement position, thus rendering this obsolete.

[0011] These objects are achieved by the method according to claim 1, the software application according to claim 19, the storage medium according to claim 20, and the examination device according to claim 21. Advantageous further developments are specified in the subclaims and are explained below.

[0012] According to the invention, a method is proposed for examining the condition of at least one component of a pump unit, in particular for determining the speed of the pump unit or with regard to a fault by analyzing the airborne noise emitted by the pump unit, comprising recording at least one first audio signal during operation of the pump unit at a first geometric location with a first distance from the pump unit, and recording a second audio signal during operation of the pump unit at a second geometric location with a second distance from the pump unit that is less than the first distance, wherein the condition of the pump unit is determined by analyzing a signal from the pump unit that has been cleaned of background noise from at least one background noise source and is reconstructed by comparing amplitude values ​​of the first and second audio signals.

[0013] For the purposes of the present invention, a pump unit is understood to be an arrangement comprising a pump unit that pumps a liquid, in particular a centrifugal pump unit, and a drive unit that drives it, such as an electric motor. In addition, pump electronics can be provided for controlling and / or regulating the electric motor. Alternatively, there can also be just a terminal box with electrical connections for the electric motor. The components mentioned can form a structural unit or simply interact functionally, and thus be arranged in a distributed manner. The electric motor can drive the pump unit directly or via a clutch or a gearbox. In the case of a direct drive, the motor shaft can simultaneously be the pump shaft and carry at least one impeller. The pump unit can be single-stage or multi-stage, i.e. have one or more impellers.The pump unit can be either a dry-running or a wet-running pump. The pump electronics can be mounted on the electric motor housing or arranged separately.

[0014] When a condition is mentioned in the context of the present invention, this refers to the operating state of a component of the pump unit or of the entire pump unit in general, i.e., whether the pump unit or the component in question is operating within normal acoustic parameters, or in other words, without any particular acoustic abnormality, or whether such an acoustic abnormality is present that indicates a fault. In particular, the operating state can also be understood as the current speed at which the pump unit is currently operating.

[0015] A fault is defined as an incorrect operating condition of the pump unit or one of its components that, from an acoustic perspective, results in at least one abnormality in the airborne noise emitted by the pump unit. Examples of such a fault include bearing damage, more specifically damage to a rolling or plain bearing, cavitation, a damaged impeller, damage to a mechanical seal, a blockage, an imbalance in a rotating part, or even the fact that one or more screws are loose. All of these technical faults lead to certain vibrations in the pump unit or at least one of its components, which are also noticeable in the airborne noise.

[0016] The acoustic abnormality and / or the speed can be detected in a known manner from the frequency spectrum of the airborne sound emitted by the pump unit due to the presence of certain frequencies or peaks there, or in the case of the speed, can be estimated.

[0017] The reconstructed signal can thus be a reconstructed frequency spectrum. Alternatively, it can be a different type of spectrum, for example, an order spectrum for order analysis. This enables analysis of the noise or vibrations of rotating machines. Unlike in frequency analysis, the amplitude representing the energy content of the noise is not plotted against the frequency, but against the order, where the order is a multiple of the rotational speed. In this respect, it is a normalized spectrum. According to another alternative, the reconstructed signal can be a time signal that has been back-transformed from a previously reconstructed spectrum.

[0018] Furthermore, a software application for a mobile examination device with a display, at least one operating element and an acoustic sensor and optionally an optical sensor is proposed, comprising program instructions which, when executed by the examination device, cause the examination device to carry out the method according to the invention.

[0019] Furthermore, the invention relates to a computer-readable storage medium with such a software application. Finally, a mobile examination device is also proposed, comprising a display, at least one control element, an acoustic sensor for making an audio recording, and optionally an optical sensor, wherein the examination device is configured to carry out the method according to the invention. In particular, the examination device can contain the aforementioned storage medium for this purpose and comprise a processor for executing the software application stored on the storage medium to carry out the method.

[0020] While the determination of the condition of the pump unit or the analysis of the airborne noise emitted by the pump unit for condition detection (speed or error) is carried out in a manner known per se to those skilled in the art, a core idea of ​​the invention is to determine amplitude values ​​of a physical or mathematical quantity from at least two audio signals recorded at different distances from the pump unit and to compare them with each other and, through the comparison, to identify which noise components originate from the pump unit and which do not originate from it, but from a secondary noise source, in order to then reconstruct a signal in which the noise components emanating from this secondary noise source are eliminated. The inventive procedure uses a distance law, i.e.the knowledge that the airborne sound emitted by the pump unit has a first value at the first location with the first distance from the pump unit, which, according to natural law, must have changed to a specific and thus expected second value at the second location with the second distance from the pump unit. This test for the validity of the distance law can be performed by comparing the amplitude values ​​of a quantity in the first and second audio signals with each other to determine whether they are in the relationship expected based on the distance law.

[0021] The airborne sound is recorded using an acoustic sensor. As an acoustic quantity of the airborne sound, the sound pressure, for example, can be measured, so that the first and second audio signals each form a sound pressure-time curve. The sound pressure is a scalar quantity that follows the distance law 1 / r, where r is the distance or radius to the pump unit. In other words, the sound pressure p is inversely proportional to the distance r (p ~ 1 / r). It follows that a sound pressure p at a first location p(r 1 ) with a distance r 1 from a reference point only has half the value if the distance to the reference point is doubling to r 2 = 2r 1 , p(r 2 ) = 0.5 p(r 1 ). This can be taken into account in the inventive comparison of the amplitude values ​​to determine whether a certain frequency in the airborne sound originates from the pump unit or from the secondary noise source.

[0022] As an alternative to sound pressure, sound power or sound intensity can be used. Sound power and sound intensity are energetic quantities that each follow the distance law 1 / r 2<, where r is again the distance or radius to the pump unit. In other words, the sound intensity I or sound power P is inversely proportional to the square of the distance r 2< (I~1 / r 2< , P~1 / r 2< ). It follows that a sound intensity I or a sound power P at a first location with a distance r 1 from a reference point only has 25% of its original value if the distance to the reference point is doubled to r 2 = 2r 1 , I(r 2 ) = 0.25·I(r 1 ), P(r 2 ) = 0.25·P(r 1 ).

[0023] Ideally, the acoustic sensor is or will be directed toward the pump unit. A microphone can serve as the sensor, preferably a directional or adjustable microphone.

[0024] It makes sense for the first and second distances to be significantly different. This means that they don't just differ by a few millimeters or a few centimeters. Rather, the first distance should be at least 20%, preferably at least 50%, greater than the second distance. Ideally, the first distance should be about twice as large as the second distance. However, the distance should also be chosen with consideration for the distance between the pump unit and the background noise source and should not be too large compared to it.

[0025] In particular, the first distance should be smaller than the distance between the pump unit and the secondary noise source.

[0026] The amplitude values ​​are, for example, those of spectral components determined from the first and second audio signals. A spectral component is a portion of a spectrum that describes a distribution function of the physical or mathematical quantity in the corresponding audio signal. An amplitude value thus contains information about the quantity in which a certain spectral component is present in the corresponding audio signal.

[0027] According to one embodiment, to eliminate at least one background noise in the first and second audio signals, a spectrum of each of the first and second audio signals is determined. Amplitude values ​​of corresponding spectral components of the two spectra or spectra derived therefrom can then be related to each other, with the reconstructed signal being formed from the amplitude values ​​of those spectral components of one of the two spectra or spectra derived therefrom whose ratio lies beyond a certain limit. The quotient of the aforementioned amplitude values ​​can be used as the ratio.

[0028] The spectra can, for example, be frequency spectra and the spectral components can be frequencies.

[0029] Alternatively, the spectra can be order spectra in which the spectral components are orders, more precisely multiples of a rotational speed.

[0030] Preferably, these are discrete spectra to enable their numerical calculation and evaluation.

[0031] It is advantageous if the reconstructed signal is formed from the amplitude values ​​of the spectral components of the spectrum of the second audio signal or the spectrum derived from it, because this was recorded closer to the pump unit and thus has the better ratio between the useful signal and noise.

