Data-driven Iterative Learning Research Articles

Acceleration waveform reproduction control of hypergravity centrifugal shaking table based on data driven iterative learning control and two-degree-of-freedom control

The hypergravity centrifugal shaking table (HCST) is the most effective means for studying the disaster effects of rock and soil earthquakes, and the accurate control of acceleration waveform recurrence is the critical problem. This paper compares the dynamic characteristics of the normal gravity shaking table (NGST) and the HCST, highlighting the latter’s particularities, and analyzes the limitations of the existing control strategy. Meanwhile, a hypergravity unidirectional airborne shaking table (HUAST) is designed, and a data-driven iterative learning control and two-degree-of-freedom control-based acceleration waveform reproduction strategy is proposed and experimentally verified.

Finite-Time Asymmetric Bipartite Consensus for Multi-Agent Systems Using Data-Driven Iterative Learning Control

Finite-Time Asymmetric Bipartite Consensus for Multi-Agent Systems Using Data-Driven Iterative Learning Control

Event-Triggered Distributed Data-Driven Iterative Learning Bipartite Formation Control for Unknown Nonlinear Multiagent Systems.

In this study, we investigate the event-triggering time-varying trajectory bipartite formation tracking problem for a class of unknown nonaffine nonlinear discrete-time multiagent systems (MASs). We first obtain an equivalent linear data model with a dynamic parameter of each agent by employing the pseudo-partial-derivative technique. Then, we propose an event-triggered distributed model-free adaptive iterative learning bipartite formation control scheme by using the input/output data of MASs without employing either the plant structure or any knowledge of the dynamics. To improve the flexibility and network communication resource utilization, we construct an observer-based event-triggering mechanism with a dead-zone operator. Furthermore, we rigorously prove the convergence of the proposed algorithm, where each agent's time-varying trajectory bipartite formation tracking error is reduced to a small range around zero. Finally, four simulation studies further validate the designed control approach's effectiveness, demonstrating that the proposed scheme is also suitable for the homogeneous MASs to achieve time-varying trajectory bipartite formation tracking.

Robust Data-Driven Iterative Learning Control for High-Speed Train With Aperiodic DoS Attacks and Communication Delays

Robust Data-Driven Iterative Learning Control for High-Speed Train With Aperiodic DoS Attacks and Communication Delays

An Improved Data-Driven Iterative Learning Secure Control for Intelligent Marine Vehicles With DoS Attacks

An Improved Data-Driven Iterative Learning Secure Control for Intelligent Marine Vehicles With DoS Attacks

Quantised data-driven iterative learning bipartite consensus control for unknown heterogeneous linear MASs with varying trial lengths

This paper aims to realise the robust output bipartite consensus for unknown heterogeneous linear time-varying multiagent systems (MASs) subject to varying trial lengths, unknown measurement disturbances and data quantisation. To this end, inspired by the idea of quantised control, a quantised data-driven adaptive iterative learning bipartite consensus (AILBC) method is proposed. Specifically, to address the problem of varying trial lengths, a distributed auxiliary output prediction system is constructed based on the agents' input-output (I/O) dynamic relationship. An adaptive update protocol is developed to estimate the measurement disturbances and unknown parameters of I/O dynamic relationship. Subsequently, a quantised distributed data-driven iterative learning control (ILC) approach based on the quantised output information is proposed for MASs to achieve robust bipartite consensus tracking, with an attempt to relax the need of explicit model information. The bipartite consensus tracking errors are ultimately bounded through rigorous analysis, and this result is further extended to switching topologies. Finally, numerical simulations are conducted to verify the validity of the AILBC method.

Distributed data-driven iterative learning point-to-point consensus tracking control for unknown nonlinear multi-agent systems

This paper studies the point-to-point consensus tracking problem for a class of nonlinear non-affine multi-agent systems, where the tracking target is a series of some given ideal output points rather than a complete ideal trajectory. For the unknown nonlinear dynamics of every agent, the relationship between its control input and the corresponding output signal at these given points is acquired, then a data-driven model is constructed based on adjacent-agent dynamic linearization technology. Furthermore, by optimizing two performance index functions, we design a novel point-to-point iterative learning tracking control scheme, which not only ensures the consensus tracking results of all agents only utilizing the I/O data of every agent at the given ideal output points, but also makes all agents reach consensus behavior at the rest of the time instants. The effectiveness of our algorithm is confirmed through rigorous theoretical analysis and two experimental simulations.

