Linear Active Disturbance Rejection Control From the Practical Aspects

In this paper, important issues that could come up in practice of active disturbance rejection control (ADRC) due to a common assumption and some inevitable practical restrictions are investigated. The main idea behind ADRC is modeling effects of both internal uncertainties and external disturbances as an extra state called the total disturbance which is timely estimated. In the majority of recent works, it is assumed that time derivative of total disturbance is bounded. First we prove that, beside the system characteristics, the validity of this assumption strongly depends on the tuning parameters, even in the case of simple classes of linear second-order systems. Thus, one should be careful upon making this common assumption independent from the tuning parameters. Then, the effect of imperfections imposed by sensor dynamics, measurement noise, actuator saturation, and discrete implementation are thoroughly studied. It is shown that these imperfections could crucially affect the control performance. Furthermore, a tuning method, with a higher degree of freedom, is proposed involving the stability considerations in its design procedure. The given theoretical analyses are supported via both simulation and experimental results. The discussed issues are recommended to be carefully considered by experts and practitioners of ADRC.