[0032] In an advantageous further development, a number of dominant spectral components can first be determined and selected from each of the two spectra. These dominant spectral components, together with their amplitude values, form the respective derived spectrum. The amplitude values ​​are then set in relation to each other. In other words, the respective derived spectrum corresponds to the respective spectrum of the first or second audio recording with regard to the dominant spectral components, but the amplitude values ​​of the remaining spectral components are set to zero. The dominant spectral components can be referred to as "peaks" in the spectrum. Figuratively speaking, the aforementioned procedure thins out the spectrum consisting of individual peaks. This filters out the spectral components caused by noise in the audio recordings. Furthermore, subsequent signal processing is less computationally intensive.

[0033] The number X of dominant spectral components may be unknown in one variant. For example, the dominant spectral components can be determined by comparing their amplitude with a threshold value, so that all spectral components that exceed this threshold are dominant. In this case, the exact number cannot be determined at the beginning of the determination.

[0034] In another variant, however, the number N of dominant spectral components to be selected can be specified. In this case, the selected frequencies are those belonging to the N spectral components with the highest amplitude. Compared to using a threshold, this variant leads to better noise filtering.

[0035] The specification of a number N of dominant spectral components to be selected can refer globally to the entire spectrum, which in the case of a frequency spectrum, for example, can extend between 0 and 20 kHz. Since the ratio of useful signal to noise varies in different areas of the spectrum (frequency bands), it is advantageous if the spectrum is divided into m intervals and a number NI,peak of dominant frequencies is determined individually in each interval I x with x = 1 to m. The number NI,peak can be the same for all intervals or different, in particular can be determined individually. This has the advantage that the overall spectrum can be divided into more or less interesting sub-ranges. The sub-ranges of interest can be those in which frequencies are known to be caused or generated by faults in the pump unit.For these interesting sub-ranges, a higher number of NI,peak dominating frequencies can be used than for the other sub-ranges.

[0036] For example, the total spectrum can be divided into 2 to 20 sub-ranges. The sub-ranges can have the same or different spectral widths. In the case of a frequency spectrum from 0 to 20 kHz, each sub-range or each interval with, for example, 10 intervals can have a spectral width (bandwidth) of 2 kHz. For each interval, the number NI,peak can be between 50 and 500, for example. With 500 selected spectral components, an interval with 2000 discrete spectral components (e.g., 2 kHz interval width) is already thinned by 75%. With 100 selected spectral components, the spectrum is even thinned by 95%, which significantly reduces the computing power for the subsequent evaluation.

[0037] The dominant spectral components can be determined by applying a sorting algorithm. The spectral components can be sorted according to their amplitude, with the N peak or N1,peak spectral components with the largest amplitudes being selected.

[0038] Preferably, the determination of the dominant spectral components is carried out in such a way that a) first, all spectral components are checked one after the other to determine which spectral component has the highest amplitude, b) then the remaining spectral components are checked one after the other to determine which of these spectral components has the highest amplitude, and c) step b) N peak -2 is repeated, where N peak is the number of dominant spectral components.

[0039] This procedure can be performed for the entire spectrum as a whole or, preferably, for each interval I x of the corresponding spectrum. In the latter case, steps a) to c) relate to the spectral components within an interval I x and are performed sequentially for each interval I x.

[0040] For example, in each interval I x , the discrete spectral lines of the spectrum are sorted according to descending or ascending amplitude. The dominant spectral components of an interval I x are those lines (peaks) with the largest amplitudes (NI,peak). While the lines of the dominant spectral components remain in their positions, all other spectral lines are removed from the spectrum. Thus, the spectrum is thinned. The resulting spectrum forms the derived spectrum.

[0041] Considered step by step, in the case of an interval-related determination of the dominant spectral components, the following steps can be carried out: a) for a first interval I i : a1) first, all spectral components of the first interval I i are checked one after the other to determine which spectral component has the highest amplitude, b1) then, the remaining spectral components of the first interval I 1 are checked one after the other to determine which of these spectral components has the highest amplitude, and c1) step b1) N i -2 times is repeated, where N i is the number of dominant spectral components in the first interval I i, b) for a next interval I i+1 : a2) first, all spectral components of the next interval I i+1 are checked one after the other to determine which spectral component has the highest amplitude, b2) then, the remaining spectral components of the next interval I i+1 are checked one after the other to determine which of these spectral components has the highest amplitude, and c2) step b2) N i+1 is repeated -2 ​​times,where N i+1 is the number of dominant spectral components in the next interval I i+1, and c) step b) is repeated m -2 times, where m is the number of intervals I x. Thus, the dominant spectral components are determined for all intervals I x successively.

[0042] As already mentioned, the number N i or NI,peak of dominant spectral components can be the same for all intervals or specified individually for all intervals and, if necessary, different from interval to interval. This means that more dominant spectral components can be determined in a first interval than in another interval. This makes it possible to take frequency band-specific characteristics of the airborne noise of the pump unit into account. For example, in those intervals in which certain mechanical faults in the pump unit are known to be evident in spectral components in the spectrum, a higher number of dominant spectral components can be determined than in other intervals in which this is not the case or in which noise predominates. In such intervals, the number of dominant spectral components can be selected accordingly smaller.

[0043] It should be mentioned again that the above steps a)-c) are carried out to determine the dominant spectral components for the spectra of all audio signals.

[0044] Preferably, the amplitude ratios are formed by calculating, for a spectral component, a ratio of the amplitude value of the spectral component in the spectrum of the second audio signal or in the spectrum derived therefrom to the amplitude value of the corresponding spectral component in the spectrum of the first audio signal or in the spectrum derived therefrom. The quotient of the aforementioned amplitude values ​​can also be used as the ratio here.

[0045] Thus, the higher amplitude value should always be in the numerator, meaning the ratio should be greater than 1. These ratios can also be referred to as peak factors.

[0046] In more detail, the amplitude ratios can be formed in one embodiment such that, for a spectral component fk, the ratio Fk of the amplitude value of this spectral component fk in the spectrum of the second audio signal or in the spectrum derived therefrom to the amplitude value of the same spectral component fk in the spectrum of the first audio signal or in the spectrum derived therefrom is calculated. This is then performed for each, or at least for each selected, spectral component fk.

[0047] If the ratio of the amplitude values ​​from the second to the first audio signal is calculated from a spectral component, it is advantageous to apply the aforementioned calculation rule only if both amplitude values ​​are greater than zero to avoid division by zero. The ratio can be set to zero if one of the two amplitude values ​​is zero. Alternatively, however, a data vector containing all ratios F k could be initialized to zero before calculating the ratios.

[0048] If only the N peak-dominating spectral components are used, the zero query is not absolutely necessary, since the amplitudes of these spectral components are always greater than zero. However, the numerical processing of the present method is simpler if the ratio calculation is performed sequentially for all N total spectral components fk, with k as the running index from 0 to N total -1.

[0049] From case to case, acoustic measurements taken at different times do not always result in exactly identical spectra, even if the source emits the same sound(s) invariably. This means that a dominant spectral component in the second measurement or second audio signal may not necessarily be located at the same spectral location—more precisely, at the same index k of the discrete spectrum—as in the first measurement or first audio signal. A spectral offset may therefore exist between the first and second audio recordings. Such a spectral offset is more likely the finer the spectral spacing, i.e., the resolution of the spectral components in the spectra of the audio signals, is chosen.At low resolution, there is a high probability that no spectral offset will be present, and the amplitude values ​​of the same spectral component fk in the spectrum of the first and second audio signals can be compared. At higher resolution, however, an offset is possible and must be taken into account.

[0050] It can therefore be provided according to a preferred embodiment that an offset correction is carried out when forming the amplitude ratios by a. for the spectral component fk the ratio F k of the amplitude value of the spectral component fk in the spectrum of the second audio signal or in the spectrum derived therefrom to the amplitude value of the same spectral component fk in the spectrum of the first audio signal or in the spectrum derived therefrom is calculated if both said amplitude values ​​are greater than zero, or b. for the spectral component fk the ratio F k of the amplitude value of the spectral component fk in the spectrum of the second audio signal or in the spectrum derived therefrom to the amplitude value of the previous spectral component f k-1 in the spectrum of the first audio signal or in the spectrum derived therefrom is calculated if both said amplitude values ​​are greater than zero, or c.for the spectral component fk, the ratio F k of the amplitude value of the spectral component fk in the spectrum of the second audio signal or in the spectrum derived therefrom to the amplitude value of the next spectral component f k+1 in the spectrum of the first audio signal or in the spectrum derived therefrom is calculated if both said amplitude values ​​are greater than zero.

