DE102018218696A1 - Method and diagnostic device for determining friction effects in continuous valves. - Google Patents

Abstract

The invention relates to an improved method and a diagnostic device for determining friction effects in continuous valves, in which at least one process value (PV) that is directly related to a valve position of a continuous valve and a desired value for the valve position (MV) are evaluated by in in a first step, first indicators (SI1 to SI6) for existing friction effects in the continuous valve are calculated. According to the invention, a resultant indicator (SIres) is then calculated based on the previously calculated indicators (SI1 to SI6), which indicates a probability of the occurrence of friction effects in the continuous valve, the first indicators being standardized, the results of the standardization sorted in descending order and counting how many of the indicators indicate friction effects, how many are undecided, and how many friction effects exclude. On the basis of the count results, a decision distinction is then made using a decision tree, according to which the resulting indicator (SIres) is derived and displayed.

Description

The invention relates to a method and a corresponding diagnostic device for determining friction effects in continuous valves according to the preamble of claim 1.

A wide range of sensors and actuators are mostly used in process or process engineering systems in connection with control loops for monitoring and influencing physical or chemical parameters in technical processes. Permanent and automatic monitoring of the sensors and actuators or the control loops is advantageous, so that when the performance of individual components deteriorates, measures for maintenance and / or optimization can be intervened in a timely and targeted manner at the correct point in the system.

Frequently, performance problems of system components are not solely due to the control loop, but rather to problems of the system component itself. The performance problems of valves that are used in every industrial system can be, for example, valve lash, dead time or oscillation and valve friction problems act. In the case of continuous valves in particular, a more precise analysis with regard to the occurrence of increased friction effects (English static friction, or stiction for short) is desirable.

In fluid technology, especially in hydraulics, the term continuous valve is a collective term for generally electrically controlled directional control valves, which do not switch discretely (open / close), but rather allow the switching positions to change continuously. The volume flow of the fluid can thus be adjusted. Proportional valves, control valves and servo valves are examples of continuous valves.

The term friction effects encompasses both static friction and sliding friction. Friction effects with continuous valves mostly arise from wear or external environmental influences and lead to increased energy consumption over time and ultimately to valve failure. It is therefore all the more important to be able to make a reliable and early assessment of the condition of a valve. From the book “Detection and Diagnosis of Stiction in Control Loops” by M. Jelali and B. Huang, State of the Art and Advanced Methods, SpringerVerlag, London, 2010 (Ref1) and from the publication of Choudhury, M .; Shah, S .; Thornhill, N .: "Detection and Quantification of control valve stiction", Department of chemical and material engineering, University of Alberta, Edmonton AB, Canada, 2006 (Ref2), methods are known for determining from the measurement and setpoints of a control circuit with a continuous valve as an actuator whether the valve has increased static and / or sliding friction (summarized here as friction effects). In practice, it has been shown that not every of these methods for detecting friction effects gives a good result in every situation and it is not possible to decide which of the methods gives the correct assessment.

The invention is therefore based on the object of specifying an improved method and an improved diagnostic device for determining friction effects in continuous valves which, in comparison with the already known methods, provide more precise results with regard to the analysis of friction effects in continuous valves.

This object is achieved in terms of the method aspects by the features specified in claim 1. Further developments of the invention are described in the dependent claims, a corresponding diagnostic device in claim 2, a computer program for carrying out the method in claim 3 and a corresponding computer program product in claim 4.

The advantage of the method according to the invention lies in a more precise and reliable statement with regard to friction effects in continuous valves. The combination of the results from the individual methods allows a significantly improved stiction recognition in comparison to the use of only one method or in comparison to a simple averaging over the results of the individual methods. Another advantage is the quick calculation of the combination of the individual results, since the method according to the invention is based on a simple case differentiation.

Based on the drawings, in which an embodiment of the invention is shown, the invention as well as configurations and advantages are explained in more detail below.

• 1 einen Ausschnitt eines Blockschaltbilds einer verfahrenstechnischen Anlage mit einem Ventil, Strecke und Diagnoseeinrichtungen,
• 2 ein Blockschaltbild einer Regelstrecke mit Diagnoseeinrichtungen,
• 3 Signalverläufe des Ausgangssignals des Reglers MV und des Messwerts PV der Ventilstellung in Abhängigkeit von der Zeit bei Auftreten von erhöhten Reibungseffekten,
• 4 ein so genannter Scatter-Plot zwischen Reglerausgang MV und Ventilstellung bei Auftreten von erhöhten Reibungseffekten,
• 5 eine Übersicht über die Bewertungsgrenzen der einzelnen Methoden, und
• 6a und 6b: für zwei unterschiedliche Ventile eine Gegenüberstellung der Ergebnisse von 6 Einzelmethoden für Daten eines simulierten Regelkreises

