WO2012167819A1

WO2012167819A1 - Method for ascertaining a remaining service life of a machine component, and computer program for carrying out the method

Info

Publication number: WO2012167819A1
Application number: PCT/EP2011/059415
Authority: WO
Inventors: Hubert Bellm; Henning Lenz
Original assignee: Siemens Aktiengesellschaft
Priority date: 2011-06-07
Filing date: 2011-06-07
Publication date: 2012-12-13

Abstract

The invention relates to a method (10) for ascertaining a remaining service life (12) of a machine component (17). A reference characteristic curve (48, 50, 56) of at least one characteristic from at least one wear test of a reference component that matches the machine component (17) is retrieved from a database (14). The reference characteristic curve (48, 50, 56) is approximated using a mathematical function. At least one current characteristic (20) is retrieved from a sensor (34, 36, 38, 40) for the machine component (17). A time value that corresponds to the current characteristic (20) is ascertained from the approximated characteristic curve (52, 54, 58), and the remaining service life (12) is the difference between an expected total service life of the machine component (17) and the ascertained time value.

Description

description

Method for determining a remaining service life of a machine component and computer program for carrying out the method

The invention relates to a method for determining a residual life of a machine component according to the Oberbe ^¬ handle of claim 1. It further relates to a computer program for performing the method.

Of machine components such as roller chains, high Anforde ^¬ stakes are, particularly in the application field of packaging machinery, provided. These are firstly the very high clock rates of machines per minute lie ^¬ gen can at several hundred cycles and thus a high speed components he call ^¬. On the other hand, due to hygiene and quality requirements, some maintenance and lubrication restrictions result, which result in increased running noise and wear. The component service life can therefore be significantly below the manufacturer's specification. Since the exchange of a compo nent ^¬ causes an interruption in production of several hours, there is the part of the machine manufacturer and users of the machine high interest the optimum time for the exchange to predict.

For estimating a residual life of the machine components, generally data obtained from the drive or supplied by external sensors is used.

"; Component ^¬ monitoring and remaining lifetime determination using microsystems technology exemplified by conveyor chains, Wissensportal construction machinery ne.de 4 (2005) Nienhaus, K. Pierburg, L." Discloses a ausgestatte- with micromechanical sensors th chain link, with for example, in whose help is continuously online, ie from current sensor data, a load integral of a conveyor chain is determined. This then becomes a residual life estimate on the basis of the so-called Wöhler characteristic derived. Similarly, DE60210747T2 describes the equipment of lift chains of a forklift truck with special chain links provided with sensors. From the data of these sensors, a remaining lifetime estimate is derived as before via a load integral. US200880004839A1 describes a device which uses non-destructive material ^tests on the basis of a database to determine the remaining service life of construction elements and notify the user via voice output. The method described in US20030158803A1 for residual life estimation, there as

Assets of designated components is based on historical, determined by sensors data from which a degree of wear is calculated. This in turn serves to derive a residual life of the assets.

The invention consists in specifying a further method that can be used to determine the remaining service life of machine components. This object is achieved with the features of claim 1. An ^¬ . Therein is provided a method of determining a remaining service life of a machine component, wherein a reference characteristic value curve a wear test of a matching to the machine component reference component is retrieved from a database of at least one characteristic value of Minim ^¬ least, with an approximation of the reference characteristic curve is carried out by a mathematical function, for the machine component at least one current characteristic value with a corresponding to the current parameter time ^¬ value determined from the approximated characteristic value curve and wherein there is the remaining life as the difference between an expected total lifetime of the machine component and the value detected is retrieved from a sensor. The advantage of the invention is that based on

Data of a reference component that is already worn out, ie historical data, and real-time data that better predicts the remaining life of a machine component so that an exchange of the machine component can be coordinated with the production planning.

Advantageous embodiments of the invention are the subject of the dependent claims. This used backlinks point to the further development of the subject matter of the main claim by the features of the respective subclaim; they are not to be regarded as dispensing with obtaining an inde ^¬ ended, objective protection for the feature combinations of the referenced dependent claims. Furthermore, with a view to an interpretation of the claims in a closer specification of a feature in a subordinate claim, it is to be assumed that such a restriction does not exist in the respective preceding claims.

If at least two wear tests historical characteristic ^¬ values are retrieved from the data base, is less than ^¬ different union total lifetimes of the respective reference components, the time axis of the characteristic curves on before Approximati- with a measure of a load which exposed the respective Refe ^¬ rence component during its operation was ^¬ prominent ska. In this way, different loads on the machine components from the wear tests, such as a clock rate or in chains a pressure on the clamping cylinder, are taken into account.