[0051] The quotient of the mentioned amplitude values ​​can also be used as a ratio here.

[0052] In the aforementioned cases, the second audio signal, i.e., the "near spectrum," is used as the reference, since its amplitude values ​​are contained in the counter. Of course, the first audio recording can also be used as a reference.

[0053] In case a. there is no spectral offset between the peaks in the near and far spectrum. In case b. there is an upward spectral offset of the far spectrum for the spectral component fk, i.e. the spectral component in the far spectrum corresponding to a spectral component fk in the near spectrum is shifted to the next higher index value k+1 at the level of the spectral resolution Δf = f k+1 - fk of the spectra under consideration, so that in order to correctly calculate the amplitude ratio the index of the spectral component in the far spectrum must be reduced by 1. In other words, a peak in the spectral component fk of the second audio signal (near spectrum) corresponds to a peak in the spectral component f k-1 of the first audio signal (far spectrum). In case c. there is a spectral offset of the far spectrum downwards in the spectral component fk, ieThe spectral component of the far spectrum corresponding to the spectral component fk of the near spectrum is shifted to the next lower index value k-1 at the level of the spectral resolution Δf = f k+1 - fk of the spectra under consideration, so that for a correct formation of the amplitude ratio, the index of the spectral component in the far spectrum must be increased by 1. In other words, a peak in the spectral component fk of the second audio signal (near spectrum) corresponds to a peak in the spectral component f k+1 of the first audio signal (far spectrum).

[0054] If none of the three cases a., b. or c. is met for the spectral component fk, i.e. if one of the two amplitude values ​​to be related is zero, the ratio F k can be set to the value zero. Alternatively, a data vector comprising the ratios F k of all spectral components fk can be initialized with zero before the ratio is formed. Thus, the ratio F k is already zero if none of the three cases a., b. or c. is met for the spectral component fk. If, on the other hand, one of the three cases a., b. or c. is met for the spectral component fk, the value for the ratio F k pre-initialized with zero is replaced by the calculated value of the ratio F k.

[0055] The simultaneous occurrence of several cases a., b., and c. is avoided by the prior selection of the dominant spectral components, because this sets the amplitudes of most spectral components to zero. However, one or more rules can also be established to handle the simultaneous occurrence of several cases. For example, if several cases a., b., and c. occur, the largest ratio F k of the cases can be used.

[0056] Preferably, after the ratios have been formed, and possibly after the offset correction, a check is carried out to determine the spectral component for which the amplitude ratio determined for this purpose lies above the limit value. This check is preferably carried out for all spectral components, at least for the selected spectral components one after the other. Those spectral components for which this condition—i.e., that the amplitude ratio determined for this purpose lies above the limit value—is met, together with the amplitude values ​​assigned to these spectral components from the spectrum of the second audio signal, then form the reconstructed signal in the form of a reconstructed spectrum.

[0057] Ideally, the limit value is set taking into account a relationship between the two distances. Thus, the limit value can take into account the aforementioned distance law, which applies depending on the acoustic quantity being measured. Therefore, if sound pressure is measured as an acoustic quantity, the limit value can be set such that it reflects a proportional relationship between the first distance and the second distance. If sound power or sound energy is measured as an acoustic quantity, the limit value can be set such that it reflects a quadratic relationship between the first distance and the second distance.

[0058] If the first and second distances are specified for carrying out the method according to the invention or if the ratio between the first and second distances is specified, the limit value is also known and fixed and can be stored as a constant in the software application.

[0059] As already noted, for example, the second distance may be half the first distance, with these distances being used as default values. In this case, the limit value may be 2 when using sound pressure or 4 when using sound power or sound energy.

[0060] If, on the other hand, it is permitted that the first and / or second audio signal is recorded at a self-selected distance, or if it is to be checked whether a specified distance actually exists during the first and / or second recording, then the first and second distances must be measured. In this case, the limit value F threshold can be calculated from the measured values ​​determined, e.g. in the case of sound pressure from the ratio of the first to the second distance: F threshold = D far / D near . In this case, the limit value is therefore variable and is always re-determined from the determined distances D far and D near. In the case of sound intensity or sound energy, the ratio of the first to the second distance can be calculated as follows: F threshold = (D far / D near ) 2< .

[0061] Preferably, the limit value can be weighted with a factor between 0.6 and 1, preferably between 0.8 and 1. This takes measurement inaccuracies into account and prevents a spectral component belonging to the pump unit from being ignored due to these measurement inaccuracies.

[0062] To improve the accuracy of the method, it is advantageous if the first geometric location, the second geometric location, and the pump unit lie on a straight measuring line. Furthermore, it is useful if the measuring line is perpendicular to a connecting line between the pump unit and the secondary noise source, as the measuring line in this case is almost tangential to the sound field of the secondary noise source. However, it is optimal if the first geometric location and the second geometric location lie on a circular path around the secondary noise source.

[0063] It should be noted that the method proposed in this description has limitations. For optimal elimination of background noise in the recorded audio signal and for reliable and accurate determination of the pump unit's condition, it is necessary that the pump unit and the background noise source(s) are sufficiently far apart compared to the measurement distances D far and D near. If this is not the case, the elimination results and, consequently, the condition detection results will be impaired.

[0064] Furthermore, the operating state of the pump unit should be the same when recording the first and second audio signals and should not change during the recording in order for the process to achieve a high quality.

[0065] As already mentioned, a further improvement of the method according to the invention consists in carrying out at least one further measurement, i.e. detecting a third audio signal, specifically at a third distance from the pump unit which is greater than the second distance but preferably smaller than the first distance. The third distance therefore lies between the first and second distances. All of the method steps described above with regard to the first and second audio signals can be applied accordingly to the third and second audio signals. Their amplitude values ​​can also be related, so that here too an amplitude ratio of the third audio signal to the second audio signal can be calculated for each spectral component fk.This respective amplitude ratio can be used to verify an assumption resulting from the spectral comparison of the second audio signal with the first audio signal that a certain spectral component fk actually belongs to the pump unit 1 under investigation. If an amplitude ratio F k calculated from the first and second audio signals for a certain spectral component fk lies beyond the limit value F threshold , so that this spectral component is a potential candidate for a spectral component originating from the pump unit, a further amplitude ratio can be calculated from the third and second audio signals for the spectral component fk, which is then also compared with a limit value to verify the assumption. This limit value is then formed by the ratio of the third distance to the second distance, if necessary also weighted with a factor between 0.6 and 1.Only if this condition is also fulfilled can it be assumed that the spectral component fk is a spectral component originating from the pump unit.

[0066] The same procedure can be used with other measuring positions, which are optionally possible.

[0067] According to a particularly advantageous development of the invention, the distances are determined by an optical measurement, in which the pump unit is detected from the first and second geometric locations using an optical sensor. The optical sensor can, for example, be a laser distance measuring device that uses a laser beam and a time-of-flight measurement to determine the distance between the examination device and the pump unit.

[0068] Alternatively, the optical sensor can be a camera with which the pump unit is visually captured in at least one image, with this image being evaluated, preferably in real time. The disadvantages mentioned in the introduction to the description of the laborious process of finding an appropriate position for the handset or microphone based on a real-time evaluation of the acoustic signal in terms of signal quality and amplitude are overcome by the proposed optical distance measurement. It provides an objective indicator for determining the correct distance for audio recording. This enables the user to position the examination device at a specified distance from the pump unit and thus record the audio signals at the correct distance.

[0069] Furthermore, the user does not need to take multiple images to detect a fault in the pump unit. Correct distance adjustment also eliminates the need for additional measuring tools such as a folding rule or tape measure. Furthermore, the measurement can be performed with one hand, holding the test device. However, the test device can also be mounted on a tripod. This provides a simple and user-friendly method for distance measurement, particularly during an acoustic measurement of a pump unit.

[0070] The visual measurement can be performed in a known manner according to the state of the art. Methods for determining the spatial distance between a camera and an object detected by it are known per se. Reference is made to the relevant literature, for example, EP 2669707 A1, WO 2015 / 144775 A1, US 2011 / 0025845 A1, US 2021 / 0254962 A1, US 2019 / 0340799 A1, EP 326249 9B1, or US 10489033 B2.