Show it:

• 1 a section of a block diagram of a process engineering system with a valve, line and diagnostic devices,
• 2nd 2 shows a block diagram of a controlled system with diagnostic devices,
• 3rd Waveforms of the controller output signal MV and the measured value PV the valve position depending on the time when increased friction effects occur,
• 4th a so-called scatter plot between the controller output MV and valve position when increased friction effects occur,
• 5 an overview of the assessment limits of the individual methods, and
• 6a and 6b : For two different valves, a comparison of the results of 6 individual methods for data from a simulated control loop

1 shows a section of a block diagram of a process plant with a valve V , a route S and various diagnostic devices D and D2 used here to detect and determine the frictional effects of the valve V be used. With the valve V it is a continuous valve, the valve position via a manipulated variable MV is controlled. The valve is monitored at its outlet or after passing through a certain process section S (e.g. pipes) measured a process variable that is directly related to the valve position (e.g. the flow through a pipe behind the valve). About the process value PV can therefore be directly deduced from the valve position. In some cases, a direct position feedback SR can also take place, ie a valve position is measured directly. The measured values SR or PV together with the desired valve position (manipulated variable MV ) a diagnostic device D supplied as input variables, in this embodiment using an algorithm 6 Indicators of friction effects related to the valve can be determined and used as output variables (here SI1 to SI6 ) are output.

According to 2nd becomes the valve V by means of a control device R supervised. At the exit of the controlled system S a process value is measured as a controlled variable that is in relation to the valve to be monitored and to a setpoint SW , which is specified, for example, by an operating unit, to the control device R returned. With the control device R can be conventional P, PD, PI or PID controllers. At the exit of the control device R is the manipulated variable MV in front. This is the desired valve position that is applied to the actuator of the valve. The parameters MV and PV be the diagnostic facility D fed. The facility D usually has a data memory and an evaluation unit. One or more data records with different process variables, here the setpoint, are stored in the data memory SW , the manipulated variable MV and the controlled variable PV , saved. The evaluation device calculates indicators for friction effects in connection with the monitored valve on the basis of the data records stored in the data memory.

In 3rd are examples of waveforms of the output signal of the controller MV and the measured value PV shown in connection with the valve position as a function of time. In the lower part of the 3rd is the signal curve for the controller output MV , that is, the setpoint of the valve position, which basically corresponds to the desired valve position, is plotted against time (here, sampling values). In the upper part of the 3rd is the associated process value PV which is measured at the output of the system or controlled system. The course of the measured value should be without friction effects PV that of the setpoint MV consequences. On the other hand, if friction effects occur, the curve shown is shown. The valve is stuck at the reversal points and suddenly changes to the other extreme position instead of constantly changing the valve position.

Numerous methods are known from the literature for detecting friction effects in valves.

Using these methods, indicators (also called stiction index, abbreviated SI) are calculated. Each of these indicators indicates how likely stiction / friction effects are to occur in a control circuit or on its control valve. Each method therefore delivers a value between 0 and 100. The respective indicator is determined by means of the desired valve position MV (Signal curve at the controller output) and a measured variable or the process value PV such as a valve position or a quantity directly related to the valve position. There are different limits for each of the methods, which define when friction effects are very likely, when they can be excluded and when an exact statement is not possible.

In the exemplary embodiment described here, 6 methods form the basis of the invention. They are briefly introduced below. For details regarding the individual methods, reference is made to the references mentioned above (Ref1 and Ref2).

Method A (Refl, p. 104f):

Here it is checked whether in the areas in which the position feedback SR (or the process value PV of the controller if there is no feedback) is constant, the manipulated variable MV changes strongly (areas I, II and III in 3rd ). For values of the SIA indicator greater than 0.75, friction effects are assumed. The evaluation limits are in 5 shown.

Method B (Refl, p. 105f):

The distance of the rising and falling path in the scatter plot ( 4th ) between controller output MV and valve position SR or PV characterized. For values of the indicator SIB friction effects are assumed to be greater than 0.75. The evaluation limits are in 5 shown.

Method C (Ref1, p. 106f):

Here is based on the changes in controller output MV and the valve position a qualitative statement about the shape of the scatter plot ( 4th ) of the two quantities, since there is a characteristic parallelogram for friction effects. For values of the indicator SIC friction effects are assumed to be greater than 0.3. The evaluation limits are in 5 shown.

Cross-correlation method (Refl, p. 117f):

Here it is checked whether the cross correlation between controller output MV and the valve position is point or axisymmetric (cf. 4th ). A cross correlation symmetrical to the origin is an indication of increased friction effects. For values of the indicator SICorr friction effects are assumed to be greater than 0.66. The evaluation limits are in 5 shown.

