Controller Performance Monitoring at Constant Set Point using Autocorrelation

R.R. Howard and D.J. Cooper (USA)


Process control, controller performance assessment, disturbance rejection, PID tuning


Efficient disturbance rejection is often the primary goal of a controller, but few industrially relevant methods exist for characterizing controller performance for disturbance rejection when the set point remains fixed. This work uses the patterns in the autocorrelation function (ACF) of a closed-loop data set with a stationary set point to directly relate controller performance. The method is advantageous because it does not require deliberate disruption to the process. Rather, it uses raw regulatory data normally found in a plant historian. Thus, it can be incorporated into a performance assessment schedule. Plots of ACF allow industrial practitioners to directly apply intuition built from analyzing disturbance rejection trends. Commonly used controller assessment pattern features such as overshoot and settling time can be measured and used as benchmarks. A flow diagram and tuning map method provide the basis for determining controller performance and retuning. This work presents an analytical study of the ACF and its implementation in simulated examples. Process characteristics such as noise level and sampling rate are included in the studies, and recommendations are made for the best practice use of ACF.

Important Links:

Go Back