The or each characteristic curve is preferably approximated with a quad ^¬ ratischen polynomial. As a result, relatively strongly scattering characteristic values can be smoothed so that a trend of the characteristics curve can be better recognized.

When retrieving data from at least two wear tests, the respective coefficients of the quadratic polynomials may preferably be averaged for each characteristic. The characteristic curves may be able to vary significantly, so that greater differences can be smoothed by averaging the coefficients due to different ready for operation ^¬ conditions. The characteristic values may be preferred weighted according to relevance before the averaging. Here, for example, characteristics, let their courses see no MACHINES SHOW ^¬ chen trend, are weighted lower. Preferably, a residual service life determined for at least a first and a second current characteristic value is determined. In various observed characteristics of the reference component, for example a first one of a torque and a second calculated from a vibration signal characteristic value, ER give, different (approximate) characteristic value curves and therefrom also have different remaining lifetimes for Ma ^¬ schin component when separately for each characteristic value ermit ^¬ telt. By forming a mean value of these residual life take ^¬ the reliability of Restlebensdauerer- increased mediation.

With an availability of several different reference characteristic curves, that is to say with a corresponding sensory equipment of a machine which comprises the machine component, it is possible to select the characteristic values which can be used to determine the remaining service life. Specifically, such characteristic values can be considered which, in the majority of wear tests, indicate a trend in the characteristic curve or in the analytically adjusted characteristic curve. In this way, characteristic values which are more sensitive to certain operating conditions and thus could lead to a falsified result of the method can be removed from the method during the corresponding wear tests. Preferably, a reference characteristic curve can, with characteristic values for a torque as a sole reference parameter course USAGE ^¬ be det. This has the advantage that, under correspondingly favorable operating conditions of the machine component, more complex evaluations of a multiplicity of characteristic values can be dispensed with.

The described method with its embodiments is preferably implemented in software. Accordingly, the concerns The invention also relates to a computer program with computer-executable program code instructions for implementing the method as described here and below, when the computer program is executed on a computer, eg an automation device with a processor-type processing unit or the like. An automation sierungsgerät ^¬, especially a Industrieautomatisierungsge ^¬ advises on which such a computer program is loaded, is an example of a computer system that He-making at which also refers. Instead of the Automatisierungsge ^¬ räts standard computers, as they are common in office automation today, come into consideration. The Computerpro ^¬ gram for implementing the method is usually on or in a computer program product, in particular SpeI chermedium, ie, for example, on a magnetic or optical disk or in a semiconductor memory, so z. B.

volatile or non-volatile memory modules, held, so that the invention in this respect also relates to a computer program product with a computer-executable computer program for implementing the method

An embodiment of the invention will be explained in more detail with reference to the drawing. Corresponding objects or elements are provided in all figures with the same reference numerals.

The or each embodiment is not to be understood as limiting the invention. Rather, numerous modifications and variations are possible within the scope of the present disclosure, in particular those variants and combinations, for example, by combination or modification of individual features described in the general or specific description part as well as in the claims and / or the drawing Elements or process steps for the expert with regard to the solution of the problem can be removed and lead by combinable features to a new subject or to new process steps or process steps. Show it

1 shows a schematic representation of an inventive

A method for determining a residual life of a machine component;

2 shows a chain test bench and an attached

Computer for carrying out a method according to the invention;

3 is a schematic representation of a first Auswerteal ^¬ gorithmus the method;

4 shows two vibration waveforms of a machine component of a wear test and a ^¬ analytical specific adaptation of the vibration waveforms;

5 shows a torque curve of a machine component from a course of wear and an analytical adaptation of the torque curve;

6 shows a graphical representation of the determined Restle ^¬ lifetimes for five different signals; 7 shows a schematic representation of a second algorithm of the method; and a graphical representation of the determined residual life, averaged out of five remaining lifetimes as shown in FIG 6 and determined solely on the basis of the torque as characteristic value.

