Idea Of Iterative Learning Control Research Articles

Adaptive PI control of ultrasonic motor using iterative learning methods

PI (proportional integral) controller is widely utilized in ultrasonic motor control systems. However, PI controllers which use fixed parameters is often difficult to ensure that the control performance meets the expectations. Based on the idea of iterative learning control (ILC), a new online adaptive adjustment method for PI control parameters is proposed. Simple P (proportional) type and PD (proportional differential) type iterative learning controllers are designed to adjust the proportional and integral coefficients of PI controller online. Using this method, the coefficients of PI controller are adaptively adjusted with the change of motor characteristics to ensure the control performance meets expectations. This method can improve the nonlinear expression ability of the controller, so that its input–output relationship can better match the characteristics of the controlled object, thereby improving the control performance. Results of comparative experiments indicate the validity of the proposed control strategy.

Two‐loop reinforcement learning algorithm for finite‐horizon optimal control of continuous‐time affine nonlinear systems

AbstractThis article proposes three novel time‐varying policy iteration algorithms for finite‐horizon optimal control problem of continuous‐time affine nonlinear systems. We first propose a model‐based time‐varying policy iteration algorithm. The method considers time‐varying solutions to the Hamiltonian–Jacobi–Bellman equation for finite‐horizon optimal control. Based on this algorithm, value function approximation is applied to the Bellman equation by establishing neural networks with time‐varying weights. A novel update law for time‐varying weights is put forward based on the idea of iterative learning control, which obtains optimal solutions more efficiently compared to previous works. Considering that system models may be unknown in real applications, we propose a partially model‐free time‐varying policy iteration algorithm that applies integral reinforcement learning to acquiring the time‐varying value function. Moreover, analysis of convergence, stability, and optimality is provided for every algorithm. Finally, simulations for different cases are given to verify the convenience and effectiveness of the proposed algorithms.

A future concern of iterative learning control : A survey

The main idea of iterative learning control (ILC) was exposed when Arimoto’s first paper was published. Industrial tasks mainly in repetition are controlled by an iterative teaching controller in both iteration and time-domain which makes ILC unique. ILC is becoming very popular among control engineers because of its very simple and effective control techniques. This paper describes the basic key knowledge about ILC and its types of applications. The core concern of this paper is to elaborate and explore the future scope and key application areas of iterative learning control in engineering as well as other all subjects of interest.

Improved Indirect Iterative Learning MIT Control Method for Ultrasonic Motor

MIT control strategy is a kind of model reference adaptive control method with the simplest structure. The reason why the structure is simple is that only the gain of the controlled object is adaptively adjusted. Because only the gain is adjusted, the ability of MIT control strategy to change the characteristics of the controlled object is limited. This also limits its application. In this paper, the simple idea of iterative learning control is introduced into MIT controller to increase the controller's ability to adjust the characteristics of the controlled object, and make it suitable for complex control objects. In the proposed control method, the output of the iterative learning controller is used to adjust the adaptive law of the MIT controller. The proposed control method is applied to the speed control system of ultrasonic motor. Experiments show that although only the gain of MIT controller can be adjusted, the learning process based on memory increases the degree of control freedom. Therefore, the dynamic characteristics of the system can be greatly changed, and the control performance can be significantly improved. Moreover, the proposed method only needs to add a simple P-type iterative learning controller to the MIT controller, and the increase of online calculation amount is small.

Precise control for the size of droplet in T-junction microfluidic based on iterative learning method

Microfluidic droplet technique is a new technology developed on the basis of microfluidics to study the formation, manipulation and application of microdroplets of a few micrometers size. It drastically enhances the advantages of microfluidics in terms of low consumption, automation and high throughput and is widely used in chemical, microelectronics, materials science, biology and biomedical engineering etc. In this paper, an iterative learning control (ILC) scheme is proposed to accurately control the droplet size at low capillary numbers. ILC is able to revise the current control input based on the error information measured during previous experimental operations and ultimately produce the desired droplet size in the T-junction microfluidic channel we design. The feasibility of the ILC scheme is verified through experiments, where two different situations are addressed in detail. The results indicate that only a few number of iterations is required to achieve the desired droplet size. The main characteristics of the proposed ILC scheme are as follow: (1) it does not require an accurate mathematical model of the microfluidic systems, which can resolve the severe uncertainties of complex droplet microfluidic system well. (2) Owing to the feedforward characteristic and simple structure of ILC, it is easier to be implemented in practical experimental environments than typical feedback controllers, e.g., proportional-integral-derivative controller. (3) Compared with the traversal methods widely used in medical and biological fields, the idea of ILC would reduce the number of experimental trials significantly.

