Data processing: generic control systems or specific application – Generic control system – apparatus or process – Optimization or adaptive control
Reexamination Certificate
2001-03-22
2003-07-22
Voeltz, Emanuel Todd (Department: 2121)
Data processing: generic control systems or specific application
Generic control system, apparatus or process
Optimization or adaptive control
C700S030000, C700S031000, C700S037000, C700S044000, C700S052000, C703S002000, C702S082000, C318S561000
Reexamination Certificate
active
06597958
ABSTRACT:
1. FIELD OF THE INVENTION
This invention relates to industrial process control systems and more particularly to the measurement of the control provided by such systems to the process.
2. DESCRIPTION OF THE PRIOR ART
Users of industrial process control systems are concerned with how well their control is performing. That concern may be framed as an inquiry asking if the control can be improved; or how well is the process doing; or how often should the control loops be retuned; or is there a number which tells the user how the system is performing!
The process control industry tries to answer these questions by calculating the standard deviation of the signal being controlled. It is thought that if the standard deviation is good or bad, then control must be good or bad. The standard deviation is a good measure of process performance, but is not a good measure of control performance. For example a high standard deviation could be the result of either a poorly tuned control loop or load disturbances that occur at frequencies beyond the capability of the controller.
Another drawback of using standard deviation as a measure of the effectiveness of the control is that it does not tell the user of the process control system how good the system could be. Many times a control system may be optimally tuned, but the standard deviation is outside of the user's product specifications. The user then spends much time trying to retune an already perfectly tuned controller. The user has no tool that allows the user to make an informed decision on whether to spend money on process control re-tuning, new algorithms, or process changes.
In the past, control performance was measured by comparing standard deviations of several hours of data collected during a “blind” run and an “on control” run. The blind run is a running of the process with the control system off and no operator intervention. The on control run is taken with all the controls in the automatic mode. The amount that the on control standard deviation is better than the off control standard deviation is an indication of how the control is doing.
Aside from the problems with standard deviation already mentioned, there are several additional problems with using the standard deviation method to measure control effectiveness. The standard deviation method shows how the system performs as compared to no control. Users rarely run their process control systems in the “no control” mode. Therefore this control performance measure has very little meaning. Also, a user will rarely allow its control system to be turned off for several hours. As a result, the “no control” mode performance measure is usually not done. Even when this performance measure is done, the disturbances that impact the “no control” run must be the same as those that impact the on control run. In the real world this is never the case.
A recent attempt to solve the problem of determining the quality of control performance is by comparing current control performance with that of an ideally tuned controller. Methods to accomplish this comparison are the well known Minimum Variance and Harris Index. These methods work fairly well for systems with limited amounts of process delay. However, systems with process delay cause problems with these methods as in such systems the methods result in forecasts that can not be achieved. The methods indicate that the control system should be able to perform better than what is possible. These methods are good for identifying the limit, but they do not identify what is the best that the specific control system can perform.
As a result, users have relied on the experience of the control engineer to tell the user when the control system is tuned as good as it can be. This usually results in user confusion. A new control engineer will visit the plant with more experience than the last control engineer that has visited the plant and the new control engineer re-tunes the loops. The results may be better than the tuning produced by the last engineer.
The user enters a pattern where the user looks for more and more experienced control engineers. The user conclusion is that control performance can always be made better with the increased experience of the control engineer. The user is never convinced that its control system is setup and tuned as good as it can be, because the user is never convinced that the control engineer is the best there is. This pattern may become dangerous if the next control engineer is not as experienced as the previous control engineer. The user pattern will continue as long as there is not a good measure of control and process performance.
The present invention answers the user questions about control and control performance in such a way that control and process performance is easily measured. The present invention allows the user to make informed decisions on when to retune a controller, when an algorithm needs to be replaced, or when the process needs to be changed.
SUMMARY OF THE INVENTION
A method for measuring the control provided by a control system to a process. The system has a controller for controlling the position of a final control element to control a process variable. The method has the step of determining process model parameters for a simple model and a high order model of the process. The method also has the steps gathering the value of the process variable and the final control element position; predicting off control data using the determined simple and high order models parameters and the gathered value of the process variable and the final control element position; determining the optimal tuning using the determined simple and high order models parameters; and forecasting the optimal process performance from the predicted off control data and the determined optimal tuning.
A method for measuring the control provided by a control system to a process. The system has a controller for controlling the position of a final control element to control a process variable. The method has the step of performing a bump test on the process to determine process model parameters for a simple model and a high order model of the process. The method also has the, steps of gathering the value of the process variable and the final control element position; predicting off control data using the simple and high order models parameters determined by performing the bump test and the gathered value of the process variable and the final control element position; determining the optimal tuning using the simple and high order models parameters determined by performing the bump test; and forecasting the optimal process performance from the predicted off control data and the determined optimal tuning.
A method for measuring the control provided by a control system to a process. The system has a controller for controlling the position of a final control element to control a process variable. The method has the step of determining process model parameters for a simple model and a high order model of the process. The method also has the steps of predicting off control data using the determined simple and high order models parameters and the value of the process variable and the final control element position; determining the optimal tuning using the determined simple and high order models parameters; and forecasting the optimal process performance from the predicted off control data and the determined optimal tuning.
REFERENCES:
patent: 4922412 (1990-05-01), Lane et al.
patent: 5257206 (1993-10-01), Hanson
patent: 5282130 (1994-01-01), Molnar
patent: 5519605 (1996-05-01), Cawlfield
patent: 5522224 (1996-06-01), Canney
patent: 5684375 (1997-11-01), Chaffee et al.
patent: 6253113 (2001-06-01), Lu
patent: 6546295 (2003-04-01), Pyotsia et al.
ABB Automation Inc.
Barnes Crystal J.
Rickin Michael M.
Todd Voeltz Emanuel
LandOfFree
Method for measuring the control performance provided by an... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for measuring the control performance provided by an..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for measuring the control performance provided by an... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3061437