[0071] According to one embodiment, the optical, particularly visual, measurement is performed simultaneously with the recording of the respective audio signal. This allows the examination device to know the distance it was at at the time the respective audio signal was recorded.

[0072] For example, the distance measurement can be performed before, after, or during a recording. It is also possible for the distance measurement to be performed multiple times, particularly continuously during the respective recording. It is advantageous if the distance is determined twice, for example, once at the start or before the recording and again at the end or after the recording. This has the advantage that the second measured distance can be used to check whether the distance has changed significantly during the recording compared to the first measured distance. The first measured distance can be used for the further process.

[0073] The recording duration can be between 5 seconds and 20 seconds, preferably 10 seconds.

[0074] Ideally, the determined distance is stored associated with the corresponding audio signal, in particular the corresponding audio recording, for example in the form of so-called metadata.

[0075] Once the distance has been determined, it can be displayed on the examination device's display, particularly while the respective recording is taking place.

[0076] According to an advantageous further development, a termination criterion can be applied, whereby the recording is terminated if a change in the distance of more than 3%, preferably more than 5%, occurs during the recording. In this case, the user can be asked to keep their hand steady by a corresponding message on the examination device.

[0077] During the visual distance measurement, the captured image can be displayed on the examination device's display. Reference points can also be entered and displayed in the image, for example, as color-coded circles overlaid on the pump unit. This can be referred to as "augmented reality measurement" (AR measurement). Furthermore, the distance determined from the captured image can be displayed on the examination device's display, for example, as a numerical value.

[0078] Preferably, the threshold value is calculated from the ratio of the two determined distances. This has the advantage of eliminating the time-consuming positioning of the examination device at a specific, predetermined distance from the pump unit. The first and second audio signals can thus each be recorded from essentially any distance. Nevertheless, distances can still be recommended as guidelines, such as 25 cm and 50 cm, 30 cm and 60 cm, 40 and 80 cm, etc.

[0079] The mobile examination device can consist of one, two, or more components. The software application is preferably loaded onto the examination device, in particular onto a central component with at least one processor and at least one data memory (RAM, EEPROM), or installed for its execution. In one embodiment, the acoustic sensor and / or the optical sensor can form an integral part of the examination device. Ideally, the examination device is a smartphone, a tablet computer, or a laptop, as these devices are often equipped with integrated cameras and microphones.

[0080] However, the optical sensor and / or the acoustic sensor can also be a separate part from the central component, more precisely with its own housing that is connected to the central component for data purposes, for example via a cable or wirelessly. In this case, the smartphone, tablet computer or laptop forms the central component. This has the advantage that a better microphone than the one integrated in the central component can be used, for example with a special directional characteristic, and / or that a better camera than the one integrated in the central component, for example with a higher resolution, or a different optical sensor can be used for distance measurement. If the optical and / or the acoustic sensor are each connected to the central component, they together form the examination device, which in this design variant is made up of three parts.

[0081] In a preferred embodiment, the examination device consists of a smartphone and a housing with a holder for holding the smartphone. At least one acoustic sensor for recording audio recordings is integrated into the housing. This acoustic sensor can be connected to the smartphone via an interface integrated into the holder. The interface can be a mechanical, electrical interface that is contacted when the smartphone is inserted into the holder, so that an electrical connection is simultaneously established between the smartphone and the acoustic sensor. The camera integrated into the smartphone can be used to capture the image.

[0082] In the same way, however, in addition to the acoustic sensor, the holder can have an optical sensor, in particular a camera, to take an image of the pump unit.

[0083] The holder may have a handle that allows it to be held in the hand in such a way that the person holding it can see the display of the smartphone.

[0084] Further features, advantages, and characteristics of the invention are explained in more detail below with reference to exemplary embodiments and the accompanying figures, which illustrate these examples. In the figures, identical reference numerals or symbols designate identical or at least functionally equivalent components, objects, or method steps.

[0085] Although a feature of the invention described above and below may be disclosed in connection with a specific embodiment, it should be clarified that this feature may also be present in another embodiment, provided that its transfer to this other embodiment is not excluded.

[0086] It should be noted that, in the context of this description, the terms "have," "comprise," or "include" in no way exclude the presence of other features. Furthermore, the use of the indefinite article for an object does not preclude its plural.

[0087] They show: Fig. 1 : a pump unit according to the state of the art Fig. 2 : perspective view of an examination device according to the invention from the front Fig. 3 : perspective view of an examination device according to the invention from behind Fig. 4 : idealized test setup with two sound sources and two measuring positions Fig. 5 : combined spectra of the first and second audio recording Fig. 6 to 12 : Flowcharts of the procedure Fig. 13 : Comparison of the reconstructed spectrum with the original spectrum of the pump unit.

[0088] Figure 1shows an exemplary pump assembly 1 according to the prior art, comprising a pump unit 2, an electric motor 3 driving the same, and pump electronics 4 for controlling and / or regulating the electric motor. The pump unit 2 is a multi-stage, normal-suction, vertical high-pressure centrifugal pump in a so-called inline design, i.e. the suction port 5 and the discharge port 6 of the pump housing 9 are in line. The electric motor 3 is a wet-running motor without a mechanical seal. The electric motor 3 and pump unit 2 have a through shaft on which several impellers are mounted in a rotationally fixed manner. The pump assembly 1, together with the pump housing 9, stands on a pump base plate 7 for foundation fastening.

[0089] Within a casing tube 8 of the pump housing 9, axially stacked stage casings are arranged, each containing one of the impellers to convey the fluid to the next stage casing. The last impeller, axially positioned at the top, conveys the fluid into an annular space between the stage casings and the casing tube 8, where it exits the pump housing 9 through the discharge nozzle 6. The casing tube 8 ensures reliable sealing.

[0090] The pump electronics 4 includes a frequency converter for continuously adjusting the speed of the electric motor 3 between a minimum and a maximum speed. The pump unit 1 is used to pump cold and hot water and other liquids without abrasive or fibrous substances. Main areas of application are water supply and pressure boosting systems, industrial circulation systems, process engineering, cooling water circuits, fire extinguishing systems, and washing and irrigation systems.

[0091] Figures 2 and 3show a mobile examination device 10 according to the invention from the front and rear, where "front" refers to the side of the examination device 10 that a user looks at during an audio recording, and "rear" is the side of the examination device 10 that is intended to face the pump unit 1 during an audio recording. The examination device 10 is made up of several parts. It comprises a housing 11 in which examination electronics are arranged, a handle 13 connected to the housing 11 for holding the examination device 10, a holder 14 for a smartphone 15, and a smartphone 15 arranged therein. The housing 11 is positively connected to the handle 13 via a dovetail joint.

[0092] The examination electronics comprise an acoustic sensor 12 in the form of a microphone, which is positioned in front of an opening on the rear side of the housing 11, facing away from the user, for recording airborne sound from an object that reaches the examination device 10 along an acoustic path 12a. The microphone can be a directional microphone. For preprocessing, e.g., amplification, filtering, and digitization, the examination electronics comprise a signal processing unit with at least one processor, for example, a digital signal processor (DSP), and at least one memory.Also part of the examination electronics is a communication interface to the smartphone 15, which consists of an electro-mechanical component that is at least partially integrated into the holder 14 and that is contacted by a corresponding counter-interface when the smartphone 15 is inserted into the holder 14, and software for sending and receiving data from the smartphone 15 in a protocol-based manner via the communication interface.

[0093] In addition, the examination electronics includes a first interface 18a in the form of a USB interface, which serves as the power supply, and a second interface 18b in the form of an RCA jack as the data interface for connecting another sensor, for example, a vibration sensor. The separation of power supply and data transmission has the advantage of reducing the risk of interference. However, in another embodiment, the power supply and data transmission can also be carried out via a single common interface, e.g., a USB interface.

[0094] The smartphone 15 has, in a known manner, a display 16, which, in the form of a touch-sensitive display, also forms a control element of the smartphone 15. A software application, commonly referred to as an "app" for short, is loaded into the memory of the smartphone 15. This software application is configured to execute a method for examining the condition of at least one component 2, 3, 4 of the pump unit 1, in particular with regard to a fault condition, by acoustically analyzing the airborne sound emitted by the pump unit 1. The software application thus constitutes software for analyzing acoustic signals. It comprises corresponding program instructions which, when executed by the examination device 10, more precisely on a processor of the smartphone 15, cause the examination device 10 to execute the method. The smartphone 15 controls the examination electronics, or rather the acoustic sensor.