Gradient-Descent-Based Velocity Observer with a Residual Displacement Term for ILSMC in Contour Following Applications

Industrial applications such as pick-and-place tasks and CNC machining often involve repetitive motions that may contain periodic disturbances. Due to the periodic nature of these applications, the idea of iterative learning control can be exploited to cope with external periodic disturbances so as to improve system performance. In addition, in order to ensure satisfactory motion accuracy for operation scenarios that may encounter significant disturbance, (e.g., machining and polishing), a robust controller is essential. In particular, iterative learning sliding mode control (ILSMC) is shown to be suitable for control applications that encounter periodic disturbances. However, when the disturbance contains high frequency components, ILSMC may become less effective in compensating for external disturbances if its velocity estimation is not accurate enough. In order to cope with the aforementioned problem, this paper develops a gradient-descent-based velocity observer with a residual displacement compensation term to provide accurate velocity feedback to be used in ILSMC. Both theoretical analysis and experimental results are provided to verify the effectiveness of the proposed approach.

Iterative Learning Control for Collaborative Tracking: Point to Point Tasks and Constraint Handling

Collaborative tracking of networked dynamical systems where a group of subsystems work together collaboratively to track a desired reference has important applications in a range of areas. To achieve high performance tracking, the idea of iterative learning control (ILC) has recently been applied with superior tracking performance. This paper considers two previously unexplored problems in ILC design of collaborative tracking, namely, point to point tracking tasks and constraint handling, which are of great practical relevance. We propose two new algorithms to solve the above problems using the idea of gradient based ILC and a projection based method. The proposed algorithms achieve monotonic convergence in the tacking error norm and can guarantee the satisfaction of system constraints. They can be applied to both homogenous and heterogenous networked systems where the subsystems might or might not have the same dynamics. Convergence properties of the algorithms are analysed in detail and numerical examples are presented to demonstrate their effectiveness.

ILC-Based Minimum Entropy Filter Design and Implementation for Non-Gaussian Stochastic Systems

A new filtering approach based on the idea of iterative learning control (ILC) is proposed for linear and non-Gaussian stochastic systems. The objective of filtering is to estimate the states of linear systems with non-Gaussian random disturbances so that the entropy of output error is made to monotonically decrease along the progress of batches of process operation. The term Batch is referred to a period of time when the process repeats itself. During a batch, the filter gain is kept fixed and state estimation is performed. Between any two adjacent batches, the filter gain is updated so that the entropy of closed-loop output error is reduced for the next batch. Analysis is carried out to explicitly determine the learning rates which lead to convergence of the overall algorithm. Experiments have been implemented on a laboratory-based process test rig to demonstrate the effectiveness of proposed filtering method.

Iterative Learning Controller Based on Particle Swarm Optimization

Iterative Learning Control (ILC) has recently emerged as a powerful control strategy that iteratively achieves a higher accuracy for systems with repetitive tasks. The basic idea of ILC is to construct a compensation signal based on the tracking error in each repetition so as to reduce the tracking error in the next repetition. In this paper, particle swarm optimization (PSO) is proposed to optimize the input of iterative learning controller. The experimental results confirm that the proposed method not only has higher tracking accuracy than that of Improved Genetic Algorithm (IGA) and traditional Genetic Algorithm based elisit strategy (EGA), but also has the advantages of simple algorithm and good flexibility. And compared with conventional iterative learning control methods, it is easy to solve the optimal input for non-linear plant models.

Iterative learning control scheme for manipulators including actuator dynamics

This paper presents the iterative learning control for the industrial robot manipulators including actuator dynamics. Motivated by human learning, the basic idea of iterative learning control is to use information from previous execution of a trial in order to improve performance from trial to trial. This is an advantage, when accurate model of the system is not available as friction and actuator dynamics, though present in the system, are not modeled to reduce the computational complexity. In this paper different aspects of ILC including the design schemes and control algorithms are covered. The learning control scheme comprises two types of control laws: a linear feedback law and a feed-forward control law. In the feedback loop, the fixed gain PD controller provides stability of the system and keeps its state errors within uniform bounds. In the feed-forward path, a learning control rule/strategy is exploited to track the entire span of a reference input over a sequence of iterations. Algorithms are verified through detailed simulation results on a two DOF robot manipulator.

Comparison of Standard and Lifted ILC on a Motion System

Iterative Learning Control (ILC) is a technique for improving the performance of systems or processes that operate repetitively over a fixed time interval. The basic idea of ILC is that it exploits every possibility to incorporate past repetitive control information, such as tracking errors and control input signals into the construction of the present control action. Past control information is stored and then used in the control action in order to ensure that the system meets the control specifications such as convergence. The goal of the research presented in this paper is to compare two different ILC techniques applied to the wafer stage of a wafer scanner motion system. Namely, we consider briefly the concepts of standard and lifted ILC and we evaluate the ILC performance in terms of tracking errors. We experience that ILC applied to the wafer stage has liability to deal with limited performance in the face of position dependent dynamics, with the fact that ILC does not account for setpoint trajectory changes and with stochastic effects.