Modelling and prediction of non-stationary optical turbulence behaviour

Doelman, Niek; Osborn, James

doi:10.1117/12.2233351

Cited by 3 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Meanwhile, the scientific camera records the corrected images and sends them for post-processing in order to get even higher resolution. disturbance 45,46 . LQG is a kind of linear prediction algorithm in which any deviation from the estimated model would degrade the performance.…”

Section: Traditional Wavefront Reconstruction and Controlmentioning

confidence: 99%

Adaptive optics based on machine learning: a review

Guo¹,

Zhong²,

Lei³

et al. 2022

OEA

View full text Add to dashboard Cite

Adaptive optics techniques have been developed over the past half century and routinely used in large ground-based telescopes for more than 30 years. Although this technique has already been used in various applications, the basic setup and methods have not changed over the past 40 years. In recent years, with the rapid development of artificial intelligence, adaptive optics will be boosted dramatically. In this paper, the recent advances on almost all aspects of adaptive optics based on machine learning are summarized. The state-of-the-art performance of intelligent adaptive optics are reviewed. The potential advantages and deficiencies of intelligent adaptive optics are also discussed.

show abstract

Section: Traditional Wavefront Reconstruction and Controlmentioning

confidence: 99%

Adaptive optics based on machine learning: a review

Guo¹,

Zhong²,

Lei³

et al. 2022

OEA

View full text Add to dashboard Cite

show abstract

“…We have described a method that can be used to model these non-stationarities and that the change in Fried parameter can be modelled following Doelman. 22 We show that using this model, the variations in Fried parameter over short time scales do not change the performance of both the integrator and the LMMSE predictors. Inserting wind gusts (into the simulation), we see that the integrator and the LMMSE predictors are less efficient at rejecting the atmospheric turbulence corresponding to a factor of two increase in the rms wavefront error.…”

Section: Resultsmentioning

confidence: 81%

“…There is no data on the Fried parameter at a frequency of 1 Hz or greater using available stereo-scidar measurements. Using wavefront sensor data, Doelman 22 determined a fractional ARIMA model to describe the evolution of the Fried parameter at La Palma on the time interval of 100 seconds. We adopt this model, creating a time series by assuming the Fried parameter remains constant over these 100 seconds.…”

Section: Fried Parameter Variationsmentioning

confidence: 99%

Performance of AO predictive control in the presence of non-stationary turbulence.

Kooten¹,

Doelman²,

Kenworthy³

2017

Proceedings of the Adaptive Optics for Extremely Large Telescopes 5

Self Cite

View full text Add to dashboard Cite

Current high contrast imaging systems are limited at small angular separations, preventing the direct imaging of Earth-size exoplanets close to their host stars. One primary cause of this is the performance of the extreme adaptive optics (AO) system which is dominated by the servo-lag error at small angular separations. Prediction can be used to help reduce this servo-lag error. Most AO systems today do not use predictive controllers (integral controller), resulting in a phase correction that is proportional to the current measured wavefront and does not take into account the evolution of the wavefront phase fluctuations between measurement and correction. To improve upon this, we focus on the development of different predictive control strategies that estimate how the wavefront phase fluctuations evolve over a time horizon equal to the servo-lag. Moreover, the statistical properties of the turbulence phase are non-stationary (change in time) which needs to be included in the AO analysis for high contrast imaging systems. First, we show that the non-stationary properties of the atmosphere can be modeled using a von Karman covariance function which can incorporate variations in time of the Fried parameter, outer scale, and wind velocity. Using this new disturbance model of turbulence, the performance of three different predictors-based on a steady state, a recursive, and an adaptive linear-minimum-mean-squareestimator (LMMSE)-is tested under non-stationary conditions. We feed the prediction algorithms wavefront slopes from a 11-by-11 subaperture Shack-Hartmann wavefront sensor. A 97 actuator deformable mirror applies the predictor's phase correction. We present the latest results of our simulations and compare our predictors with the common integrator. We show that under our varying wind conditions, the root-mean-square wavefront error can increase by a factor of two for both the integrator and our predictors. In our simulations, we do not observe an increase in performance using an exponential forgetting factor adaptive LMMSE.

show abstract

“…Identification of models 30 and adaptation to unstationary conditions can be found in. 47 Recent work is reported in. 48…”

Section: Robustness and Model Identificationmentioning

confidence: 99%