[0095] The smartphone 15 further comprises a camera 17 which, when the smartphone 15 is held in the holder 14, is located at the rear, ie on the back of the smartphone 15 facing away from the user, so that it can capture images along an optical path 17a which is substantially parallel to the acoustic path 12a.

[0096] The method for investigating the pump unit condition by acoustic analysis of the airborne sound emitted by the pump unit 1 is explained below. Figures 6 to 12 show flow charts with the individual process steps.

[0097] The basic idea of ​​the investigation method according to the invention is to measure several simultaneously existing sound sources S main , S beside (see Figure 4) based on their spectra, whereby in this example the frequency spectrum is used and the spectral components are therefore frequencies. By varying the measurement position and a subsequent algorithmic data analysis, the measured acoustic signals are divided into combined spectra 30, 40 (see Figure 5 ), the main sound source S main is spectrally separated from the secondary sound source(s) S secondary, and the separated or reconstructed spectrum of the main sound source S main is output for any further processing.

[0098] In the present example, a combined spectrum is a frequency spectrum that includes frequency components of both the main sound source S main and the secondary sound source S secondary.

[0099] Figure 4shows an idealized test setup that illustrates the basic concept of the invention for suppressing background noise from a secondary noise source S secondary during the test of the pump unit 1, which here constitutes the main noise source S main . The secondary noise source S secondary can, for example, be another pump unit 1a that is structurally installed and operated next to the pump unit 1 under test.

[0100] Although in the present example only two sound sources S main , S adjacent and only two measuring positions P far , P near are used for the implementation of the method, the method can also be applied to several secondary sound sources and, if necessary, include more than two measurements at different locations. In the simplest case, however, the test setup consists of Figure 4shows, from two sound sources, namely the main sound source S main and the secondary sound source S next to as well as from two measuring positions, namely a first measuring position P far with a greater distance D far to the main sound source S main and a second measuring position P near with a shorter distance D near to the main sound source S main .

[0101] It would be optimal if the two measurement positions P far and P near lie on a measurement curve that corresponds to a circular path around the secondary sound source S adjacent. However, this is not always feasible in practice.

[0102] The main sound source S main and the two measuring positions D far , D near therefore lie on a measuring line 20 and in a horizontal measuring plane. This line 20 also runs orthogonally to a connecting line 21 between the two sound sources S main , S next , although in practice it can lie at a different angle α to the connecting line 21. The orthogonality between the measuring line 20 and the connecting line 21 as well as the use of two measuring positions P far , P near at the shortest possible distance to the main sound source S main (compared to the distance between the main sound source S main and the secondary sound source S next ), preferably smaller than the distance between the main sound source S main and the secondary sound source S next , ensure that one moves as tangentially as possible to the sound field of the secondary sound source S next. The secondary sound source S next can lie outside the horizontal measuring plane.

[0103] The examination method comprises recording a first audio signal in a first audio recording during operation of the pump unit 1 at a first geometric location P far with a first distance D far from the pump unit 1, and recording a second audio signal in a second audio recording during operation of the pump unit 1 at a second geometric location P near with a second distance D near to the pump unit 1 that is less than the first distance D far . The first audio recording thus contains audio data from a remote measurement, while the second audio recording contains audio data from a close measurement. The two locations P far and P near lie on a measurement line 20 with the pump unit 1 to be acoustically measured. The recordings are each made using the acoustic sensor 12, controlled by the software application on the smartphone 15.

[0104] Sensor 12 detects the sound pressure p of the airborne sound emanating from pump unit 1, which, in the example considered, is superimposed by the airborne sound of a neighboring sound source and is thus also detected. Sensor 12 delivers an analog electrical signal representing the sound pressure-time curve. This sensor signal is digitized by the examination electronics, amplified and / or filtered before or afterward if necessary, buffered, and then made available as an audio recording to the software application, which reads the audio recording accordingly for processing. This occurs for both the first and the second audio recording. An audio recording thus contains the digitized sound pressure-time curve.

[0105] The condition of pump unit 1 is determined by analyzing a frequency spectrum 50 of pump unit 1 that has been cleaned of the background noise from background noise source 1a and reconstructed by comparing amplitude values ​​of frequencies determined from the first and second audio recordings. This is explained below.

[0106] Figure 6 shows the rough procedure, starting with the already described recording of the first and second audio signals from pump unit 1, including the background noise of the background noise source(s) 1a S adjacent at the measurement positions P far , P near in step S1. This represents the starting point of the following analysis. This is followed by the following three steps, which form the core of the procedure for a spectral-based differentiation of several sound sources S main , S adjacent : In step S2: Calculation and analysis of the combined spectra for all measurement positions. This involves calculating the frequency spectrum 30, 40 of each of the audio signals, followed by post-processing. In step S3: Calculation of the spectral amplitude ratios. This involves evaluating the audio signals in the frequency domain by forming amplitude ratios for corresponding frequencies in spectra 30, 40 of the first and second audio signals. In step S4: Reconstruction of the spectrum of the main sound source S main . This involves reconstructing the spectrum 50 of pump unit 1, cleansed of background noise.

[0107] These three core steps S2, S3, S4 are then followed by the evaluation of the cleaned spectrum 50 of the pump unit 1 in step S5, which includes the actual determination of the state of the pump unit 1, for example the determination of the speed and / or the general detection of a fault state and, if applicable, the identification of the specific fault.

[0108] It should be noted that this evaluation as such is not part of the invention. Reference is made to the relevant specialist literature for further details.

[0109] Figure 7 illustrates the individual steps of step S1. The user first moves the examination device 10 to the first measurement position P far .

[0110] First, the distance D far from the pump unit 1 is visually determined in step S1.1. For this purpose, the camera 17 in the smartphone 15 is used and controlled by the software application. The camera 17 detects the pump unit 1 and records at least one image, optionally a video, which is evaluated in real time. This is done by identifying prominent reference points 19 on the surface of the pump unit 1 in the image, forming a reference plane from these, and determining the distance D far between the camera and the reference plane along a measuring line orthogonal thereto. The determined distance D far is stored associated with the first audio signal.

[0111] When determining the distance (measurement), the user must hold the examination device 10 as still as possible so that the distance remains constant during the measurement. Information about the quality of position maintenance can be shown on the display 16, for example graphic information. For example, a green circle around the pump unit 1 can indicate that the distance is being maintained. In contrast, a red circle around the pump unit on the display 16 can indicate that the position changed too greatly during the distance determination. The measurement must then be repeated. Thus, compliance with the measurement position can be visually monitored on the display 16, in particular the smartphone display, during the measurement.

[0112] Following the visual determination of the distance D far, the first audio signal is recorded in a first audio recording at the first measurement position P far in step S1.2. As already explained above, this includes the detection of the airborne sound of the pump unit 1, including the background noise of the secondary sound source(s) 1a, using the acoustic sensor 12, possibly also preprocessing such as amplification and / or filtering, as well as digitizing and storing the sensor signal as a sound pressure-time curve in order to provide it to the software application on the smartphone 15 as the first audio recording.

[0113] Although the visual distance determination in step S1.1 in Fig. 7While the distance determination precedes the recording in step S1.2, the visual distance determination can of course also be performed during or after the recording in step S1.2. It is even preferred that the distance determination be performed throughout the entire duration of the recording, in particular repeatedly, so that the currently determined distance can be shown on the display 16. This is because the distance determination requires less time than the recording.

[0114] The user now moves the examination device 10 to the second measurement position P near . This is followed by a visual determination of the distance D near to the pump unit 1 in step S1.3. This is performed analogously to step S1.1. For this purpose, the camera 17 in the smartphone 15 is also used and controlled by the software application, and the distance D near between the camera 17 and the reference plane 19 is determined analogously to step S1.1. The determined distance D near is saved and associated with the second audio recording.

[0115] Subsequently, the second audio signal is recorded in a second audio recording at the second measurement position P remote in step S1.4. This occurs analogously to step S1.2. This also includes the detection of the airborne sound of the pump unit 1, including background noise from the secondary sound source(s), using the acoustic sensor 12, optional preprocessing such as amplification and / or filtering, as well as digitizing and storing the sensor signal as a sound pressure-time curve in order to provide it to the software application on the smartphone 15 as a second audio recording.

