EP2047118A1

EP2047118A1 - Method for fault localization and diagnosis in a fluidic installation

Info

Publication number: EP2047118A1
Application number: EP07703456A
Authority: EP
Inventors: Jan Bredau; Reinhard Keller
Original assignee: Festo SE and Co KG
Current assignee: Festo SE and Co KG
Priority date: 2007-02-14
Filing date: 2007-02-14
Publication date: 2009-04-15
Anticipated expiration: 2027-02-14
Also published as: US7917325B2; US20100153026A1; ATE529643T1; CN101427033A; WO2008098589A1; TW200846275A; KR20100014067A; EP2047118B1

Abstract

The invention relates to a method for fault localization and diagnosis in a fluidic installation, wherein the fluidic volume flow of the overall installation, or of at least a partial region of the same, or a variable as a measuring variable that is dependent thereon, are each recorded during an operating cycle and compared to stored references. At the time of a variation or change of the variation from the reference, it is determined in which component or components (10-14) of the installation a process influencing the fluid consumption has taken place in order to then identify the same as faulty. In case of such a variation, or change of variation, and in case of a simultaneous occurrence of a plurality of processes influencing the fluid consumption by a plurality of components (10-14), an exclusion process is carried out, in which subsequent activities, in which at least one of these components (10-14) is involved, are tested in subsequent test steps in order to see if in turn a variation, or change of variation occurs, wherein the components involved are excluded from further testing as non-faulty in each of these testing steps, if no variation, or change of variation, occurs.

Description

t February 12, 2007

FESTO AG & Co, 73734 Esslingen

Method for fault isolation and diagnosis on a fluidic system

The invention relates to a method for limiting errors and diagnosis in a fluidic system, wherein the fluidic volume flow of the entire system or at least a portion of the same or a dependent size is detected as a measured variable in each case during one operating cycle and compared with stored references, and wherein at the time a deviation or change in the deviation from the reference is determined, in which component or components of the system, the FIUI idverbrauchverbrauching process has taken place, in order to recognize them then as faulty.

In such a method, known from WO 2005/111433 A1, the air consumption curve for error localization is evaluated. In the case of deviations from a reference, i5 is closed from the time of the deviation on the faulty subsystem (for example valve actuator unit) or the faulty component. Such errors, which can occur in fluidic systems, have their causes for example in the wear of the components, in improper Mon-

20 days, loose glands, porous hoses, process disturbances and the like, in the movements of the fluidic drives outer, and other leaks of various kinds. To diagnostic errors due to the change of certain boundary conditions, such as pressure and temperature to

P 26118 / PCT. ,

February 12, 2007 avoid, this document mentions the possible correction of air consumption with pressure and temperature. In particular, in large fluidic systems in which a plurality of subsystems are partly active simultaneously, s can not be determined with the known method, which of these components is faulty.

It is therefore an object of the present invention to improve the method of the type mentioned at the beginning in such a way that even with simultaneously active components and sub-systems an error, in particular a leak, can be unambiguously assigned to a specific component or to a specific subsystem ,

This object is achieved by a method having the features of claim 1.

By virtue of the method according to the invention, the leakage location can be delimited stepwise in an advantageous manner so that the fault location can be determined in a simple manner even with a large number of simultaneously active components or subsystems. This is all the more a special advantage

20, as a strictly sequential sequence in fluidic systems, especially in large fluidic systems, is relatively rare. Another advantage is that only the Aktorstellsignale and a volume flow sensor for determining the leakage location are required, that is, limit switch

25 th of actuators are not absolutely necessary. The more different the axis movements are and the more different cycles occur with simultaneously moving subsystems or components or combinations thereof, the more advantageously the method according to the invention can be used.

30 are set. Not only is it trying to find the leak-causing subsystems or components, but also it

P 26118 / PCT. ,

February 12, 2007 Also definitely excluded subsystems, components or actuator chambers are excluded.

The measures listed in the dependent claims advantageous refinements and improvements of the method 5 in An-5 specified method are possible.

As stored references, fluid consumption reference curves formed from integrated volume flow values or conductance reference curves formed from integrated conductivity values (Q / P) have proven to be particularly suitable, which are compared with lo corresponding measurement curves.

An even greater degree of diagnostic accuracy and accuracy in finding leak locations is achieved by parameter-dependent compensation of the volume flow values or Leitwertgrδßen, the compensation particular temperature dependent and / or fluid-dependent and / or moisture-dependent and / or the particle content of the fluid and / or time or event-dependent for different operating states.

