Half-gain observer tuning for noise reduction in discrete-time ADRC

Abstract

Both the ability to suppress disturbances and the simplicity of plant modeling within the active disturbance rejection control (ADRC) approach are enabled by its observer and largely dependent on its sufficiently fast tuning. This, however, may require high observer gain values, which increase the controller’s susceptibility to measurement noise. To reduce the noise sensitivity without requiring any change to the controller structure, this article transfers the results of a continuous-time method called half-gain tuning to the discrete-time domain. Applied only to ADRC’s observer, the closed-loop dynamics will remain almost unaffected. Explicit tuning equations for the discrete-time observer gains are derived. A detailed examination performed analytically, in simulation, and in experiment reveals how much of the theoretical noise reduction promised by the continuous-time method can still be achieved in the discrete-time domain. In summary, an observer tuning method is presented that delivers a substantial reduction in noise sensitivity in practically relevant scenarios and can be applied minimally invasively to existing ADRC control loops.