The above-mentioned and other features and advantages of this invention, and the manner of attaining them, will become more apparent and the invention will be better understood by reference to the following description of an embodiment of the invention taken in conjunction with the accompanying drawing, wherein:
Corresponding reference characters indicate corresponding parts throughout the several views. The exemplification set out herein illustrates one embodiment of the invention, in one form, and such exemplification is not to be construed as limiting the scope of the invention in any manner.
The present invention will be described in more detail in the following text using an exemplary embodiment and with reference to the drawing.
The single FIGURE of the drawing shows a simplified schematic representation of an apparatus for controlling at least one quality feature of a running material web, in particular a fibrous web, during its production. This apparatus is assigned to a plant having a wet end and a paper machine (or coater, or calender) for producing a paper web or paperboard web. Hence in this case the apparatus is used, for example, to control at least one quality feature of a paper web or paperboard web.
The control apparatus includes a measurement system 10 for the repeat measurement of the respective quality feature on a running material web or paper web over the entire width of the material web. As is evident from
In addition the control apparatus includes an electronic control and/or evaluation unit 14 for determining the variability of the quality feature through a variance component analysis on the basis of the recorded profile measured values, of a measured value set including several consecutive CD profiles in the machine direction. The set having been recorded during a selectable previous time window.
The electronic control and/or analysis unit 14 is designed such that the control of the quality feature takes place according to the result of the variance component analysis.
The variability of the quality feature can be divided by way of the variance component analysis into a CD variability component in the cross direction (CD), an MD variability component in the machine direction (MD) and a residual variability component. Alternatively, or in addition, it is possible for confidence intervals for the CD profile and/or the MD profile to be calculated from determined variance values. The calculated variability components or confidence intervals are drawn on to control the quality feature.
The individual CD and MD profile measured values of a respective measured value set, which is to be subjected to a variance component analysis, can be arranged by way of the control and/or analysis unit in a matrix in whose lines or columns there are CD profile measured values respectively and in whose columns or lines there are MD profile measured values respectively. The profile measured values, contained in a respective measured value set, which is to be subjected to variance component analysis, can be variously weighted in respect of their contribution to the variability of the quality feature to be determined, at least partially according to how far back they lie in time as compared to the actual moment of determining the variability. In this case, profile measured values lying relatively far back can be weighted in particular relatively more lowly.
The variance component analysis can be repeated when a selectable number of new profile measured values are measured by way of measurement system 10. The result of a respective variance component analysis can be drawn on in order to change accordingly the behavior of at least one control system provided to control the quality feature. For example, at least one control parameter and/or at least one filter parameter of at least one control system, provided to control the quality feature, can be variably adjustable according to the result of a respective variance component analysis.
The variance component analysis can be based, for example, on at least one of the following algorithms: TAPPI TIP 1101, TAPPI T585, DAHLIN Exact Variance Analysis (see the literature cited above). The Exact Variance Analysis is also described below in more detail.
The division into a CD variability component, an MD variability component and a residual variability component must be sufficiently exact in order to be able to draw the correct conclusions for the control in question. An underestimated variable in the assessment of a respective paper quality feature is the residual profile, also referred to simply as residual. The residual fraction of the web deviation is defined in that it cannot be assigned to either the machine direction (MD) or the cross direction (CD). Hence it characterizes random quality faults. The residual fraction is perceived by the paper producer as “live” cross profiles. The cross profile is poor and changes constantly; the peaks in the cross profile are not fixed but sometimes become inverted.
Often the paper producer concludes that the cross profile is poor, which is misleading. The fact is this type of fault cannot be corrected even by an improved cross profile control, for example with more actuating elements.
The purpose of the variance component analysis is to put the relationships into perspective and to clearly distinguish which fraction of the displayed profile faults has its origin in stable cross profile deviations (CD), which fraction comes from temporary production faults (MD), and how large the random fraction of the deviations is.
The variability of the paper web arises therefore from the superimposition of three causes:
It has been shown that these three causes of quality deviations can be considered as statistically independent of each other, which is an important precondition for the variance component analysis.
The basic idea behind the Exact Variance Analysis, for example, is to find an algorithm which correctly determines the amplitudes of the three mentioned causes from the measured values of the paper web. The algorithm for this Exact Variance Analysis is presented below. It meets the following requirements:
The relations in question for the Exact Variance Analysis are presented in the following:
Equation letters:
X: matrix with measured data (each gap is a cross profile)
M, N: number of measured data in the MD and CD direction
i, k counter for the summation
Mean MD profile and CD profile:
Mean value of all data values:
Intermediate variables for calculating the CD and MD variance:
Variance fractions (result of the variance analysis):
Knowledge of the variance scatter can be used to specify confidence intervals of the variance analysis. The smaller the random sample, the greater the statistical uncertainty of the analysis and the larger the confidence intervals.