1 shows a schematic representation of a method 10 for determining a remaining service life 12 of a machine component according to an exemplary embodiment of the invention. In the method 10, from a database 14, historical data from at least one wear test with a reference component matching the machine component are converted into a first Evaluation algorithm 16 loaded. The data base 14 comprises ^¬ be vorzugt data from wear tests with a plurality of reference components. In the exemplary embodiment explained here, the considered machine component is a chain 17, eg a roller chain. Chains are widely used in packaging machines such as hygiene article packaging machines as transport elements, as they represent a robust and economical machine component with their space-saving design, easy installation and low maintenance costs. In addition, these have an exact positioning and high rigidity, which, for example, in favor of a parallel operation. Despite these fundamentally good properties, chains in packaging machines are subject to high wear, so that a diagnosis and estimation of the remaining service life makes sense in order to avoid unexpected failures. Ket ^¬ shoots nowadays take on a variety of tasks in production machines. The choice of the chain used is adapted to the particular field of application, so that different types of chains have developed. The most popular type of chain is the roller chain. The advantages of chain drives include slip-free power transmission, constant transmission and insensitivity to external influences such as temperature, humidity and dirt. In contrast to other types of transmissions, chain drives make it possible to easily bridge larger center distances. Disadvantages of chains are in particular the Polygonef ^¬ effect, which causes a fluctuating chain speed, and the susceptibility to vibration and maintenance. In the exemplary embodiment explained here, it is assumed that historical data from wear tests are available as reference components for three different chains. These chains have reached a final wear condition, ie the chains can no longer be operated. The data of these three reference chains are once evaluated offline in the method 10 for determining the remaining life of a chain in the first evaluation algorithm 16 for each machine. The method 10 furthermore comprises a second evaluation algorithm. in which the data from the first evaluation algorithm 16 and characteristic values 20 determined from current sensor data are included online. The first evaluation algorithm 16 will be described in more detail with reference to FIG. 3, the second evaluation algorithm 18 with reference to FIG.

FIG. 2 shows a known chain test bench 22, from which data is transmitted to a computer program 24, which comprises program code instructions 26 for implementing the method 10. The computer program 24 is stored on a computer, e.g. an automation device 28 with a processing unit 30, executed.

The first evaluation algorithm 16 according to an embodiment of the invention is shown in FIG. Therein, in a first step 32, for each signal, and calculates each of the three Refe rence ^¬ chain characteristics. The signals come from ver ^{¬ different} sensors of a corresponding KettenprüfStands on which the reference chains have been tested, for example, the Kettenprüf- stand 22 shown in FIG 2. The track tester 22 includes vibra ^¬ tion sensors 34, three sensors horizontal and three other sensors can measure vertical vibrations , In this case, horizontal and vertical vibration signals are measured in a rear area of the chain test bench 22 and at two positions (left and right) in a front area of the chain test bench 22. The test rig 22 comprises chain wei ^¬ terhin a pressure sensor 36, position transmitter 38 which transmit a position of the clamping cylinder, and means for transmitting a torque and a clock signal. As a means for transmitting the torque and the clock signal, for example, is a module, such as a converter 40 (for example, a device known under the name SINAMICS S120 modules), with which the signals can also be recorded. The individual signals recorded on the chain test bench for the reference chains and the corresponding calculated characteristic value for the abovementioned exemplary embodiment are listed in Table 1. Description i Calculated characteristic i

i FFT, no further i i signal 1 drive signal (8 kHz)

i Evaluation i

Vibration signal, front ii FFT -> meshing-i

: Signal 2 j

left vertical i frequency (peak, PiB ¹ )!

Vibration signal, front ii FFT -> meshing signal i Signal 3

left horizontal i frequency (peak, PiB ¹ )!

Vibration signal, front ii FFT -> meshing-i

Signal 4

right vertical ü frequency (peak, PiB ¹ )!

Front vibration signal ii FFT -> meshing signal i Signal 5

right horizontal ii frequency (peak, PiB ¹ ):

Vibration signal rear FFT -> meshing signal i i Signal 6 j

left vertical ii frequency (peak, PiB ¹ );

Vibration signal at rear FFT -> meshing-i

Signal 7

left horizontal i frequency (peak, PiB ¹ )!

