Robust evolving cloud-based PID control adjusted by gradient learning method

In this paper an improved robust evolving cloud-based controller (RECCo) for a class of nonlinear processes is introduced. The controller is based on parameter-free premise (IF) part. The consequence in this case is given in the form of PID-type controller. The three adjustable parameters of PID controller are updated on-line with a stable adaptation mechanism based on Lyapunov approach such that the output of the process tracks the desired model-reference trajectory. The proposed algorithm has also ability to add new rules or new clouds when this is necessary to improve the whole behaviour of the controlled process. This means that RECCo controller evolves the control structure and adjusts at the same time the parameters of the controller in an on-line manner, while performing the control of the plant. This approach is an example of almost parameter-free approach, because it does not use any off-line pre-training nor the explicit model of the plant and requires almost no parameter tuning. The proposed algorithm is tested on an artificial nonlinear first-order process and on a simulated hydraulic plant.