Comparison of Different Model-Free Control Methods Concerning Real-Time Benchmark

The design of accurate model often appears as the most challenging tasks for control engineers especially focusing to the control of nonlinear system with unknown parameters or effects to be identified in parallel. For this reason, development of model-free control methods is of increasing importance. The class of model-free control approaches is defined by the nonuse of any knowledge about the underlying structure and/or related parameters of the dynamical system. Therefore, the major criteria to evaluate model-free control performance are aspects regarding robustness against unknown inputs and disturbances and related achievable tracking performance. In this contribution, a detailed comparison of three different model-free control methods (intelligent proportional-integral-derivative (iPID) using second-order sliding differentiator and two variations of model-free adaptive control (using modified compact form dynamic linearization (CFDL) as well as modified partial form) is given. Using a three-tank system benchmark, the experimental results are validated concerning the performance behavior. The results obtained demonstrate the effectiveness of the methods introduced.