Tip-tilt adaptive correction based on stochastic parallel gradient descent optimization algorithm

Abstract

Adaptive optics correcting technique based on stochastic parallel gradient descent (SPGD) control algorithm is an alternative approach which is independent of wavefront sensor and optimizes the performance metric directly. In this paper we establish a simulation model of tip-tilt adaptive optics system, where SPGD optimization algorithm is used to correct the tip-tilt aberration induced by dynamic turbulence. The distance between the measured centroid of a blurring image and the demarcated centroid of the ideal image is considered as the system performance metric, and signals applied to compensation of the drifting atmosphere are generated for tip-tilt mirror. The sensitivity of the performance metrics is investigated, and the convergence rate and the stability are analyzed. By means of this optimization algorithm, the centroid of the image is modified in real time, which results in the improvement of the long exposure image.

Rapid prediction of thrust force coupling scale-span model and revised ANN in drilling CFRPs

“…The Adam algorithm [33] based on gradient descent is adopted to modify the weight indices in the established ANN model. The training process ends when the calculation error is not greater than the preset precision.…”

Rapid prediction of thrust force coupling scale-span model and revised ANN in drilling CFRPs

“…With increasing training times, the weight indices are constantly modified in accordance with Eq. (5) and(6).The Adam algorithm[33] based on gradient descent is adopted to modify the weight indices in the established ANN model. The training process ends when the calculation error is not greater than the preset precision.As can be observed from Fig.9, 4 hidden layers and 18 neurons in the hidden layers exhibit satisfactory training and hence are selected.…”