Data-Driven Spatial Adaptive Terminal Iterative Learning Predictive Control for Automatic Stop Control of Subway Train With Actuator Saturation

A data-driven spatial adaptive terminal iterative learning predictive control (SATILPC) scheme with actuator saturation is proposed for automatic stop control of the subway train. Considering the outstanding repetitive operation pattern and the unavailable accurate model of a subway train, the unknown train dynamics is firstly transformed into a nonlinear discrete form in spatial domain via the spatial differential operator. Since the train automatic stop control (TASC) only concentrates on the tracking performances of the terminal position and terminal speed, an iterative dynamic linearization approach considering the terminal operation point is devised to formulate the relationship of the train input and output (I/O) into a linear affine form. Then, a terminal iterative learning prediction mechanism is introduced to reconstruct the developed train data model to forecast the future train behaviors through rolling optimization process. As a result, by removing the constraints on the unimportant points, the optimal control input (braking force) can be obtained by minimizing the objective function through terminal I/O data. Further, actuator saturation is considered to address the passenger comfort and the reliable operation of the train. The proposed SATILPC approach is a pure data-driven iterative learning control scheme and no model information is involved in the whole design processes. By utilizing a newly space-based contraction mapping method, the convergence of the proposed approach is strictly proved. Finally, the simulation results further demonstrate the feasibility of the proposed algorithm.

Compensating Aerodynamics of Over-Actuated Multi-Rotor Aerial Platform With Data-Driven Iterative Learning Control

Compensating Aerodynamics of Over-Actuated Multi-Rotor Aerial Platform With Data-Driven Iterative Learning Control

ESO-Based Model-Free Adaptive Iterative Learning Energy-Efficient Control for Subway Train With Disturbances and Over-Speed Protection

An extended state observer based model-free adaptive iterative learning energy-efficient control (ESO-based MFAILEEC) scheme for subway train speed tracking with external disturbances and over-speed protection under the constraint on traction/braking force is proposed. Firstly, the continuous-time train motion dynamics is formulated into a discrete-time data model with consideration of external disturbances by applying the iterative dynamic linearization. Meanwhile, the external disturbances and the unknown nonlinear uncertainties of the train are transformed into a new state, which is estimated by an ESO designed in the iteration domain. Then, the ESO-based energy-efficient controller with learning ability is designed, which enables the train to achieve the purpose of energy efficiency by reducing the input force. Further, over-speed protection with trigger mechanism is developed to ensure the train operates within safe speed range. All the control strategies are designed under the constraint on traction/braking force by considering the practical limitation of the train system. No model information is involved in the whole design processes and it is a pure data-driven iterative learning approach. Rigorous mathematical analysis proves the feasibility and the robustness of the proposed method, which can guarantee the train operates safely and reliably. Finally, the simulation results further demonstrate the effectiveness of the proposed algorithm.

A Data-Driven Iterative Learning Approach for Optimizing the Train Control Strategy

The energy-efficient train control (EETC) problem is investigated in this paper. And a soft actor-critic-based method is proposed to optimize the train driving strategy. Firstly, EETC problem is converted to the inverse problem, i.e., minimizing the trip time of the journey with constant energy consumption. Based on the conversion, the EETC problem is reformulated as a finite Markov decision process which can be solved by deep reinforcement learning algorithms. Secondly, an optimization method based on the soft actor-critic (SAC) method is designed to calculate the optimal driving strategy of the train with introducing the reservoir sampling method. Finally, some case studies are conducted to verify the effectiveness and performance of the proposed method. Simulation results demonstrate that a good energy-saving performance can be achieved. In single interval, the SAC-based method can reduce about 1.65% of the energy consumption compared with numerical method. And the energy consumption reduction can be extended to be 6.49% when the proposed approach is applied in multiple intervals.

Neural network-based iterative learning control for trajectory tracking of unknown SISO nonlinear systems

This paper proposes a neural network-based (NN-based) data-driven iterative learning control (ILC) algorithm for the tracking problem of nonlinear single-input single-output (SISO) discrete-time systems with unknown models and repetitive tasks. The control objective is to make the output of the system track the reference trajectory in each iteration process. Therefore, at each relative time during every iteration process, a generalized regression neural network (GRNN) is used as the estimator to solve the key parameters of the system, and a radial basis function neural network (RBFNN) is used as the controller to solve the control input. Compared with the traditional ILC algorithm, the two complex solving processes, i.e., dynamic linearization and criterion function minimization, are replaced and simplified into the iterative training of GRNNs and RBFNNs. The proposed algorithm is out-of-the-box and uses a point-to-point method to calculate the control input for each relative time of the system iteration, driving the tracking error of the system to approach zero. In addition, it is proved that the tracking error of the system under the proposed control algorithm is uniformly ultimately bounded. Finally, a numerical example shows the effectiveness and superiority of the control algorithm, and a path tracking experiment of an unmanned vehicle further verifies its practicability.

