Data-driven model reference control design by prediction error identification

Campestrini, Lucíola; Eckhard, Diego; Bazanella, Alexandre Sanfelice; Gevers, Michel

doi:10.1016/j.jfranklin.2016.08.006

Cited by 62 publications

(28 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…It is extremely beneficial to work with linearized feedback control systems because their behavior generalizes well in wide operating ranges [37]. The ORM tracking problem concerns the control system behavior from the reference input to the controlled output, neglecting potential load disturbances [38]. Extension of the proposed theory to nonlinear ORMs is not difficult.…”

Section: Output Model Reference Control For Unknown Systemsmentioning

confidence: 99%

“…Extension of the proposed theory to nonlinear ORMs is not difficult. Under classical control rules, the process's delay and non-minimum-phase character should be included in M. However, the non-minimum-phase zeroes make M non-invertible in addition to requiring their knowledge via identification [38], affecting the subsequent VRFT design, motivating the minimum phase assumption on the process.…”

Section: Output Model Reference Control For Unknown Systemsmentioning

confidence: 99%

See 1 more Smart Citation

Data-Driven Model-Free Tracking Reinforcement Learning Control with VRFT-based Adaptive Actor-Critic

Rădac

Precup

2019

Applied Sciences

View full text Add to dashboard Cite

This paper proposes a neural network (NN)-based control scheme in an Adaptive Actor-Critic (AAC) learning framework designed for output reference model tracking, as a representative deep-learning application. The control learning scheme is model-free with respect to the process model. AAC designs usually require an initial controller to start the learning process; however, systematic guidelines for choosing the initial controller are not offered in the literature, especially in a model-free manner. Virtual Reference Feedback Tuning (VRFT) is proposed for obtaining an initially stabilizing NN nonlinear state-feedback controller, designed from input-state-output data collected from the process in open-loop setting. The solution offers systematic design guidelines for initial controller design. The resulting suboptimal state-feedback controller is next improved under the AAC learning framework by online adaptation of a critic NN and a controller NN. The mixed VRFT-AAC approach is validated on a multi-input multi-output nonlinear constrained coupled vertical two-tank system. Discussions on the control system behavior are offered together with comparisons with similar approaches.

show abstract

Section: Output Model Reference Control For Unknown Systemsmentioning

confidence: 99%

Section: Output Model Reference Control For Unknown Systemsmentioning

confidence: 99%

Data-Driven Model-Free Tracking Reinforcement Learning Control with VRFT-based Adaptive Actor-Critic

Rădac

Precup

2019

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…The Optimal Controller Identification method was presented in Campestrini et al (2017) and extended to MIMO systems in Huff et al (2019), both presenting applications to real systems. Different from VRFT, which identifies the ideal controller through least squares, the OCI method identifies it through the prediction error approach.…”

Section: Optimal Controller Identificationmentioning

confidence: 99%

“…Among several data-driven control techniques, the noniterative ones (also known as one-shot) have been largely applied to real systems due to the fact that only one batch of input-output data is required in order to design the controller. Virtual Reference Feedback Tuning (VRFT) (Campi et al, 2000), Correlation-based Tuning (CbT) (Karimi et al, 2007) and the Optimal Controller Identification (OCI) (Campestrini et al, 2017) are representative of such group. Even though data can be obtained in open or closed-loop experiments, the majority of data-driven control literature presents practical results using openloop data (Bazanella et al, 2012;Formentin et al, 2013;Rojas et al, 2011;Campestrini et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

“…Virtual Reference Feedback Tuning (VRFT) (Campi et al, 2000), Correlation-based Tuning (CbT) (Karimi et al, 2007) and the Optimal Controller Identification (OCI) (Campestrini et al, 2017) are representative of such group. Even though data can be obtained in open or closed-loop experiments, the majority of data-driven control literature presents practical results using openloop data (Bazanella et al, 2012;Formentin et al, 2013;Rojas et al, 2011;Campestrini et al, 2017). However, with the growth of historic data availability, the use of closedloop data, collected directly from historic data becomes of great interest, specially because it is not necessary to add disturbances to the process or to open the control loop, which could result in production losses and safety concerns.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

The effects of disturbances in non-iterative data-driven controller design

Silva

Campestrini

2019

Anais Do 14º Simpósio Brasileiro De Automação Inteligente

View full text Add to dashboard Cite

Non-iterative data-driven control design techniques are very compelling for their ease of implementation and the convenience of only using one set of collected data to do the design. During the period of the data set collection, disturbances that may have occurred to the plant will influence on the design results. On this paper, we use a refrigeration system model to test the effects of different disturbance signals during a closed-loop test to the performance of the controllers obtained through two different non-iterative data-driven controller design methods. The controllers obtained were compared on how far the performance obtained was from the desired response and on the Ratio of Integrated Absolute Variation of Control signal (RIAVU). The results show that the disturbances do interfere in the estimated controllers, changing the closed-loop response. Filtering the collected data before applying the data-driven methodologies improved the results significantly, considering the chosen disturbance signals.

show abstract

Data‐assisted control: A framework development by exploiting NASA Generic Transport platform

Eslami,

Banazadeh

2023

Intl J Robust & Nonlinear

View full text Add to dashboard Cite

The increasing capabilities of control systems, driven by abundant data and computational resources, necessitate the incorporation of data‐based techniques with model‐based approaches. While mathematical models exist for air and space vehicles, they may not fully meet the requirements for expanding control system capabilities. Consequently, the development of data‐based extensions in collaboration with these models becomes crucial to provide specific analysis and synthesis tools in the hybrid realm for practical cases. This article introduces the data‐assisted control (DAC) framework for aerospace vehicles, with the NASA generic transport model serving as the platform for study. By leveraging data, the model‐based controller enhances performance during a damage event. Real‐time decisions are made to update the control law based on information obtained from the data, while the model‐based controller may not exhibit consistent performance. The closed‐loop system is shown to be stable in the transition phase between the data and the model, and robust against uncertainties. The dual unscented Kalman filter is employed to estimate fixed dynamic parameters and generalized force‐moment, and the Koopman estimator is used to estimate the evolution equation of the estimated forces and moments. The dual estimation model ensures observability by considering the persistency of excitation. Simulations have shown that the purely model‐based robust control leads to degradation of the closed‐loop performance in case of damage, suggesting the need for data assistance. The DAC framework offers a robust approach to integrate data and models, significantly enhancing control performance in practical aerospace applications.

show abstract

Data-driven model reference control design by prediction error identification

Cited by 62 publications

References 13 publications

Data-Driven Model-Free Tracking Reinforcement Learning Control with VRFT-based Adaptive Actor-Critic

Data-Driven Model-Free Tracking Reinforcement Learning Control with VRFT-based Adaptive Actor-Critic

The effects of disturbances in non-iterative data-driven controller design

Data‐assisted control: A framework development by exploiting NASA Generic Transport platform

Contact Info

Product

Resources

About