[0116] It should also be noted that although the visual distance determination in step S1.3 in Fig. 7temporally precedes the recording in step S1.3, the visual distance determination can of course also be carried out during or after the recording from step S1.4. Here, too, it is preferred that the distance determination be carried out throughout the entire duration of the recording, in particular repeatedly, so that the currently determined distance can be displayed on the display 16.

[0117] Finally, it should be noted that although the recording of the second audio signal (close-up recording) at the measuring position closer to the pump unit 1 (P near ) in the example shown here takes place after the recording of the first audio signal (long-distance recording) at the measuring position further away from the pump unit 1 (P far ), this order can of course also be reversed. The visual distance measurement is very advantageous because it allows the software application to immediately know the distance at which the first and second audio signals were recorded, meaning that these recordings do not necessarily have to be made at a predetermined distance expected by the software application. Thus, steps S1.1. to S1.4 can be carried out in any order, for example: S1.1. → S1.2 → S1.3 → S1.4 S1.2. → S1.1 → S1.3 → S1.4 S1.1. → S1.2 → S1.4 → S1.3 S1.2. → S1.1 → S1.4 → S1.3 S1.3.→ S1.4 → S1.1 → S1.2 S1.4. → S1.3 → S1.1 → S1.2 S1.3. → S1.4 → S1.2 → S1.1 S1.4. → S1.3 → S1.2 → S1.1.

[0118] Furthermore, steps S1.1 and S1.2 and / or S1.3 and S1.4 can be performed simultaneously in one embodiment, as already mentioned. The only thing to consider is that an audio recording and a distance measurement always form a related pair of data and must therefore be performed concurrently.

[0119] The aforementioned storage of the first and second audio signals in one audio recording each can be carried out in a conventional audio data format, e.g. in the form of a WAV file (waveform format), preferably in a format in which the corresponding sound pressure-time curve is contained as uncompressed raw data.

[0120] The recording of the first and second audio signals, i.e., the digital sound pressure-time curves, is initiated by the software application. In this respect, the examination electronics can be considered a connected peripheral device to the software application, which provides the data required for analysis. Figures 6 and 7 An intermediate step (not shown) occurs after the recording of the first and second audio signals: their transmission from the examination electronics to the smartphone 15, where the software application processes and / or evaluates the audio recordings. From the perspective of the software application, this represents the reading of the sound pressure-time curve or sensor signal.

[0121] It should be noted that, as an alternative to the smartphone 15, the processing and / or evaluation of the audio signals can take place on a server, to which the smartphone 15 transmits the recorded audio signals, preferably including the determined distances D near and D far. Thus, at least part of the software application can run on a server (cloud).

[0122] After this transfer or reading, the sound pressure frequency spectra of the audio signals are calculated in step S2 using an FFT analysis (Fast Fourier Transformation) and their post-processing. The individual steps are shown Figure 8 in the form of two parallel paths S2a, S2b, which comprise the same steps in content, but are applied to the first audio signal on the one hand and to the second audio signal on the other. The parallel paths S2a, S2b can be executed sequentially or alternately step by step.

[0123] In the first path S2a, the sound pressure frequency spectrum 30 of the first audio signal is calculated in step S2.1a, and then in the second path S2b or following the first path S2a in step S2.1b, the sound pressure frequency spectrum 40 of the second audio signal is calculated. The two frequency spectra 30, 40 are in Figure 5 represented, more precisely, in a single diagram, with the amplitude values ​​of the spectrum 40 of the second audio signal being provided with a negative sign. It can be seen that the amplitude values ​​of the frequencies in the spectrum 30 of the first audio signal are lower in magnitude than the amplitude values ​​of the frequencies in the spectrum 40 of the second audio signal, because the second audio signal (in Figure 5below) was recorded closer to pump unit 1 than the first audio signal and the airborne sound at the second measuring location is correspondingly stronger than at the first measuring location, or the noises near the pump unit are louder than further away from it.

[0124] The subsequent post-processing involves eliminating noise from the signal and simultaneously thinning the frequency spectrum, simplifying subsequent signal processing and requiring less computing power. This post-processing involves selecting dominant frequencies within the sound pressure-frequency spectra. Dominant frequencies are those frequencies with a comparatively high amplitude value. In other words, the most relevant "peaks" in the spectrum are identified and used for further analysis. In the embodiment presented here, the selection is performed interval-related, i.e., for each interval of a number of frequency intervals (frequency bands) into which the frequency spectrum is divided. The post-processing / frequency selection takes place in steps S2.2a-S2.4a and S2.2b-S2.4b, respectively.

[0125] In a first step S2.2a in the first path S2a, the first frequency spectrum 30, i.e. the spectrum 30 of the first audio recording, is divided into m intervals. For this purpose, the frequency spectrum 30 is viewed along the frequency axis as a total interval I total, which ranges from a minimum frequency f min to a maximum frequency f max, I total = [f min , f max ]. Alternatively, the total interval I total can be cut out of the total frequency spectrum 30, so that the minimum frequency f min and the maximum frequency f max are defined differently from the actual frequency range in the frequency spectrum 30. For example, the minimum frequency f min = 20 Hz and the maximum frequency f max = 20 kHz. It is expedient for all intervals to have the same spectral width to simplify signal processing. However, they can also have different spectral widths.

[0126] The same procedure is followed in the first step S2.2b in the second path S2b, in which the division of the second frequency spectrum 40, ie the spectrum 40 of the second audio recording, into m intervals takes place.

[0127] After the first step S2.2a, frequency selection takes place in a second step S2.3a, which, more precisely, is initially an identification. For this purpose, for each interval I x (with x from 1 to m), a number NI,peak of those frequencies fk in the interval I x that have the largest amplitude values ​​are identified. The number NI,peak can be the same for all intervals I x or different. The latter offers the possibility of treating the intervals differently. For example, fewer frequencies fk can be selected where noise is more prevalent, and more frequencies fk can be selected where a useful signal is more likely to be found. For example, between 50 and 500, preferably 100, dominant frequencies fk can be selected per interval.This can be done by sorting all frequencies fk of an interval I x , each identified by an index value k, according to their amplitude (ascending or descending). In descending sorting order, the first NI,peak frequencies are selected, while in ascending sorting order, the last NI,peak frequencies fk are selected. This procedure is performed for each interval I x , i.e., repeated iteratively from x = 1 to m.

[0128] The same procedure is followed in the second step S2.3b in the second path S2b, in which for each interval I x (with x from 1 to m) a number NI,peak of those frequencies fk in the interval I x which have the largest amplitude values ​​is identified.

[0129] After identifying the dominant frequencies fk, they are selected in the third step S2.4a in the first path S2a by setting the amplitudes of all other, i.e., non-dominant frequencies fk to zero, so that essentially only the dominant frequencies fk form the spectrum of the first audio signal, which, due to this change, can be considered a derived (modified) frequency spectrum of the first audio signal. This automatically selects the dominant frequencies fk. Alternatively, the identified dominant frequencies can be written into a new data vector, which essentially describes an empty spectrum.

[0130] The same procedure is followed in the third step S2.4b in the second path S2b, in which the amplitudes of all other, ie unidentified frequencies fk are set to zero, so that virtually only the dominant frequencies fk form the spectrum of the second audio signal, which, due to this change, is to be regarded as a derived (modified) frequency spectrum of the second audio signal.

[0131] The described post-processing in the form of frequency selection eliminates noise in the sound pressure-frequency spectra and significantly reduces the amount of data, so that further processing of the derived (modified) frequency spectra requires less computing power. For example, if the frequency range from 1 Hz to 20,000 Hz is divided into 10 equal intervals of 2,000 frequencies each, selecting the 100 dominant frequencies results in a 95% thinning of the frequency spectrum, since the amplitudes of 1,900 frequencies are set to zero.

[0132] An implementation of step S3 is in Figure 9shown. It is intended that for each frequency fk, the ratio F k of the amplitude A k, near the frequency fk in the derived frequency spectrum of the second audio signal (near spectrum) to the amplitude A k, far of the corresponding frequency fk in the derived frequency spectrum of the first audio signal (far spectrum) is formed. The ratios F k can be referred to as peak factors and are calculated according to F k = A k ,nah / A k ,fern

[0133] The calculation is performed iteratively for all N total frequencies fk in succession, where k is the running index of the discrete frequency axis of the spectra and takes values ​​from k = 0 to N total.