Fault-tolerant design of non-linear iterative learning control using neural networks

The design of neural-network-based iterative learning control for non-linear systems is addressed in the setting of a fault tolerant control regime. Taking advantage of the repetitive character of the control task, the inherent uncertainty related to a potential faulty system state can be properly accommodated in terms of a data-driven iterative learning scheme with neural networks used for forward/inverse modeling as well as for controller synthesis. The resulting control technique is supposed to be flexible enough to accurately compensate the faults occurring on both the sensors and actuators and, additionally, take into account the disturbances and noise acting on the system. A complete characterization of the novel fault-tolerant iterative learning scheme is provided including system identification, fault detection and accommodation. Also, the painstaking convergence analysis is presented and the resulting sufficient conditions can be constructively used to determine the update of control law in the consecutive process trial. The excellent performance of the developed control scheme is illustrated by a nontrivial example of tracking control for a magnetic brake system on various scenarios involving actuator and/or sensor faults.

Kernel-based regularized iterative learning control of repetitive linear time-varying systems

For data-driven iterative learning control (ILC) methods, both the model estimation and controller design problems are converted to parameter estimation problems for some chosen model structures. It is well-known that if the model order is not chosen carefully, models with either large variance or large bias would be resulted, which is one of the obstacles to further improve the modeling and tracking performances of data-driven ILC in practice. An emerging trend in the system identification community to deal with this issue is using regularization instead of the statistical tests, e.g., AIC, BIC, and one of the representatives is the so-called kernel-based regularization method (KRM). In this paper, we integrate KRM into data-driven ILC to handle a class of repetitive linear time-varying systems, and moreover, we show that the proposed method has ultimately bounded tracking error in the iteration domain. The numerical simulation results show that in contrast with the least squares method and some existing data-driven ILC methods, the proposed one can give faster convergence speed, better accuracy and robustness in terms of the tracking performance.

Genetic algorithm-based optimization framework for control parameters of ventricular assist devices

In this work, a novel, genetic algorithm-based optimization framework (GAOF) has been developed and evaluated to optimize the control parameters of ventricular assist devices (VADs). This framework enables the optimization of complex control structures based on VAD- and patient-specific characteristics by allowing the selection of the numerical model of the human cardiovascular system and the VAD to represent the patient scenario of interest accurately. Additionally, the GAOF can incorporate treatment-specific goals during the definition of the objective functions of the optimization problem and, consequently, promotes the development of treatment-specific VAD controllers. The efficacy of the GAOF was assessed with one- and two-degree-of-freedom physiologic proportional-integral-derivative controllers and a physiologic data-driven iterative learning controller. Two VAD designs and various patient disease scenarios were used to further explore and evaluate the capabilities of the GAOF. The optimized controllers outperformed substantially the hand-tuned controller, which was used as the benchmark, in all the investigated cases. This highlights the potential improvement in the performance of any VAD controller by deploying the GAOF and, consequently, the possibility to increase the survival rates and enhance the quality of life of VAD patients.

Adaptive iterative extended state observer-based data-driven iterative learning model predictive control for semiconductor silicon single crystal batch process

Aiming at the problems of unstable batch control of key crystal quality parameters and susceptibility to batch-to-batch non-repetitive disturbances during repeated operation of single crystal furnaces, this paper proposes a data-driven iterative learning model predictive control method based on an adaptive iterative extended state observer (IESO) for designing melt temperature and crystal diameter learning controllers with disturbance suppression. By applying dynamic linearization techniques and model predictive control strategies along the iterative axis, an ILMPC scheme with disturbance compensation terms using only input and output data of the system is designed. Among them, adaptive IESO is used to estimate the disturbance compensation terms. Then, the theoretical analysis shows that the tracking error of the ILMPC scheme can converge to a bounded range as the number of iterations increases. The experimental results verify the effectiveness of the proposed control method, which not only ensures that the control system has learning ability, but also achieves stable and accurate control of crystal quality parameters.