Position clamping cylinder

Signal 8 ii MAX-MIN i

Left

Position clamping cylinder

i Signal 9 ii MAX-MIN i

right i signal IO! Pressure on the clamping cylinder ii MAX-MIN i

Signal 11 Torque (1 kHz) ii Mean i

PiB = "Power in Band

Table 1 The characteristics for the signals 1 to 7 were calculated using the

Fast Fourier Transform (FFT) and for the signals 2 to 7 with a PiB algorithm ( "power in band" -Algo- algorithm) evaluated. The PiB algorithm sums the Ampli ^¬ amplitudes of all the frequencies in a predetermined frequency range and describes it as a characteristic value. this is explicitly emphasized by foreign body engagement, such as dirt, excited vibrations. In the present embodiment, an evaluation for the an ^¬ drive signal (signal 1) was found to be not useful, because in the Fast Fourier Transform has been calculated to have a meshing frequency similar to that of the sprocket. Therefore, a PiB (Power in Band) evaluation due to an overlay by sidebands of both frequencies is not meaningful, nor was a trend found in a peak evaluation For the signals 2 to 7, the meshing frequencies according to PiB and Peaks were evaluated signals 8 to 10 were evaluated according to maximum and minimum values. From signal 11 (only torque ^¬ ment) a mean value is formed.

In a second step 42, the characteristic values of the reference chains over time are graphically represented and there is vorzugt be ^¬ a selection of relevant signals and characteristics. This selection takes into account, for example, that a trend can not be recognized in all characteristic curves or that certain characteristic curves do not produce any representative results due to special operating conditions of a chain. If a chain is repeatedly hit by, for example, intervals of several weeks time, resulting torque peaks so that the torque curve deviates greatly from a torque curve of a Ket ^¬ te, which is operated continuously. In this exporting ^¬ insurance for the characteristics have made of five signals detect a trend; these are the signals 2 and 3 (PiB evaluation), 8, 10 and 11 according to Table 1.

If there are different service lives for the reference components in the wear tests, whereby these have been operated with different loads, such as different clock rates or different pressure on the clamping cylinder, it makes sense to normalize the lifetimes of the individual reference components. For this purpose, in a third step 44, the time axis of the characteristic ^{curves is} scaled with a measure of the load to which the respective reference chain was exposed during its operation. For example, if a first chain was operated at 120 strokes, a second chain at 100 strokes, and a third chain at 140 strokes, and at the wear tests, a life span for the first chain of 43 days, for the second chain of 57 Days and for the third chain of 35 days, a lifetime scaled to a load of 100 strokes equals 51.6 days for the first chain, 57 days for the second chain, and 49 days for the third chain.

In a fourth step 46, an approximation of the scaled characteristic curves is performed by a mathematical function for the selected characteristic values. The mathematical function used for the approximation is, for example, a quadratic polynomial.

If appropriate, the fourth step 46 can also be carried out before the third step 44. FIGS. 4 and 5 show examples of characteristic curves and an approximation by a quadratic polynomial. In FIG. 4, the time in days and on the ordinate a PiB tooth ^engagement frequency is plotted in a coordinate system on the abscissa. 4 shows the trend of

Oscillation signal as a reference characteristic curve 48, 50, gemes ^¬ sen front left in the vertical (signal 2) and horizontal (signal 3) direction, as well as approximated by a quadratic polynomial reference characteristic curve 52, 54 for each Refe ^¬ renzkennwertverlauf 48, 50 for the third chain ( so). The chain bar was Betrie ^¬ ben at 140 bars and a pressure of. 5 In FIG. 5, the time in days and on the ordinate a torque in Nm are plotted in a coordinate system on the abscissa. 5 shows the trend of the torque (mean value ^¬) (signal 11) of the same chain as the reference characteristic value curve 56 and the correspondingly with a square polynomial appro ^¬ ximierten reference characteristic value curve 58th

In FIG. 6, the time in days is plotted in a coordinate system on the abscissa and the remaining life in days is plotted on the ordinate. Figure 6 shows the signals for the above mentioned five he ^¬ mediated remaining lifetimes of 5 already considered in reference to Figures 4 and chain. For determining the remaining service life 12 (FIG. 1), for each of a sensor 34, 36, 38, 40, a corresponding time value is determined from the corresponding approximated reference characteristic curve 52, 54, 58. From the difference between the expected total lifetime and this time value, the remaining life 12 (FIG. 1) of the chain results. In general, the expected Gesamtle ^¬ service life of a machine component, for example, may be the overall life of the reference component of the reference characteristic curve of value was measured, or all in the United ^¬ schleißversuchen used machine components averaged total lifetime, or even a specified by the manufacturer of the machine component overall life.

To reduce the scatter of the remaining lifetime values of the individual signals are in a fifth step 60 of the first evaluation algorithm 16 (FIG 3) the coefficients of the quadrati ^¬ rule polynomials (58 matched characteristic curves 52, 54,) are averaged characteristic value specific of all chains. In addition, a weighting factor can be included that expresses the relevance of the corresponding characteristic value. This results in wear patterns that are averaged over several chains.