[0134] To avoid division by zero, before calculating each ratio F k , a condition can be checked to determine whether the amplitude A k,far from the frequency f k in the derived frequency spectrum of the first audio signal (far spectrum) is greater than zero. The amplitude ratio F k is set to zero if this is not the case, otherwise it is calculated as above. Alternatively, the condition can check whether the amplitude A k,far from the frequency f k in the derived frequency spectrum of the first audio signal (far spectrum) and the amplitude A k,near from the frequency f k in the derived frequency spectrum of the second audio signal (near spectrum) are greater than zero. The amplitude ratio F k is then calculated from F k = A k,near / A k,far if this is the case, otherwise it is set to zero.

[0135] When calculating the amplitude ratios F k , a possible offset error can be taken into account, which is characterized by the fact that corresponding frequencies in the spectra 30, 40 of the first and second audio signals do not have the same index value k. Consequently, corresponding frequencies are not located at the same position k in the derived frequency spectra either. Figure 10 shows the process flow for an offset correction implemented in step S3, which alternatively to step S3 in Fig. 9 can be executed.

[0136] The offset correction builds on and extends the last-mentioned condition. Thus, for the frequency fk, the ratio Fk of the amplitude value Ak,near of the frequency fk in the derived frequency spectrum of the second audio signal to the amplitude value Ak,far of the same frequency fk or the frequency with the same index k in the derived frequency spectrum of the first audio signal is calculated only if both of the amplitude values ​​Ak,near and Ak,far are greater than zero (first condition). Then, as above, Fk = Ak,near / Ak,far applies, whereby this case concerns an exact spectral match.

[0137] If this first condition is not fulfilled, ie the amplitude value A k,far of the far spectrum is zero, the two immediate neighboring frequencies f k-1 and f k+1 of the frequency fk in the derived frequency spectrum of the first audio recording, ie in the far spectrum, are checked one after the other.

[0138] Thus, for the frequency fk, the ratio F k of the amplitude value A k,near of the frequency fk in the derived frequency spectrum of the second audio recording to the amplitude value A k-1,far of the frequency f k-1 preceding fk in the derived frequency spectrum of the first audio recording is calculated if these two amplitude values ​​A k,near , A k-1,far are greater than zero (second condition). This results in a spectral offset Δ f of the far spectrum upwards, and then F k = A k,near / A k-1,far . However, this calculation is only performed for k > 0, so that A k-1,far can be calculated.

[0139] If this second condition is not met, a check is carried out to determine whether the amplitude value A k,near of the frequency f k in the derived frequency spectrum of the second audio recording and the amplitude value A k+1,far of the frequency f k+1 closest to f k in the derived frequency spectrum of the first audio recording are both greater than zero (third condition). In this case, the ratio F k = A k,near / A k+1,far is calculated for the frequency f k. However, this calculation is only performed for k < N total, so that A k+1,far can be calculated.

[0140] If this third condition is also not fulfilled, the ratio F k is set to the value zero for the frequency f k , because in this case either the amplitude value A k,near the frequency f k in the derived frequency spectrum of the second audio signal is equal to zero or the amplitudes of all three frequencies f k , f k-1 , f k+1 in the derived frequency spectrum of the second audio signal are equal to zero.

[0141] Overall, the offset correction, excluding the edge frequencies fk for k = 0 and k = N total, is mathematically represented as follows, where the following applies to the amplitude values ​​A k at the discrete frequencies fk: A k fern = A k fern f k mit f k = f min + k ⋅ Δ f und k = 0 , … , N total A k nah = A k nah f k mit f k = f min + k ⋅ Δ f und k = 0 , … , N total where Δ f the resolution of the frequency axis. a) Exact spectral match: A k fern > 0 ∧ A k nah > 0 ⇒ F k = A k nah / A k fern b) Slight spectral offset Δ f of the long-range spectrum upwards A k − 1 fern > 0 ∧ A k nah > 0 ∧ k > 0 ⇒ F k = A k nah / A k − 1 fern c) Slight spectral offset Δ f of the long-range spectrum downwards Ak+1far>0∧Aknear>0∧k <NI,total⇒Fk=Aknah / Ak+1fern

[0142] If none of these three cases occurs, the amplitude ratio F k is set to zero.

[0143] The simultaneous occurrence of several of the above-mentioned cases a), b), and c) is unlikely due to the selection of the dominant frequencies in the frequency spectra of the first and second audio signals. However, rules can also be defined for a case preference in the case of the simultaneous occurrence of several cases. One such rule is exemplified in Figure 11 which shows the individual steps of an alternative implementation of step S3 in a flowchart.

[0144] Compared to the implementation of step S3 according to Figure 10 the three conditions mentioned are not used alternatively as in Figure 10 , but are checked cumulatively, whereby both in the case of "yes" and in the case of "no" of a condition the next condition is checked, whereby the order is basically irrelevant.

[0145] To enable the test to determine whether two or even all three cases a), b), and c) are met simultaneously, not only a single amplitude ratio F k is calculated, but also a separate ratio F k1, F k2, F k3 for each case a), b), and c). By testing which of these ratios F k1, F k2, F k3 is greater than zero, it is possible to determine which cases were present cumulatively. This allows a priority system to be applied that determines which ratio F k1, F k2, F k3 is used in this case as the desired ratio F k. Figure 11 the priority regulation is such that of the ratios F k1, F k2, F k3 the one which is largest is used as the sought ratio F k.

[0146] While the design variants in the Figures 10 and 11 only a spectral offset of a frequency resolution of 1 · Δ fie a frequency shift by a single index value + / - 1 can be considered, in alternatives not shown, a spectral shift by 2 · Δ f , ie two index values ​​+ / - 2, and if necessary by 3 · Δ f , ie three index values ​​+ / - 3 are examined, so that more cases are added. The higher the resolution Δ f is chosen, the higher the probability that a particular peak in the far spectrum is shifted by one, two or even three indices k.

[0147] If the ratios F k of the amplitudes of all corresponding frequencies f k in the first and second audio signals are determined in step S3, the reconstruction of the background noise-free spectrum 50 of the main sound source S main , i.e. the pump unit 1, follows in step S4. This is done by comparing the amplitude ratios F k with a threshold value F threshold , which is calculated from the distances D near , D far determined in steps S1.2 and S1.4, taking into account a distance law. The individual steps of the reconstruction method of step S4 are shown in Figure 12 .

[0148] In step S4.1, the limit value F threshold is first determined from the determined distances according to the calculation rule F schwelle = δ ⋅ D fern / D nah .

[0149] The limit value F threshold is therefore determined from the ratio of the distances D far , D near between the respective measuring position P far , P near and the main sound source S main , more precisely from the ratio of the first distance D far of the long-distance measurement to the second distance D near of the close-up measurement.

[0150] The calculation rule reflects the distance law p - 1 / r, according to which, for sound pressure values ​​p, as recorded or has been recorded by the acoustic sensor 12, the amplitudes change inversely proportional to the distance r of the sensor 12 from the sound source. This circumstance is now specifically used to separate the main sound source S main from the other sound source S secondary.

[0151] The ratio D far to D near is here increased by a factor δ weighted between 0.8 and 1 in order to achieve a certain tolerance in assessing whether the distance law is fulfilled for a particular frequency.

[0152] In step S4.2, those frequencies fk whose amplitude ratios Fk exceed the limit value Fthreshold are then found and selected. For this purpose, a check is carried out for each frequency fk (with k=0 to Ntotal) to determine whether the amplitude ratio Fk determined for this frequency fk and thus assigned to it exceeds the limit value Fthreshold. If this is the case, this frequency fk is part of the frequency spectrum of the main sound source Smain and is selected for the next step S4.3. If, on the other hand, the frequency ratio Fk is below the limit value Fthreshold, this frequency fk is part of the frequency spectrum of a secondary sound source Ssecondary and is ignored.

[0153] Finally, in step S4.3, the spectrum of the main sound source S main is extracted from the combined spectrum of the second audio signal. In other words, a reconstruction of the spectrum 50 of the pump unit 1 is performed by selecting the detected frequencies fk and combining them, together with the associated amplitude values ​​of the spectrum 40 of the second audio signal, to form the reconstructed spectrum 50. Alternatively, the spectrum 30 of the first audio signal could also be used here; however, the spectrum 40 of the second audio signal has the advantage that the amplitudes there are larger.

