Proceedings Article | 11 October 2024
KEYWORDS: Neural networks, Education and training, Doppler effect, Signal detection, Signal processing, Data modeling, Scattered light, Sampling rates, Spiral phase plates, Optical simulations
Vortex beam has shown great potential in target rotational motion parameter detection due to it’s unique helical spatial phase structure. The basic principle is the rotational Doppler effect (RDE), which, unlike the classical linear Doppler effect, can be observed even if the moving target does not have a velocity component in the direction of beam propagation, thus effectively extracting target motion information when classical Doppler shift is difficult to observe. The potential of vortex beams to detect the rotational motion parameters of targets has been fully exploited with the intensive research in recent years, including detection of the rotational speed (ω), angular acceleration (a), rotational direction, position of the rotating axis (γ,d) and even the attitude of the rotating object. These studies have accelerated the progress of rotational speed measurement principles based on vortex beams RDE from theory to engineering applications. However, currently most of the information on rotational motion parameters is obtained through frequency transformation of the echo signal, and in the actual detection process, manual interpretation is mainly used to ensure accuracy of the signal, which has disadvantages such as low efficiency and difficulty in large-scale promotion and application. If there is a method that can automatically obtain target speed information directly through time-domain signals, it may greatly advance the process of this technology from theory to practical application. The intelligent extraction based on neural networks provides a new approach to solving this problem. Due to the strong coupling between parameters such as rotational speed, topological charge of vortex beam, and time-domine signals during the detection process, it is possible to simulate the patterns through artificial neural network on the basis of a large amount of detection data, thereby intelligently and accurately extracting of the rotation parameters. In this article, we conduct research on intelligent extraction of target speed motion information based on artificial neural networks. The constructed artificial neural network is trained using a large amount of simulation data, and the neural networks model is verified to achieve high-precision acquisition of target speed information directly based on time-domine signals.