Distributed Data-driven Iterative Learning Control for Consensus Tracking

High performance consensus tracking of networked dynamical systems working repetitively has found applications in a range of areas. Existing iterative learning control (ILC) designs for this problem either require a system model that can be difficult or expensive to obtain in practice, and/or cannot guarantee the monotonic convergence of the tracking error norm. They often have difficulties handling varying networks too. This paper proposes a data-driven norm optimal ILC (DD-NOILC) framework to address these limitations using the recent development in data-driven control, in particular, the so called Willems’ fundamental lemma. The novel design guarantees that even without using any model information, the proposed DD-NOILC framework can achieve the same convergence performance as the model-based NOILC framework, i.e., monotonic convergence of the tracking error norm to zero. Furthermore, using the alternating direction method of multipliers (ADMM), a distributed implementation of the framework is developed such that each subsystem's input can be updated locally, making the proposed distributed DD-NOILC algorithm suitable for large-scale and varying networks. Convergence properties of the proposed algorithms are analysed rigorously, and numerical examples are provided to verify the effectiveness of the distributed DD-NOILC algorithm.

A Data-Driven and Personalized Stance Symmetry Controller for Robotic Ankle-Foot Prostheses: A Preliminary Investigation.

People with unilateral transtibial amputation generally exhibit asymmetric gait, likely due to inadequate prosthetic ankle function. This results in compensatory behavior, leading to long-term musculoskeletal impairments (e.g., osteoarthritis in the joints of the intact limb). Powered prostheses can better emulate biological ankles, however, control methods are over-reliant on able-bodied data, require extensive amounts of tuning by experts, and cannot adapt to each user's unique gait patterns. This work directly addresses all these limitations with a personalized and data-driven control strategy. Our controller uses a virtual setpoint trajectory within an impedance-inspired formula to adjust the dynamics of the robotic ankle-foot prosthesis as a function of stance phase. A single sensor measuring thigh motion is used to estimate the gait phase in real time. The virtual setpoint trajectory is modified via a data-driven iterative learning strategy aimed at optimizing ankle angle symmetry. The controller was experimentally evaluated on two people with transtibial amputation. The control scheme successfully increased ankle angle symmetry about the two limbs by 24.4% when compared to the passive condition. In addition, the symmetry controller significantly increased peak prosthetic ankle power output at push-off by 0.52 W/kg and significantly reduced biomechanical risk factors associated with osteoarthritis (i.e., knee and hip abduction moments) in the intact limb. This research demonstrates the benefits of personalized and data-driven symmetry controllers for robotic ankle-foot prostheses.

An iterative data-driven learning algorithm for calibration of the internal model in advanced wind turbine controllers

Modern industrial wind turbine controllers for partial-load region control are becoming increasingly complex by progressively relying on modeled aerodynamic characteristics. These advanced turbine controllers generally consist of a combined wind speed estimator and tracking controller, allowing for a granular trade-off between energy capture maximization and (fatigue) load minimization. Because of the limited measurements available to the controller, the control scheme's internal model quality is of utmost importance in satisfying performance and stability requirements. Therefore, the calibration thereof is of particular interest. To date, little work has been performed on the direct calibration of the model information. This work proposes a data-driven iterative learning algorithm for calibrating the internal physical model parameters. The learning algorithm uses generally available closed-loop turbine measurements, complemented with an external measurement of the rotor effective wind speed (REWS), and is thereby largely nondisruptive. The algorithm is based on steady-state assumptions and performs iterative batch-wise updates of the internal control model toward convergence. As the algorithm corrects at the actual turbine operating point, short-term relocations of the turbine's operating point can be used to calibrate in a broader operational domain. Results show outstanding learning capabilities for an aerodynamically degraded wind turbine under realistic turbulent wind conditions. Moreover, a sensitivity study is performed to expose the algorithm's susceptibility to measurement errors, algorithm tuning, and the size of the data set.

Data-Driven Iterative Learning Predictive Control for Power Converters

This letter proposes a data-driven iterative learning predictive control architecture for power converters. The main objectives of this letter are to enhance the robustness and remain the high performance of finite control-set model predictive control (FCS-MPC) under unmodeled dynamics and parameter mismatch conditions. More specifically, an iterative dynamic linearization technique is utilized to equivalently reformulate the nonlinear power converter system at each operating point. Based on this, a model-free adaptive control scheme is presented to iteratively determine the optimal control actions. Due to the incorporation of iterative learning control and data-driven concept into the FCS-MPC framework, the effect of parameter perturbations can be alleviated in the proposed method, while creating a positive effect on the tracking error. Finally, a convergence analysis is provided and experimental investigations on a three-level neutral-point-clamped (NPC) converter confirm the effectiveness of the proposed method.