Multi-variable control loop assessment

Details Multi-variable control loop assessment Multi-variable control loop assessment

2001-03-23

2003-11-18

Follansbee, John (Department: 2121)

Data processing: generic control systems or specific application

Generic control system, apparatus or process

Optimization or adaptive control

C700S032000, C700S044000, C700S045000, C700S047000, C700S051000, C700S029000

Reexamination Certificate

active

06650947

ABSTRACT:

FIELD OF THE INVENTION
This invention relates, in general to control systems, and more particularly to a multi-variable control loop assessment method and apparatus for analyzing, assessing and trouble shooting control loops in complex industrial processes such as papermaking operations.
BACKGROUND OF THE INVENTION
In a large industrial process environment, such as a paper mill, there are hundreds or even thousands of process control loops operating within highly automated pieces of equipment. Many of these loops interact with each other and influence each other's performance. Paper quality and production efficiencies depend on the performance of these loops and suffer severely if the control loops are not optimized.
It is known in the prior art to compile a list of relevant variables of interest and then plot and trend the variables for visual examination in order to trouble-shoot process and/or control problems and to identify sources of variations. Knowledgeable engineers may then examine the cross-correlation between pairs of variables, auto-correlation, power spectrum or similar classical time series analysis functions on the variables. However, the selection of trends for observation is based on the engineer's process knowledge, which is a manual, very time-consuming and subjective process. Another drawback of using cross-correlation is that normally only the most dominant component is visually observable. Often the most dominant influence of a loop is its control signal. Interactions from other variables may have weaker influences on the loop, thus making such interactions difficult to determine using such prior art cross-correlation methods.
Automatic process and control diagnostic tools have been developed for assessing and troubleshooting large numbers of control loops and thereby overcoming some of the disadvantages inherent in prior art manual observation and cross-correlation methodologies. Such tools incorporate automatic data logging and data mining functions for collecting and storing measurements from control loops and other key process measurements. The estimation of control loop capability is normally based on a single-variable approach by examining the loop output frequency content. The variability improvement capacity is then related to the observation of the low frequency content in the power spectrum of the control loop output. There exists in the prior art no suitable measure on the minimum achievable loop output variation, and no method to predict the influence of interactive loops.
Professor Thomas Harris of Queen's University, Canada, published an article on control loop assessment in
The Canadian Journal of Chemical Engineering,
Vol. 67, October 1989, in which he proposed a method of calculating an index for a single-variable loop. This index is now known as the Harris index. Industrial applications of the Harris index began appearing in 1992, as reported in such articles as “Towards mill-wide evaluation of control loop performance” by M. Perrier and A. Roche in
Control Systems
'92 conference, and “An expert system for control loop analysis” by P. Jofriet, et. al. in
CPPA annual meeting,
1995. Successful applications in paper mills are limited to lowest level loops due to the fact that the method is based on the single-variable approach.
It has been recognized in the prior art that control loop performance assessment based on a single variable approach provides an erroneous loop performance index due to cyclical perturbations. Because of the interactions between multiple control loops, only a correctly implemented multivariable analysis is capable of revealing the true process and control information, and is thereby suitable for assessment and troubleshooting purposes.
However, the direct extension of Harris's method to a multivariable process gives rise to practical difficulties that require extensive process modeling. An alternative approach to determining signal correlation involves dividing potentially malfunctioning loops with approximately coincident spectral peaks into possible interacting classes. Loop interaction is accounted for by the calculation of a “modified” index, as set forth in U.S. Pat. No. 5,838,561, entitled “Automatic Control Loop Monitoring and Diagnostics” by James Owen. However, the approach advocated by Owen is only valid when interacting loops have common primary or secondary frequencies. This normally requires that the loop is clearly oscillating. Also, only the most dominant component may be determined with this technique such that weaker correlation is ignored.
SUMMARY OF THE INVENTION
According to the present invention, a multivariable analysis tool is provided wherein an orthogonal decomposition method such as the Partial Least Squares algorithm is applied to a disturbance model relating the known loop disturbances to the loop model residue. The tool according to the invention first extracts the most dominant correlation to the loop model residue and then uses the residue to search for secondary dominant correlation in an orthogonal space. This process is repeated until no further output variation can be significantly attributed by the next dominant correlation.
In this way, the analysis tool of the present invention is able to estimate the performance potential of each control loop under different disturbance conditions and provide a control performance index by comparing the achieved performance to the performance potential in a multi-variable environment. This index indicates whether or not a given loop is under optimal operation and, in fact, shows the variance of the loop from the best achievable loop performance.
The analysis tool also predicts potential control improvement when any control solution is used to reduce a known disturbance variation (for example, feed-forward control or advanced controls such as multi-variable control using MPC techniques, etc.). This prediction can then be used to decide and justify the use of specific control solutions.
Importantly, the analysis tool of the present invention identifies sources of process variations through the disturbance model parameters associated with the contribution of each known disturbance signal to the loop model residue through latent variables. This information is then used to troubleshoot the process, measurement and/or control functions or malfunctions in a multi-variable process environment.
In the disturbance model, known disturbance signals with little contribution towards the loop model residue could be dropped out. Those disturbances are identified by the insignificance of their coefficients associated with the disturbance model output. Further, looking for signals greatly correlated with the disturbance model residue and adding them in the disturbance model as known disturbances will expand a user's knowledge of process. Such added loop disturbances may have great influence on the loop performance and once their variations are eliminated, a significant reduction on the minimum achievable loop variation could be realized.
Using the novel ideas of this invention and the control loop assessment functions, the analysis tool of the present invention has clear and useful application for process engineers, project engineers, service engineers, trouble-shooting personnel and sales engineers in paper making mills. Moreover, this invention applies more generally to loop monitoring and process insight/troubleshooting products that may be useful or desirable in other process industries.


REFERENCES:
patent: 5442562 (1995-08-01), Hopkins et al.
patent: 5489829 (1996-02-01), Umida
patent: 5568377 (1996-10-01), Seem et al.
patent: 5877954 (1999-03-01), Klimasauskas et al.
patent: 6110214 (2000-08-01), Klimasauskas
Harris, Thomas J, “Assessment of Control Loop Performance” The Canadian Journal of Chemical Engineering, vol. 67, Oct., 1989.
Jofriet, et al. “An Expert System for Control Loop Performance Analysis,” 81stAnnual Meeting, Technical Section, CPPA, Feb. 2-3, 1995.
Perrier, et al., “Towards Mill-Wide Evaluation of Con

Affiliated with

Also associated with

Rating

Multi-variable control loop assessment does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Multi-variable control loop assessment, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Multi-variable control loop assessment will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3172378