Expediently, a plurality of parameter-dependent or parameter-dependent compensated fluid consumption

Reference curves or conductance reference curves stored in a selection matrix and can be selected or specified for the respective cycle, for example, by successively checked for correlation with the respective 25 duty cycle.

The reference curves are expediently detected in a learning mode, in particular also during the later operation of the fluidic system.

P 26118 / PCT

February 12, 2007 In order to rule out that detected deviations from the measured curve and the reference curve are based on a time error, a curve comparison with respect to possible time shifts is preferably carried out prior to the diagnosis of leakage, wherein a time shift exceeding a tolerance value is switched to further stored reference curves for their checking or an error message and / or a stop of another leakage diagnosis is triggered.

In the case of the leakage diagnosis according to the invention, differential values or a difference curve between the measured variable curve and the reference curve are formed for a particularly advantageous evaluation. This difference curve is expediently filtered in a frequency-dependent manner by means of an integrator, which in particular has a phase shift of -90 °, in order to filter out interference signals and spikes. A filtered compensation curve is then formed by calculating the slope of the integral of the difference values or the difference curve, which then provides a particularly simple, purposeful evaluation

20 allows.

An embodiment of the invention is illustrated in the drawing and explained in more detail in the following description. Show it:

FIG. 1 shows a pneumatic system, in the supply of which a flow meter is connected;

Figure 2 is a Leitwertdiagramm to explain the occurrence of a time shift between the trace and reference curve and

P 26118 / PCT

February 12, 2007 FIG. 3 Conductance diagrams for explaining the diagnosis of leakage.

In Figure 1, a pneumatic system is shown schematically, which could in principle also be another fluidi- 5 plant, such as a hydraulic system, act.

The pneumatic system consists of five subsystems 10-14 or components, which may each be actuators, such as valves, cylinders, linear actuators and the like, as well as combinations of the same. These subsystems 10-14 are fed by a pressure source 15, wherein a flow meter 17 for measuring the flow or the volume flow is arranged in a common supply line 16. The subsystems 11, 12, on the one hand, and the subsystems 13, 14, on the other hand, in turn each form an i5 system with a common supply line.

An electronic control device 18 is used to specify the process of the system and is electrically connected to the subsystems 10-14 via corresponding control lines. The subsystems 10-14 receive control signals from the electronic control device 18 and send sensor signals back to them. Such sensor signals are for example position signals, limit switch signals, pressure signals, temperature signals and the like, which are not absolutely necessary in the simplest case.

The flow meter 17 is connected to an electronic diagnostic device 19 which additionally supplies the signals of a temperature sensor 20 and a pressure sensor 21 for measuring the temperature T and the pressure P in the supply line 16, ie the temperature and the pressure of the fluid

30 are. Furthermore, a fluid sensor 23 for detecting the kind

P 26118 / PCT. ,

February 12, 2007 the fluid used and a moisture and / or particle sensor 24 for detecting the moisture content and the particle content of the fluid connected to the diagnostic device 19. This additionally has access to the Ablaufpro- program 5 of the electronic control device 18. The diagnostic results are supplied to a display 22, these diagnostic results of course also stored, printed, optically and / or acoustically displayed or a center via lines or wirelessly can be transmitted.

The sensors 20, 21 as well as 23 and 24 can also be dispensed with in a simplest embodiment, although at least one temperature sensor 20 and one pressure sensor 21 can be expediently provided.

Of course, the diagnostic device 19 can also be integrated in i5 of the electronic control device 18, which may contain, for example, a microcontroller for carrying out the sequence program and optionally for diagnosis.

In the case of a very large number of subsystems or components, these can be divided into several groups, each group having its own flow meter 17 in order to independently diagnose the subregions of the system assigned to the groups, as in the prior art mentioned at the outset is described.

The method for error limitation and diagnosis will now be explained in the following with reference to the described pneumatic system and the conductivity diagrams shown in FIGS. 2 and 3.

The diagnosis can be made in the simplest case by comparing stored and selected fluid consumption reference curves

P 26118 / PCT

February 12, 2007 be carried out with corresponding measured value curves, wherein the fluid consumption reference curves are formed from integrated or totalized volumetric flow values. Better results are achieved by using diagnostic control values, where the diagnostic control value is a characteristic variable of a fluidic system or of a fluidic system that consists of various subsystems. The conductance characterizes the behavior of the entire system over a defined cycle. Conductance reference curves are formed in the simplest case from integrated Leitwertgrδßen Q / P, where Q is the respective volume flow value and P is the measured working pressure. These conductance reference curves are compared with corresponding measured value curves, ie with measured value curves formed from integrated conductance variables. i5 The conductance values or conductance curves and conductance reference curves can be compensated and refined by further measurement parameters, for example by the measured operating temperature T, the moisture content and / or the particle content of the fluid, the type of fluid and of each

20-time time- or event-dependent operating states. Such operating states are, for example, the warm-up, the operation after a long standstill, the reclosing when retrofitting or the operation at predeterminable time intervals, that is, for example, after a one-hour or ten-hour

25 or more hours of operation. The following discussion of fault isolation and diagnosis is based on conductance, and fluid consumption values could be used accordingly.