[0154] The reconstruction can be conveniently performed by "copying" the selected frequencies from the combined spectrum 40 for the near measurement position P near and "writing" them into a still empty spectrum, thus creating the reconstructed spectrum 50. Alternatively, the opposite procedure can be used: in the frequency spectrum 40 of the second audio signal, the amplitudes of those frequencies fk for which the associated amplitude ratio F k lies below the threshold value F threshold are set to zero. The reconstructed spectrum 50 corresponds to an approximation of the actual spectrum of the main sound source S main .

[0155] Figure 13shows a comparison of the reconstructed spectrum 50 of the main sound source S main with the original spectrum 60 of the main sound source S main or of the pump unit 1 in a common frequency diagram, wherein the amplitude values ​​of the original spectrum 60 are given a negative sign for this purpose, so that the original spectrum 60 is shown in the lower half of the diagram and the reconstructed spectrum 50 in the upper half of the diagram.

[0156] It should be noted that the foregoing description is given merely by way of example for illustrative purposes and in no way limits the scope of the invention. Features of the invention referred to as "may," "exemplary," "preferred," "optional," "ideal," "advantageous," "optionally," "suitable," or the like are to be considered purely optional and do not limit the scope of protection, which is determined exclusively by the claims. The invention is defined by the scope of the claims.

[0157] Although the above description of the invention mentions a multitude of physical, non-physical, or method-related features in relation to one or more specific embodiments, these features can also be used in isolation from the specific embodiment, at least as long as they do not require the mandatory presence of further features. Conversely, these features mentioned in relation to one or more specific embodiments can be combined with each other as desired, as well as with other disclosed or undisclosed features of shown or not shown embodiments, at least as long as the features do not exclude each other or lead to technical incompatibilities.

Claims

1. Method for analysing the state of at least one component (2, 3, 4) of a pump assembly (1), notably to determine the rotational speed of the pump assembly or with regard to an error through analysis of the airborne sound emitted by the pump assembly (1), comprising the recording of at least a first audio signal during operation of the pump assembly (1) at a first geometric location (Pfar) and at a first distance (Dfar) from the pump assembly (1), and the recording of a second audio signal during operation of the pump assembly (1) at a second geometric location (Pnear) and at a second distance (Dnear) from the pump assembly (1), which is less than the first distance (Dfar), characterised by the state of the pump assembly (1) being determined through the analysis of a signal (50) of the pump assembly (1) corrected for the ambient noise of at least one secondary sound source (1a), which is reconstructed through a comparison of amplitude values of the first and second audio signals.

2. Method according to claim 1, characterised by the amplitude values being spectral components determined from the first and second audio signals.

3. Method according to claim 1 or 2, characterised by determining respectively one spectrum (30, 40) of the first and second audio signals and forming a ratio of amplitude values of corresponding spectral components (fk, fk-1, fk+1) of the two spectra (30, 40) or spectra derived from them (Fk), and by forming the reconstructed signal (50) from the amplitude values of those spectral components (fk) of one of the two spectra (30, 40) or spectra derived from them for which the ratio (Fk) lies beyond a certain established limit value (Fthreshold), in order to eliminate the ambient noise in the first and / or second audio signal.

4. Method according to claim 3, characterised by the spectra (30, 40) being frequency spectra and the spectral components being frequencies.

5. Method according to claim 3 or 4, characterised by the reconstructed signal (50) being formed from the amplitude values of the spectral components (fk) of the spectrum (40) of the second audio signal or a spectrum derived from the same.

6. Method according to claims 3 through 5, characterised by a number of dominant spectral components (fk) first being determined from each of the two spectra (30, 40) and by these being selected, in which these dominant spectral components (fk) with their amplitude values form the respective derived spectrum, and by a ratio of the amplitude values subsequently being formed.

7. Method according to one of the claims 3 through 6, characterised by the two spectra (30, 40) being divided into intervals and a number of dominant spectral components (fk) being determined for each interval and these being selected.

8. Method according to claim 6 or 7, characterised by the number being specified.

9. Method according to claim 6, 7 or 8, characterised by the dominant spectral components being determined through the application of a sorting algorithm, notably in such a manner that a) all spectral components (fk) are initially checked sequentially to determine which spectral component (fk) has the highest amplitude, b) subsequently, the remaining spectral components (fk) are checked sequentially to determine which of these spectral components (fk) has the highest amplitude, and c) step b) is repeated Npeak -2 times, where Npeak is the number of dominant spectral components (fk).

10. Method according to one of the preceding claims 3 through 9, characterised by the amplitude ratios being formed in such a manner that, for one spectral component (fk), a ratio (Fk) of the amplitude value of the spectral component (fk) in the spectrum (40) of the second audio signal or in the spectrum derived from that to the amplitude value of the corresponding spectral component (fk,fk-1, fk+1) in the spectrum (30) of the first audio signal or in the spectrum derived from that is calculated.

11. Method according to claim 10, characterised by the ratio Fk of the amplitude value of the spectral component fk in the spectrum (40) of the second audio signal or in the spectrum derived from that to the amplitude value of the same spectral component fk in the spectrum (30) of the first audio signal or in the spectrum derived from that being calculated for a spectral component fk, with this taking place in particular only when both aforementioned amplitude values are greater than zero.

12. Method according to claim 10 or 11, characterised by an offset correction being performed in forming the amplitude ratios, by a) for the spectral component fk, calculating the ratio Fk of the amplitude value of the spectral component fk in the spectrum (40) of the second audio signal or in the spectrum derived from that to the amplitude value of the same spectral component fk in the spectrum (30) of the first audio signal or in the spectrum derived from that, when both aforementioned amplitude values are greater than zero, and / or b) for the spectral component fk, calculating the ratio Fk of the amplitude value of the spectral component fk in the spectrum (40) of the second audio signal or in the spectrum derived from that to the amplitude value of the previous spectral component fk-1 in the spectrum (30) of the first audio signal or in the spectrum derived from that, when both aforementioned amplitude values are greater than zero, and / or c) for the spectral component fk, calculating the ratio Fk of the amplitude value of the spectral component fk in the spectrum (40) of the second audio signal or in the spectrum derived from that to the amplitude value of the next spectral component fk+1 in the spectrum (30) of the first audio signal or in the spectrum derived from that, when both aforementioned amplitude values are greater than zero.

13. Method according to claim 10, 11 or 12, characterised by sequentially checking, for all spectral components (fk), for which spectral component (fk) a determined amplitude ratio exceeds the limit value (Fthreshold) and by those spectral components (fk) where this is the case, along with the amplitude values assigned to these spectral components (fk) from the spectrum (40) of the second audio signal, forming the reconstructed signal (50).

14. Method according to one of the claims 3 through 13, characterised by the limit value (Fthreshold) being established, notably calculated, under consideration of a ratio of the two distances (Dfar, Dnear) to each other.

15. Method according to one of the preceding claims, characterised by the first geometric location (Pfar), the second geometric location (Pnear) and the pump assembly (10) lying on a straight measuring line (20) or the first geometric location (Pfar) and the second geometric location (Pnear) lying on a circular path around the secondary sound source (1a).

16. Method according to one of the preceding claims, characterised by the sound pressure level being measured using an acoustic sensor (12) and the first and second audio signal respectively forming the sound pressure level / time gradient of the sensor (12).

17. Method according to one of the preceding claims, characterised by the second distance (Dnear) being half the first distance (Dfar).

18. Method according to one of the preceding claims, characterised by the distances (Dfar, Dnear) respectively being determined by an optical measurement, in which the pump assembly (1) is recorded using an optical sensor (17), respectively from the first and second geometric locations (Pfar, Pnear), notably recorded visually in at least one image using a camera, and this image is respectively analysed, preferably in real time.

19. Software application for a mobile analysis device (10) with a display (16), at least one control element and an acoustic sensor (12) and optionally an optical sensor (17), comprising program instructions that, when they are executed by the analysis device (10), cause the analysis device (10) to execute the method according to one of the claims 1 through 18.

20. Machine-readable storage medium with a software application according to claim 19.

21. Mobile analysis device (10) with a display (16), at least one control element and an acoustic sensor (12) to record audio and optionally an optical sensor (17), characterised by the analysis device (10) being configured to execute the method according to one of the claims 1 through 18.