The study of digital PID autotuning procedure

Przedstawiono procedurę autotuningu altorytmu PID zaimplementowaną w przemysłowym regulatorze cyfrowym. Procedura składa się z dwóch etapów: identyfikacji obiektu regulacji oraz wyliczenia nastaw regulatora PID z jednego z dostępnych kryteriów. W badaniach skupiono się na etapie identyfikacji obiektu regulacji, który można przeprowadzić za pomocą metody Zieglera-Nicholsa i Cohena-Coona. Badania wykonano w warunkach laboratoryjnych co umożliwiło określenie powtarzalności wyników.The procedure of PID controller autotuning is presented in the paper. It consists of two steps: identification of the control object and calculation of PID controller settings with one of the available criteria. A user has the possibility to interfere in the procedure of autotuning. By the definition value of the specified control register, the user selects the identification method, determines the value of stabilization time and the stabilization deviation as well as defines the criterion of controller settings. The stabilization of deviation and the time of stabilization determine the steady state of the object, and so the duration of the autotuning procedure. This paper is focused on the identification process of the object control, which can be realized by the Ziegler-Nichols and Cohen-Coon methods. Investigations were performed in a laboratory for different combinations of the deviations of stabilization and the stabilization times. This allowed determining the repeatability of the autotuning procedure results. These results show that the identification process accuracy of the object of control depends on the value of the deviation of stabilization and the stabilization time defined by the user. Regardless of the identification method, the value of the deviation of stabilization should be as small as possible and should be on 1% level. In case of identification of the control object by the Ziegler-Nichols method, the minimum value of the stabilization time should be about 25% of the time constant of the control object. In case of identification of the control object by the Cohen-Coon method, the minimum value of the stabilization time should be 45% value of the time constant of the control object. Use of both methods of the control object identification gave reproducible results, which demonstrates the stability of the autotuning procedure