Adaptive controller tracks process changes

Vance J. VanDoren, Ph.D., P.E. -- Control Engineering, 2/1/1999

In some sense, all process controllers are "adaptive" because they force a process to adapt its outputs to the values operators desire. However, most process controllers do so according to algorithms designed (or at least fine tuned) by the operators before the controller ever starts its work. The operators may periodically re-tune parameters of a traditional controller, but this is generally a manual operation performed only after the controller's performance has begun to deteriorate for some reason.

A truly adaptive controller can update its tuning parameters while it is in operation so that it's performance remains optimal, even if the behavior of the process changes. For example, an adaptive controller that is initially tuned to provide aggressive control for a sluggish process will substitute more conservative tuning parameters if it detects the process has somehow become more responsive to control efforts. A traditional controller with fixed tuning parameters would continue to control the process aggressively, causing the process outputs to excessively fluctuate.

QuickStudy 2.0 from Adaptive Resources (Pittsburgh, Pa.) is a PC-based control system that includes adaptive control functions. Using QuickStudy's graphical programming tool, users draw their control strategies in block diagram format specifying the flow of data from block to block by drawing lines between them. Adaptive control is one of the available function blocks. QuickStudy also includes all of the necessary hooks for connecting the adaptive controller to any OPC- or DDE-compliant process interface such as a DCS or PLC.

Like most adaptive controllers, QuickStudy bases its control decisions on an empirical model of the process. It constantly collects process data and computes a mathematical relationship that describes how the outputs have responded to the inputs over time. Assuming that the same input/output relationships will remain valid in the near future, QuickStudy then predicts where the process is heading and determines the control efforts required to steer it in the right direction.

The biggest problem I've encountered with such online modeling techniques is collecting sufficient information about the behavior of the process, while the controller is trying diligently to prevent the process output from moving at all. Many adaptive control schemes require the controller to periodically stop and apply a test input just to see if the process responds differently than it did before. In contrast, QuickStudy uses a statistical modeling technique that is designed to eliminate the need for such artificial process stimulation. It makes do with just the historical data available from normal closed-loop operations.

I also found QuickStudy's ability to constrain its control efforts as well as the process outputs interesting. Not only can operators specify the allowable ranges for the process inputs and outputs, but they can specify how hard the controller must work to prevent violations of each individual constraint. Unfortunately, this unique approach requires a bit of not-so-quick study to use properly. And, in spite of its self-adapting features, QuickStudy can't model the process without a few hints from an operator familiar with the process. Fortunately, there are simpler configurations of the QuickStudy controller available for novice users, and Adaptive Resources offers training, installation, and commissioning services to help configure QuickStudy's more sophisticated features.

QuickStudy 2.0 and FactorySoft Control run under Microsoft Windows NT 4.0 on a 200 MHz Intel Pentium PC or better. A minimum of 20 to 30 Mbytes hard disk space and 64 Mbytes of RAM are recommended. QuickStudy's price ranges from $16,000 to $38,000 depending on application complexity.

For more information on QuickStudy, visit www.controleng.com/info.