To generate the reference curves, a repetitive 30 cycle of the entire process must be selected. Non-cyclic processes can be subdivided into subcycles, to which the diagnostic procedure is then applied. Various operating

P 26118 / PCT

February 12, 2007 Stands in a process can be considered by including and storing a set of reference curves in a selection matrix. This also applies to the influence of different parameters.

5 For evaluation, the respective measurement curve must now be synchronized with the selected or selected reference curve, ie, without leakage, the two curves are congruent, with leakage they run synchronously in time, but show deviations in the amplitude. The two curves to be compared must therefore first be checked for correlation, that is, it must be checked whether temporal shifts have occurred, for example due to changed processes within a cycle. If time shifts have been detected over a defined tolerance i5, the further evaluation of leaks is stopped and a message regarding changes in the times of subsystems is generated. A time error is detected if the value of the air consumption at the end of the cycle is within a tolerance range, but the cycle time

20 is different, as shown in Figure 2.

There, the two curves run synchronously until the time ta, and from this point on a time difference Δt occurs between the curve Km and the reference curve Kref, which remains constant until the end of the cycle at the point in time tb. If a time

As errors increase more and more as the cycle progresses, an attempt can be made to correlate by choosing a different reference curve. Only when all stored reference curves have been checked and no correlation could be achieved, is there a faulty time shift

30 before, and a subsequent diagnosis for leakage is dispensed with. A corresponding message can then be displayed, saved or forwarded.

P 26118 / PCT

February 12, 2007 If no time error is detected, the difference between the nominal value or measured value and the reference value, that is to say between the measured quantity curve Km and the reference curve Kref, is set in the next step, as shown in FIG. 3, above. The formed difference curve, which is shown in FIG. 3, below, defines the summed distance of the measured variable curve from the reference curve at any time. The time points of leaks show the staircase increases in the difference. In the following evaluations, these differences in the difference are assigned to the leak-causing subsystems or components or actuator chambers.

In order to remove also unwanted fluctuations, glitches and the like, the calculated difference or difference curve can be filtered. In conventional filtering, the change in phase and amplitude is frequency dependent. In order to achieve a frequency-independent filtering, an integrator is used, which has a fixed phase shift of -90 °. Thus, in the later evaluation of the signals no different phase shift

20 exercise. The amplitude response can be adjusted by changing the sampling time so that there is a constant attenuation of the amplitude in the desired frequency range, while other frequencies are filtered.

For evaluation, a compensation function of the integral of the calculated difference is subsequently formed. The choice of the corresponding compensation function can be made according to the Gaussian least squares principle. It must be determined which curve best suits the calculated measurement points of the difference. In the following, a straight line is chosen as the simplest possibility of a compensation function. Of course, other compensation functions are possible. Any occurring leakage leads

P 26118 / PCT. ,

February 12, 2007 to a change in the slope and the center distance of the regression line to the abscissa. In determining the slope from the integral of the difference, a representation is obtained which corresponds to the difference curve shown in FIG. 3, but is phase-shifted by minus 90 °. In the calculation of the axial distance from the integral of the difference is also a representation that corresponds to the differential curve shown in Figure 3, but is phase-shifted by minus 90 ° and is mirrored on the abscissa. The advantage of calculating the regression line is that leaks, ie changes in the slope, always have the same effect in terms of time. Leaks at a later point in a cycle affect the center distance significantly more than leaks at the beginning of a cycle. In the later temporal range of references, higher, accumulated errors occur at the current value. Actual leakages therefore change the center distance much more clearly at a later point in the cycle than any deviations from the reference, for example due to aging of the system.

20 would be. The evaluation described below therefore takes into account both changes in the pitch, as well as changes in the axial distance.

