An example of self-tuning controllers on distributed system

Deals with a theoretically known, but in practice seldom used procedure for nonparametric model identification. The essential point of the method is in applying the pseudo random binary signal and evaluating the crosscorrelation function between input and output signals which results in the estimated impulse response of the process. The tuning procedure is applied on the distributed system. The method was improved by applying the signals twice in succession, which resulted in all signals becoming periodic. It turned out that the period of 127 (seven bit PRES register) is sufficient for practical application. Automatic detection of inappropriate frequency range enables the procedure to be applied as a self-tuning method for PID controllers. The procedure was applied as a self-tuning method for PID controller tuning by Chien Hrones Reswick rules. Due to these rules the method is applicable to proportional overdamped processes only. Implementation of procedure on the distributed system has advantage, that we have more self-tuning controllers on the same hardware and time needed for parameters estimating is shorter. Implementation of the method in IDR 052 software for Mitsubishi Programmable Logic Controllers, SCADA program Factory Link from USDATA and application to a hydraulic plant are described.