The specification of confidence intervals helps to relativize the value of the statistical statements.
For a control system it can be expedient to calculate the variance fractions on the basis of a relatively small number of profiles. In conjunction with the confidence intervals it is thus possible to distinguish random profile fluctuations from controllable fluctuations with a real cause.
It has been shown that the variability of the paper web can be assessed far more informatively if all the individual measured data of the paper web, in a previous time window, are assessed simultaneously. For the example mentioned at the beginning it is possible to assess the MD variability of the paper web in the last 100 seconds on the basis of 500 individual measured values instead of typically 20 MD values as was commonly used hitherto. The fast and precise knowledge of the variability permits a fast change of the corresponding parameters of the controls and filterings if the statistical properties of the paper web change. Faults of the production process due to overly severe control interventions are reduced to a minimum. Such faults can arise when, because of random or non-controllable faults, the control responds severely nevertheless and thus carries out unnecessary process interventions.
The profile measured values contained in a respective measured value set, which is to be subjected to variance component analysis, can be variously weighted, in respect of their contribution to the variability of the quality feature to be determined at least partially according to how far back they lie in time compared to the actual moment of determining the variability. In this case, profile measured values lying relatively far back can be weighted to a lesser degree.
The variance component analysis can be repeated when a selectable number of new profile measured values was measured by way of the measurement system.
The result of a respective variance component analysis can be drawn on in order to change accordingly the behavior of at least one control system provided to control the quality feature. For example, at least one control parameter and/or at least one filter parameter of at least one control system provided to control the quality feature can be variably adjustable according to the result of a respective variance analysis.
The variance component analysis can be based not only on the Exact Variance Analysis just described, but also on at least one of the following algorithms: TAPPI TIP 1101, TAPPI T585, DAHLIN Exact Variance Analysis.
With the algorithm used as a basis for the variance component analysis it is possible to calculate short-term deviations, which can also be roughly interpreted as residual components.
The CD variability component and the MD variability component can each be divided into a controllable fraction and a non-controllable fraction.
A relatively large residual variability component, compared to the current CD and MD variability components or controllable CD and MD fractions, can be used by way of the control and/or analysis unit 14 in particular such that at least one control system provided to control the quality feature acts relatively more slowly and/or relatively more weakly on the production process.
In principle it is also possible for a relatively large residual variability component, compared to the current CD and MD variability components or controllable CD and MD fractions, to be taken into account, such that the measured data are filtered relatively more intensively prior to calculation of the respective set-point variable for at least one control system provided to control the quality features.
As already mentioned it is possible for confidence intervals for the CD profile and/or the MD profile to be calculated by way of the control and/or analysis device 14 from determined variance values and, in the light of these confidence intervals, a distinction can be drawn between a respective profile deviation based on an existing cause and a deviation based, at least essentially, only on a residual variability component. At least one control system provided to control the quality feature is then activated by way of the control and/or analysis device 14 to perform an intervention in the production process if it was established, in the light of at least one confidence interval, that the deviation in question is based on a really existing cause.
It is also conceivable, in particular for at least one control system provided to control the quality feature to be actuated by way of control and/or analysis device 14 to intervene relatively more weakly or not at all in the production process, if it was established, in the light of at least one confidence interval, that the deviation in question is based, at least essentially, only on a residual variability component.
The calculated variability components and/or confidence intervals can also be taken into account in the assessment of the quality of a process model which serves the quality control and is saved in control and/or analysis device 14 as software. Any deviation of the default values of the process model from the recorded profile measured values is determined by way of control and/or analysis device 14 and any determined deviation on a relatively high variability of the quality features and/or a relatively low confidence is weighted relatively more lowly and/or is filtered relatively more intensively prior to further use.
The calculated variability components and/or confidence intervals can also be taken into account, in particular when there is a change of the process model, which serves the quality control, and is saved in control and/or analysis device 14 as data and/or software. The calculated variability components and/or confidence intervals can also be taken into account, for example, through adaptation parameters.
It is also conceivable in particular for the process model to be corrected by way of control and analysis device 14 relatively more slowly to the process measured in particular by way of the profile measured values in the event of a relatively higher variability of the quality feature and/or a relatively lower confidence.
While this invention has been described with respect to at least one embodiment, the present invention can be further modified within the spirit and scope of this disclosure. This application is therefore intended to cover any variations, uses, or adaptations of the invention using its general principles. Further, this application is intended to cover such departures from the present disclosure as come within known or customary practice in the art to which this invention pertains and which fall within the limits of the appended claims.
Number | Date | Country | Kind |
---|---|---|---|
10 2006 045 786.2 | Sep 2006 | DE | national |