A tutorial and overview of retrospective cost adaptive control

Rahman, Yousaf; Xie, Antai; Hoagg, Jesse B.; Bernstein, Dennis S.

doi:10.1109/acc.2016.7525440

Cited by 14 publications

(3 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this section, we propose the recursive algorithm to calibrate or adjust the parameters of a generic controller. Given the parameters at the current time step, θ k , we perform an update, ∆θ k , based on data of closed-loop system operation as in (7). In particular, the learning method makes (8) follow a nominal value y k "as closely as possible", i.e., the dynamical system is controlled perfectly if…”

Section: Kalman Filter For Control Parameter Calibrationmentioning

confidence: 99%

See 1 more Smart Citation

Automated Controller Calibration by Kalman Filtering

Menner¹,

Berntorp²,

Cairano³

2021

Preprint

View full text Add to dashboard Cite

This paper proposes a method for calibrating control parameters. Examples of such control parameters are gains of PID controllers, weights of a cost function for optimal control, filter coefficients, the sliding surface of a sliding mode controller, or weights of a neural network. Hence, the proposed method can be applied to a wide range of controllers. The method uses a Kalman filter that estimates control parameters rather than the system's state, using data of closed-loop system operation. The control parameter calibration is driven by a training objective, which encompasses specifications on the performance of the dynamical system. The calibration method tunes the parameters online and robustly, is computationally efficient, has low data storage requirements, and is easy to implement making it appealing for many real-time applications. Simulation results show that the method is able to learn control parameters quickly (approximately 24% average decay factor of closed-loop cost), is able to tune the parameters to compensate for disturbances (approximately 29% improvement on tracking precision), and is robust to noise. Further, a simulation study with the high-fidelity vehicle simulator CarSim shows that the method can calibrate controllers of a complex dynamical system online, which indicates its applicability to a real-world system.

show abstract

Section: Kalman Filter For Control Parameter Calibrationmentioning

confidence: 99%

“…If h(θ k ) is differentiable with respect to the control parameters θ k , the following implementation of the extended Kalman filter (EKF) can be used to compute the parameter update in (7),…”

Section: A Extended Kalman Filtermentioning

confidence: 99%

Automated Controller Calibration by Kalman Filtering

Menner¹,

Berntorp²,

Cairano³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…The failure of the Honeywell MH-96 self-adaptive controller used on the X-15-3 was analyzed in [12], and L 1 adaptive control with an uncertain flight envelope has been flight tested on the NASA AirStar scaled aircraft [13].The goal of the present paper is to investigate the performance of an alternative technique for adaptive flight control, specifically, retrospective cost adaptive control (RCAC). RCAC is a direct discrete-time adaptive control technique for stabilization, command following, and disturbance rejection [14]. As a discrete-time approach, RCAC is motivated by the desire to implement control algorithms that operate at the sensor sample rate without the need for controller discretization.…”

mentioning

confidence: 99%

Retrospective Cost Adaptive Control of Generic Transport Model Under Uncertainty and Failure

Ansari

Bernstein

2017

Journal of Aerospace Information Systems

Self Cite

View full text Add to dashboard Cite

η = ground velocity alignment error with a ⇀ Θ, Φ, Ψ = pitch, roll, yaw angles θ = controller coefficients τ = turn rate ϕ = feedback vector I. IntroductionA LONG with aerodynamics, lightweight structures, and propulsion, feedback control is one of the key technologies that enable aviation.Feedback control uses inertial and noninertial sensors to provide measurements of position, velocity, acceleration, attitude, and angular rates to specify thrust and aerodynamic surfaces to apply corrective forces and moments. Closed-loop control enables the operation of autopilots, which can either assist the pilot or assume complete control of aircraft operation.The standard approach to feedback control of aircraft is based on classical control techniques [1][2][3]. With the advent of optimal control methods, state-space-based control techniques have also been successful [4,5]. Although classical optimal control does not account for model uncertainty, aircraft flight control has benefited from advances in robust control [6,7].For conventional aircraft operating under emergency flight conditions, as well as for unconventional aircraft, recent research has focused on adaptive control techniques [8][9][10][11][12][13]. The failure of the Honeywell MH-96 self-adaptive controller used on the X-15-3 was analyzed in [12], and L 1 adaptive control with an uncertain flight envelope has been flight tested on the NASA AirStar scaled aircraft [13].The goal of the present paper is to investigate the performance of an alternative technique for adaptive flight control, specifically, retrospective cost adaptive control (RCAC). RCAC is a direct discrete-time adaptive control technique for stabilization, command following, and disturbance rejection [14]. As a discrete-time approach, RCAC is motivated by the desire to implement control algorithms that operate at the sensor sample rate without the need for controller discretization. This also means that the required modeling information can be estimated based on data sampled at the same rate as the control update.RCAC was originally motivated by the notion of retrospectively optimized control, where past controller coefficients used to generate past control inputs are reoptimized in the sense that if the reoptimized coefficients had been used over a previous window of operation, then the performance would have been better. Unlike signal processing applications such as estimation and identification, however, it is impossible to change past control inputs, and thus the reoptimized controller coefficients are used only to generate the next control input.RCAC was originally developed within the context of active noise control experiments [15]. The algorithm used in [15] is gradient based, where the gradient direction and step size are based on different cost functions. In subsequent work [16], the gradient algorithm was replaced by batch least-squares optimization. In both [15,16], the modeling information is given by Markov parameters (impulse response components) of the openloop transfer function ...

show abstract

Improving the Electrified Powertrain Performance based on Retrospective Behaviour of Battery state of charge and engine fuel consumption

Ahmed

Rahman

2018

IFAC-PapersOnLine

View full text Add to dashboard Cite

A tutorial and overview of retrospective cost adaptive control

Cited by 14 publications

References 39 publications

Automated Controller Calibration by Kalman Filtering

Automated Controller Calibration by Kalman Filtering

Retrospective Cost Adaptive Control of Generic Transport Model Under Uncertainty and Failure

Improving the Electrified Powertrain Performance based on Retrospective Behaviour of Battery state of charge and engine fuel consumption

Contact Info

Product

Resources

About