Curve fitting method (Refl, p. 149f):

Here it is checked whether the curve of the controller output MV a sawtooth shape and the curve shape of the valve position resembles a rectangular shape (cf. 4th ). For values of the indicator SICurve friction effects are assumed to be greater than 0.6. The evaluation limits are in 5 shown.

Ellipse adjustment method (Ref2):

Here the scatter plot ( 4th ) examined for an elliptical shape. For values of the indicator SIEllipse friction effects are assumed to be greater than 0.75. The evaluation limits are in 5 shown.

The methods presented are very specifically tailored to the different signs of stiction and are therefore not always meaningful. In many cases, not all methods will work with stiction, and vice versa, some methods will also tend to stiction with normal measurement noise. The formation of a simple average is not sufficient here. It is also not possible to weight the methods differently, since it strongly depends on the strength of the stiction and the quality of the data (noise, etc.), which method is the better. The individual results must therefore be brought together, which looks at the overall picture and draws the correct conclusions from it. In order to make these decisions correctly, countless test cases (with and without stiction) were considered.

In this exemplary embodiment, the invention combines the six individual methods for the analysis of stiction presented here in such a way that the analysis delivers the best possible result in every situation. Similarly good results can also be expected from a combination of 5 or 7 methods.

• Fall 1: Eine oder mehr Methoden zeigen stiction an (numStic ≥ 1):
  • Wenn mehr als eine Methode stiction zeigt (numStic ≥ 2), findet eine Mittelwertbildung aus den Top 3 der sortierten Liste statt. Bei drei oder mehr Methoden ist stiction sehr sicher anzunehmen. Bei nur zwei Methoden entscheidet der dritte Wert in der Liste, ob das Ergebnis auf stiction hinweist, oder das Ergebnis doch als unschlüssig gewertet wird.
  • Wenn nur eine Methode stiction zeigt (numStic = 1), wird die ersten Methode in der sortierten Liste ausgeschlossen und es findet eine Mittelwertbildung über alle anderen (Index 2 bis 6) statt. In diesem Fall ist stiction unwahrscheinlich, da nur eine Methode anschlägt. Dieser Ausreiser, kann ausgeschlossen werden.
• Fall 2: Keine Methode zeigt stiction an und eine oder mehr Methoden sind unschlüssig (numKA ≥ 1):
  • Wenn drei oder mehr Methoden unschlüssig sind (numKA ≥ 3), findet eine Mittelwertbildung über die Top 5 der sortierten Liste statt.
  • Wenn zwei Methoden unschlüssig sind (numKA = 2), findet eine Mittelwertbildung über alle Sticiton-Indizes statt. Das Ergebnis ist sehr schwammig. Es hängt von den Werten der Verfahren ab, welche stiction ausschließen. Sind diese nahe an der Grenze, kann das Ergebnis dennoch unschlüssig ausfallen.
  • Wenn nur eine Methode unschlüssig ist (numKA = 1), wird die erste Methode in der sortierten Liste ausgeschlossen und es findet eine Mittelwertbildung über alle anderen statt (Index 2 bis 6). Der Großteil schließt stiction aus. Die eine unschlüssige Methode kann vernachlässigt werden.
• Fall 3: Alle Methoden schließen stiction aus:
  • In diesem Fall findet eine Mittelwertbildung über alle Sticiton-Indizes statt. Das Ergebnis ist eindeutig.

For the method according to the invention for determining friction effects in continuous valves, the results of the individual methods (SIA, SIB , SIC , SICorr , SICurve , SIEllipse ) are first put on a common scale. The indicators SIA, SIB , SIC , SICorr , SICurve , SIEllipse are simplified here with SI1 to SI6 abbreviated (see also 1 and 2nd For this purpose, the results in this exemplary embodiment are standardized linearly to the same limits. After normalization, each indicator has SI1 to SI6 a lower threshold of 30% and an upper threshold of 70%. After normalization, the results are sorted in descending order and counted, how many of the indices show friction effects and how many are undecided. The counting result for the indices with friction effects is called numStic. The counting result for indices that are undecided is called numKA. A case distinction is now made on the basis of these values:

• Case 1 : One or more methods indicate stiction (numStic ≥ 1):
  • If more than one method shows stiction (numStic ≥ 2), the top is averaged 3rd the sorted list instead. With three or more methods, stiction is very safe to assume. If there are only two methods, the third value in the list decides whether the result indicates stiction or whether the result is considered to be undecided.
  • If only one method shows stiction (numStic = 1), the first method in the sorted list is excluded and averaging over all others (index 2nd to 6 ) instead of. In this case, stiction is unlikely because only one method works. This departure can be excluded.
• Case 2 : No method indicates stiction and one or more methods are undecided (numKA ≥ 1):
  • If three or more methods are undecided (numKA ≥ 3), the top is averaged 5 the sorted list instead.
  • If two methods are undecided (numKA = 2), averaging takes place across all sticitone indices. The result is very vague. It depends on the values of the procedures which exclude stiction. If these are close to the border, the result can still be undecided.
  • If only one method is undecided (numKA = 1), the first method in the sorted list is excluded and averaging takes place over all others (index 2nd to 6 ). The majority excludes stiction. The one undecided method can be neglected.
• Case 3 : All methods exclude stiction:
  • In this case, averaging takes place across all sticitone indices. The result is clear.

