Although conventional two-dimensional model predictive iterative learning control (2D-MPILC) based on an extended non-minimum state space (ENMSS) model can avoid designing an observer, it only relies on feedback to pa...
详细信息
Although conventional two-dimensional model predictive iterative learning control (2D-MPILC) based on an extended non-minimum state space (ENMSS) model can avoid designing an observer, it only relies on feedback to passively deal with time delay. This passive treatment for time delay hinders the further improvement of control performance. To address this shortcoming, a two-dimensional model predictive iterative learning control strategy based on the set point learning (2D-SPL-MPILC) is proposed. Firstly, a set point learning strategy is developed to improve the ability to deal with time delay. Based on the error of the previous batch, the set point learning strategy perceives the system dynamics in advance, and then outputs this advance perception in the form of the predictive set point. Secondly, a novel ENMSS model is constructed on the basis of the predictive tracking error between the predictive set point and the actual output. Since the predictive tracking error integrates the predictive set point, this novel ENMSS model has certain predictive ability. Finally, based on this novel ENMSS model, a novel two-dimensional model predictive iterative learning control (2D-MPILC) method, which is called 2D-SPL-MPILC method, is designed. Because this controller contains the perception of future process dynamics, it can reduce the impact of time delay and improve the control performance. The case studies on the packing pressure control in injection molding process and batch reactors demonstrate the effectiveness of the presented 2D-SPL-MPILC strategy.
A new data-driven predictive iterative learning control(ILC) is proposed for same category discrete nonlinear systems in this work. The controller design only depends on the input/output data of the system and does no...
详细信息
ISBN:
(纸本)9781509054626
A new data-driven predictive iterative learning control(ILC) is proposed for same category discrete nonlinear systems in this work. The controller design only depends on the input/output data of the system and does not need explicit mathematical model. More prediction information along the iteration axis is utilized in the learningcontrol law to improve the control performance. The applicability of the proposed methods is proved by simulation experiments.
This paper proposes a data-driven predictive iterative learning control (DDPILC) for nonlinear nonaffine systems. First, we develop an iterativepredictive model (IPM) where an iterative dynamic linearisation techniqu...
详细信息
This paper proposes a data-driven predictive iterative learning control (DDPILC) for nonlinear nonaffine systems. First, we develop an iterativepredictive model (IPM) where an iterative dynamic linearisation technique is introduced for addressing the strong nonlinearities and nonaffine structure. Then, an auto-regressive model is employed to estimate the unavailable parameter of IPM along with the iteration direction. Next, the outputs in the future iterations are predicted pointwisely over the finite operation interval that are further incorporated into the optimal objective function to obtain the optimal control input sequence. In addition, a constrained-DDPILC is extended for systems with I/O constraints which are reformulated as a linear matrix inequality (LMI). The two proposed methods do not have model requirement except for I/O data. Simulation study verifies the results.
In this paper, predictivecontrol has been implemented by several methods in a most simplified manner for quadruple tank system, i.e., a multi-input multi-output control problem. Quadruple tank system is a nonlinear s...
详细信息
In this paper, predictivecontrol has been implemented by several methods in a most simplified manner for quadruple tank system, i.e., a multi-input multi-output control problem. Quadruple tank system is a nonlinear system which can be adjusted to obtain two different class of system i.e., minimum and nonminimum phase, therefore tuning for each phase of system has to be tackled separately. control system design has been obtained by manual proportional-integral controller, traditional generalized predictivecontrol, predictive proportional-integral-derivative on the basis of generalized predictivecontrol, predictive iterative learning control and finally adaptive predictive proportional-integral-derivative control has been developed successfully. a Stability and performance of the each control system design has also been discussed.
暂无评论