In the case of the exclusion principle according to the invention, certain areas can be left out of the error analysis during the course of the error analysis

25 exclude, so that the number of for a leak in

Considering subsystems and components or actuator chambers increasingly reduced. This makes use of the fact that in machine processes, the same groups of subsystems never move at the same time, ie

3o are active, or there are never the same Aktuator chambers simultaneously under pressure. Thus, the candidate actuator chambers are more and more limited, and the

P 26118 / PCT

February 12, 2007 say regarding leakage is becoming more and more defined. For example, actuator chambers are always excluded from further consideration regarding leakage if they are vented at a time and at the same time no leakage occurs. In the following, a diagnostic cycle will be explained by way of example with reference to FIG. 3:

At the time tθ leakage occurs. At this time, the chamber A of the subsystem 10, the chamber B of the subsystem 11 and the chamber A of the subsystem 12 are vented. These lo three chambers are thus considered as the cause of the leakage. At the same time, the chamber B of the subsystem 10, the chamber A of the subsystem 11 and the chamber B of the subsystem 12 are inactive, ie not ventilated, so that these actuator chambers are excluded from further consideration.

At time t 1, another leakage occurs. At this time, the chamber A of the subsystem 10, the chamber B of the subsystem 13 and the chamber A of the subsystem 12 is vented. This means that the chamber B of the subsystem 11 is excluded for further consideration and only the

2o chambers A of the subsystems 10 and 12 come into question for the leakage.

At time t2, chamber A of subsystem 10, chamber B of subsystem 14 and chamber A of subsystem 11 are vented. The chamber A of subsystem 11 has already been excluded for further consideration. The chamber A of the subsystem 12 is now also excluded as the cause of the leak, so that it can be finally concluded that the chamber A of the subsystem 10 is responsible for the leakage.

P 26118 / PCT

February 12, 2007 It is often possible to detect the causative system even with a single increase of ΔK, that is to say with a single occurrence of a leak. For example, in a modification of the previously described example, leakage would occur only at the time point tθ at which the chamber A of the subsystem 10, the chamber B of the subsystem 11 and the chamber A of the subsystem 12 are vented, and subsequently become later In turn, once the chamber B of the subsystem 11 and the chamber A of the subsystem 12 vented, while the lo chamber A of the subsystem 10 is not involved, and there is no leakage, then the chamber B of the subsystem 11 and the chamber A of Subsystem 12 are excluded as causative components, and it can then finally the chamber A of the subsystem 10 are recognized as leakage-inducing.

A particularly suitable type of evaluation, in particular in the case of a very large number of subsystems or components, is that each chamber of an actuator, which in the case of a working cylinder is for example two chambers, two each

20 counters are assigned. Furthermore, each chamber is assigned a timer. The timer is used to exclude additional actuator chambers or components from the consideration of a leak. If a chamber or a component is under pressure and no leakage occurs

25 within a preselected time value of the timer, this chamber is also treated as not involved in the leakage and excluded for further leak detection. The electrical components, ie counters and timers, are located, for example, in the diagnostic device

30 19. At the start of an operating cycle, the timers are started and, if a leak occurs, they are reset to zero and held there until

P 26118 / PCT

February 12, 2007 Termination of leakage held. If the relevant chamber is now under pressure during the reset state of the timer or at least during a part of the reset state, this chamber is considered to be responsible for the leakage and it is checked whether the gradient and the center distance of the compensation straight line or another compensation function has increased by a predefinable value or by a predeterminable percentage (based, for example, on the respective maximum value of the or one of the preceding lo cycles). In this case, the one for the

Incline responsible counter and / or the counter associated with the axis distance increased by the value 1. The more different the axis movements of a plurality of simultaneously moving subsystems or components and the more different cycles occur, the more accurate this method becomes. For each leak where the component or chamber of a component is pressurized, the associated counters will be incremented by a further counter, depending on the increase in the slope and / or the center distance.

20 value increased. The counts of both counters of a chamber or component are added together at the end of the cycle. The chamber with the highest total count at the end of an operating cycle is most likely to be responsible for the leakage.

25 The chamber or component with the second highest total count is the second largest likely to leak. This is important if several leaks occur in the system. If more than a fixed percentage of chambers, e.g. more than 50%, as the

30 ckage has been detected, this is defined as system leakage. This method involves a graded evaluation with the goal, even if not clear

P 26118 / PCT

February 12, 2007 Detection of the leak site at least give the maintenance personnel hints.

To increase the accuracy of the analysis, several cycles can be considered. The sum of the multiple analyzes 5 then gives more accurate information about the chamber or component causing the leakage or the chambers or components causing the leakage.

In a simpler version, a single timer may be provided for all chambers or components, each of which is reset to zero during the occurrence of a leak and held there during the occurrence of the leakage. During this period, it is then checked which chambers or which components are active, that is to say are pressurized.