The second evaluation algorithm 18 is shown in detail in FIG. 7 and comprises a first step 62, in which an associated time value from a corresponding approximated and optionally averaged reference characteristic curve 52, 54, 58 is determined for each current, selected parameter. The estimated remaining life for each selected characteristic value results from the difference between the expected total lifetime of the chain and the time value determined for each characteristic value. In a second step 64 of the second evaluation algorithm 18, the remaining lifetimes determined in the first step 62 of the second evaluation algorithm 18 are averaged. The averaged value then corresponds to a residual service life 12 determined by method 10 (FIG. 1) (FIG. 1).

In FIG. 8, the time in days and on the ordinate the remaining life in days is plotted in a coordinate system on the abscissa. FIG 8 shows the with the invention Method 10 (FIG 1) determined residual life 12 of a chain, wherein the remaining lifetimes of five selected signals (see above) were averaged. In addition, FIG 8 shows a calculated remaining lifetime, wherein the approximated Referenzkennwert- extending 58 has been used for a torque as a sole reference parameter ^¬ extending for determining the remaining service life. A residual service life curve 66 of the residual service life determined on the basis of the torque is similar to a residual life curve 68 for the reference chains, which is averaged over five signals, so that a determination of the residual service life 12 alone with characteristic values for a torque is also promising.

Applications of the method have also shown that no rapid sampling of the torque for the Restlebensdauerer- mediation is necessary, but on the basis of highly sampled drive signals from the (SINAMICS) inverter 40 (torque, speed) transmission monitoring would be conceivable.

The method described above with reference to an embodiment using chains can also be applied to other machine components, e.g. Ball screws, apply.

Individual aspects of the description entered here can thus be briefly summarized as follows: A method 10 for determining a remaining service life 12 of a machine component 17 is specified,

wherein a reference characteristic ^curve 48, 50, 56 of at least one characteristic value is retrieved from at least one wear test of a reference component matching the machine component 17 from a database 14,

wherein an approximation of the reference characteristic curve 48, 50, 56 is performed by a mathematical function,

wherein for the machine component 17 at least one current characteristic value 20 is called up by a sensor 34, 36, 38, 40,

- Where from the approximated characteristic curve 52, 54, 58 a the current characteristic value 20 associated time value determined will and

wherein the remaining life 12 results as the difference between an expected total service life of the machine component 17 and the determined time value.

Claims

claims

A method (10) for determining a remaining service life (12) of a machine component (17),

wherein a reference characteristic value profile (48, 50, 56) at least ei ^¬ nes characteristic value of at least one attempt at wear to the machine component (17) appropriate reference component is retrieved from a data base (14),

wherein an approximation of the reference characteristic curve (48, 50, 56) is performed by a mathematical function, wherein for the machine component (17) at least one aktuel ^¬ ler characteristic value (20) by a sensor (34, 36, 38, 40) is called,

wherein from the approximated characteristic curve (52, 54, 58) a time value associated with the current characteristic value (20) is determined, and

wherein the remaining service life (12) results as the difference between an expected total service life of the machine component (17) and the determined time value.

2. The method of claim 1, wherein, if at least two wear tests historical characteristic values are retrieved from the database (14), with different total lifetimes of the respective reference components, the time axis of the characteristic curves (48, 50, 56) before the approximation with a measure for a load that has been exposed to the respective reference components during their operation.

3. The method of claim 2, wherein the or each Kennwert- extending (48, 50, 56) with a quadratic polynomial is approxi ^¬ mized.

4. The method of claim 3, wherein, when retrieving data from at least two wear tests, the respective coefficients of the quadratic polynomials are averaged for each characteristic.

5. The method according to claim 4, wherein the characteristic values are weighted before relevance.

6. The method according to any one of claims 2 to 5, wherein a for at least a first and a second current characteristic value (20) determined residual life (12) is averaged.

7. The method according to any one of the preceding claims, wherein when availability of several different reference characteristic run (48, 50, 56) a selection of the usable for determining the remaining life (12) characteristic values is performed

8. The method according to any one of the preceding claims, wherein a reference characteristic curve (56) is used with characteristic values for a torque as the sole reference characteristic curve (48, 50, 56).

A computer program comprising computer executable program code instructions for implementing the method of any one of claims 1 to 8 when the computer program is executed on a computer.

10. Computer program product, in particular storage medium, with a computer-executable computer program according to claim 9.

11. Computer system on which a computer program according to claim 9 is loaded.