(369t) Relay Feedback Identification for Processes Under Drift and Noisy Environments

Lee, J., Kyungpook National University
Sung, S. W., Kyungpook National University
Edgar, T. F., Dept. of Chem. Eng.,The University of Texas at Austin

Outputs of industrial processes are often contaminated with high frequency noises and slow drifts. For a process with noise the conventional relay feedback method suffers from chattering. Chattering should be avoided for mechanical actuators and, for this, a relay with hysteresis can be used. However, its switching time fluctuates and average switching period is not the ultimate period. For a process with drift, the conventional relay feedback method shows asymmetric oscillations, resulting poor estimations of ultimate data. For a constant bias, asymmetric oscillations can be made to be symmetric by adjusting the input or output bias iteratively. Here utilizing various filters, simpler relay feedback methods have been proposed. Problems due to high frequency noises and slow drifts have been resolved. In addition, the proposed methods can reduce estimation errors due to high order harmonic terms in relay feedback oscillations. With one more information about the process in addition to relay oscillation information, a first order plus time delay (FOPTD) model can be obtained. For this, the process steady state gain which can be estimated with a biased relay or integrals of relay transients is usually used. Here a simpler method to estimate the process steady state directly and the ultimate information accurately is proposed. A band pass filter and a hysteresis relay are used. The band pass filter cuts both slow drifts and high frequency noises. Utilizing the ability for the band pass filter to cut the zero frequency disturbance, a method to obtain the process steady state gain is devised.