In a simpler embodiment of the method, for example, only the center distance or only the slope or its change can be evaluated. Per chamber or per component or per subsystem then only one counter is required. A further simplification of the procedure

20 rens can still be done by completely dispensing with the determination of the center distance or the slope and only the counter of a chamber or a component is increased by the value 1, if this chamber or component at least during a sub-period of a leakage interval

25 was ventilated.

P 26118 / PCT

12. February 2007

Claims

claims

1. A method for error limitation and diagnosis at a fluidic system, wherein the fluidic volume flow of the entire system or at least a portion of the same or a dependent thereon variable as a measured variable in each case during a

5 nes operating cycle is detected and compared with stored references, and is determined at the time of a deviation or change in the deviation from the reference at which component or components of the system a fluid consumption influencing Vorlo gang has taken place to this then to recognize as faulty, characterized in that in such a deviation or variation of the deviation and simultaneous occurrence of several activities influencing the fluid consumption of several components (10-14) a Ausschlussverfah- ren is performed ren, in which at subsequent activities in which At least one of these components (10-14) is involved, is checked in each case in further test steps, whether in turn a deviation or change of deviation occurs, in each of these further test steps, the Betei-

Components which have not been subject to error may be excluded from further testing if no deviation or variation of the deviation occurs.

2. The method according to claim 1, characterized in that in each case in further test steps in turn occurring

25 Deviation or variation of the deviation to this

When the components are not actively involved, they can be excluded from further testing as not subject to errors.

P 26118 / PCT

February 12, 2007

3. The method according to any one of the preceding claims, characterized in that the stored references from integrated volume flow values (Q) formed fluid consumption reference curves or from integrated Leitwertgrö-

5 ßen (Q / P) are formed, where P is the measured working pressure, which are compared with corresponding Meßgrößenkurven.

4. The method according to claim 3, characterized in that the volumetric flow values (Q) or conductance variables (Q / P) are parameter-dependent compensated, in particular temperature-dependent and / or fluid-dependent and / or moisture-dependent and / or dependent on the particle content of the fluid and / or or time or event-dependent for different operating states.

5. The method according to claim 4, characterized in that i5 a plurality of parameter-dependent fluid consumption reference curves or conductance reference curves are stored in a selection matrix.

6. The method according to claim 5, characterized in that the reference curves are detected in a learning mode, in particular

20 special also in the later operation of the fluidic system.

7. The method according to any one of the preceding claims, characterized in that prior to the diagnosis of leakage, a curve comparison with respect to possible temporal shifts takes place, wherein at a tolerance value over-

25 is switched over to other stored reference curves for their verification or an error message and / or a stop of another leakage diagnosis is triggered.

P 26118 / PCT

February 12, 2007

8. The method according to any one of the preceding claims, characterized in that for leakage diagnosis difference values or a difference curve (.DELTA.K) between measured variable curve (Km) and reference curve (Kref) is formed.

5 9. The method according to claim 8, characterized in that the difference curve (.DELTA.K) is frequency-dependent filtered by means of an integrator, which in particular has a phase shift of -90 °.

10. The method according to claim 8 or 9, characterized in lo net, that a compensation function of the integral of the calculated difference values or the difference curve is formed, which best matches the calculated measuring points of the difference.

11. The method according to claim 10, characterized in that i5 the compensation function is calculated according to the Gaussian least squares principle.

12. The method according to any one of the preceding claims, characterized in that during the period of a deviation or change in the deviation of a timer on a

20 is set and a comparison is made as to which component or components were active during at least a sub-interval of that period of time.

13. The method according to claim 12, characterized in that each component (10-14) or each chamber of a component

At least one counter is assigned, the count of which is incremented by one count each time the component (10-14) or chamber of the component is pressurized during at least a sub-interval of the presence of the set value of the timer.

P 26118 / PCT. ,

February 12, 2007

14. The method according to claim 13, characterized in that each component or each chamber of a component is associated with a slope counter whose count is only increased in each case when the slope of the compensation radio

5 tion at least by a predetermined value or percentage during the presence of the set value of the timer o- the active state of this component or chamber during the presence of this set value increases.

15. The method according to claim 13 or 14, characterized in lo, that each component or each chamber of a component is assigned to a distance difference counter whose count is only increased if the center distance of the compensation function at least by a predetermined value or percentage during the presence of the set value of the i5 timer or the active state of that component or chamber during the presence of this set value.

16. The method according to claim 14 and 15, characterized in that at the end of an operating cycle at each component or each chamber of a component, the counter readings of the

20 counter and the Achsabstandszählers are added, the highest total count or the highest total counter readings are considered the highest leakage probability for the respective component or chamber of a component.

P 26118 / PCT

February 12, 2007