Performance Improvement of Single-Frequency CW Laser Using a Temperature Controller Based on Machine Learning

Abstract

The performance improvement of an all-solid-state single-frequency continuous-wave (CW) laser with high output power is presented in this paper, which is implemented by employing a temperature control system based on machine learning to control the temperature of laser elements including gain crystal, laser diode and so on. Because the developed temperature controller based on machine learning combines the back propagation (BP) neural network algorithm with the proportion-integration-differentiation (PID) control algorithm, the parameters of the PID are adaptive with the variation of the environment. As a result, the control speeds and control abilities of the temperatures of the elements are dramatically enhanced. In this case, the output characteristic and the adaptability to the environment as well as the stability of the single-frequency CW laser are also improved greatly.

Auto-Regression Model-Based Off-Line PID Controller Tuning: An Adaptive Strategy for DC Motor Control

“…Modern control and other techniques such as adaptive and artificial intelligence control for changing gains are currently used as alternatives for adaptation mechanisms to system changes in time. In [7], machine learning was used to optimize temperature control for laser instruments. Selecting the attention to specific tasks as an exhibition of cognitive and neural mechanisms was mentioned in [8] for control purposes as an efficient strategy toward adaptive control.…”

Editorial for the Special Issue on Optics and Photonics in Micromachines

“…In this Special Issue, we are glad to collect 15 research articles covering a broad area, including optical field modulation [ 11 ], laser fabrication techniques [ 12 , 13 ], optical measurement [ 14 , 15 , 16 , 17 , 18 ], on-chip photonic devices [ 19 , 20 , 21 , 22 ], super-resolution imaging [ 23 , 24 ], and related theoretical [ 25 ] investigations.…”

“…By the post-processing methods, the raw confocal microscopy images with spatial resolution over the diffraction limit can be enhanced to below the diffraction limit. There is also another machine-learning-based study on the optimization of CW laser [ 25 ].…”