With this case distinction, a better statement can be made than with a simple mean over all results. As a result of the case distinction, a resulting indicator SIres is output, which indicates a probability that valve friction is present.

Depending on the number of combined procedures, the values used in the case distinction can be varied slightly.

The described method and a corresponding diagnostic device (cf. D2 in 1 , 2nd ) for determining friction effects in continuous valves according to the present application can be implemented for example by means of a process control system or advantageously in a software environment for cloud-based system monitoring. Such a software environment is represented, for example, by the data-based "Control Performance Analytics" remote service from Siemens AG. Data from a customer's system are collected with the help of software agents, aggregated and sent to a Siemens Service Operation Center, where they are stored on a Remote service computers can be saved. There, the data is evaluated using various "data analytics" software applications. The results of the diagnosis can be displayed on a monitor of the remote service computer and / or made available on a website so that they can be viewed by the end customer, ie the operator of the automated process engineering system, for example in a browser.

The method according to the invention and corresponding diagnostic device are thus preferably implemented in software or in a combination of software and hardware, so that the invention also relates to a computer program with program code instructions that can be executed by a computer for implementing the method on a suitable computing unit. In this context, the invention also relates to a computer program product, in particular a data carrier or a storage medium, with such a computer program that can be executed by a computer. Such a computer program can, for example, be held in a memory of a control system of an automated process engineering system or loaded into it, so that the method according to the invention is carried out automatically when the system is operating, or the computer program can be stored in a memory of a remote controller in a cloud-based environment. Service computer available or be loadable in this. In addition, the computer program can also be held in a computer connected to the control system and the remote service computer, which communicates with the other two systems via networks.

In the 6A and 6B two examples are shown which illustrate the advantages of the invention.

The two figures each show the non-standardized results of four individual methods in a bar chart. A scale is sketched to the left of each bar by which the 5 indicated limits for each individual method. The gray shading shows whether the method has found signs of stiction (dark gray), whether it is undecided (medium gray) or whether it excludes stiction (light gray).

In 6A The results of the six methods for stiction detection for data from a simulated control loop with stiction are shown in the actuator. Methods A and Ellipse in particular show high probabilities for stiction. The methods C and cross correlation on the other hand exclude stiction. Averaging all results would give a very low value here. According to the invention, this is the case 1 consider. Two methods display stiction and the resulting averaging of the top 3rd comes to the correct assessment that there is stiction in this case.

In 6B the results of the six methods for stiction detection for data from a control loop with high quantization at the measuring point are shown. In this case, averaging would provide too high a value. According to the invention, this is also the case 1 consider. The first method in the sorted list can be excluded. The invention hides the too high dropout and ultimately excludes stiction correctly.

QUOTES INCLUDE IN THE DESCRIPTION

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

Non-patent literature cited

• Choudhury, M .; Shah, S .; Thornhill, N .: "Detection and Quantification of control valve stiction", Department of chemical and material engineering, University of Alberta, Edmonton AB, Canada, 2006 [0005]

Claims (4)

Method for determining friction effects in continuous valves, in which at least one process value (PV) that is directly related to a valve position of the continuous valve and a desired value for the valve position (MV) are evaluated by first indicators (SI1 to SI6) for existing friction effects in the continuous valve, characterized in that a resultant indicator (SIres) is then calculated based on the previously calculated indicators (SI1 to SI6), which indicates a probability for the occurrence of friction effects in the continuous valve, whereby - the first indicators are standardized, - the results of the standardization are sorted in descending order, - the number of indicators showing friction effects, how many are undecided and how many friction effects are excluded, and then, based on the counting results, a decision tree Fallu A distinction is made according to which the resulting indicator (SIres) is determined and displayed. Diagnostic means for determining friction effects in proportional valves, wherein the diagnosis means is adapted, at least comprising at least one process value (PV), which is directly related to a valve position of the continuous valve in relation, and storing a desired value for the valve position (MV) in a record, by at least one evaluation device which is designed to perform the method according to Claim 1 perform. Computer program with program code instructions executable by a computer for implementing the method according to Claim 1 when the computer program is running on a computer. Computer program product, in particular data carrier or storage medium with a computer program executable according to Claim 3 .

Images