2D Temperature Field Reconstruction Using Optical Frequency Domain Reflectometry and Machine-Learning Algorithms.

We present experimental results on the reconstruction of the 2D temperature field on the surface of a 250 × 250 mm sensor panel based on the distributed frequency shift measured by an optical backscatter reflectometer. A linear regression and a feed-forward neural network algorithm, trained by varying the temperature field and capturing thermal images of the panel, are used for the reconstruction. In this approach, we do not use any information about the exact trajectory of the fiber, material properties of the sensor panel, and a temperature sensitivity coefficient of the fiber. Mean absolute errors of 0.118 °C and 0.086 °C are achieved in the case of linear regression and feed-forward neural network, respectively.

Deep reinforcement learning for self-tuning laser source of dissipative solitons.

Increasing complexity of modern laser systems, mostly originated from the nonlinear dynamics of radiation, makes control of their operation more and more challenging, calling for development of new approaches in laser engineering. Machine learning methods, providing proven tools for identification, control, and data analytics of various complex systems, have been recently applied to mode-locked fiber lasers with the special focus on three key areas: self-starting, system optimization and characterization. However, the development of the machine learning algorithms for a particular laser system, while being an interesting research problem, is a demanding task requiring arduous efforts and tuning a large number of hyper-parameters in the laboratory arrangements. It is not obvious that this learning can be smoothly transferred to systems that differ from the specific laser used for the algorithm development by design or by varying environmental parameters. Here we demonstrate that a deep reinforcement learning (DRL) approach, based on trials and errors and sequential decisions, can be successfully used for control of the generation of dissipative solitons in mode-locked fiber laser system. We have shown the capability of deep Q-learning algorithm to generalize knowledge about the laser system in order to find conditions for stable pulse generation. Region of stable generation was transformed by changing the pumping power of the laser cavity, while tunable spectral filter was used as a control tool. Deep Q-learning algorithm is suited to learn the trajectory of adjusting spectral filter parameters to stable pulsed regime relying on the state of output radiation. Our results confirm the potential of deep reinforcement learning algorithm to control a nonlinear laser system with a feed-back. We also demonstrate that fiber mode-locked laser systems generating data at high speed present a fruitful photonic test-beds for various machine learning concepts based on large datasets.

Highly Dense FBG Temperature Sensor Assisted with Deep Learning Algorithms.

In this paper, we demonstrate the application of deep neural networks (DNNs) for processing the reflectance spectrum from a fiberoptic temperature sensor composed of densely inscribed fiber bragg gratings (FBG). Such sensors are commonly avoided in practice since close arrangement of short FBGs results in distortion of the spectrum caused by mutual interference between gratings. In our work the temperature sensor contained 50 FBGs with the length of 0.95 mm, edge-to-edge distance of 0.05 mm and arranged in the 1500-1600 nm spectral range. Instead of solving the direct peak detection problem for distorted signal, we applied DNNs to predict temperature distribution from entire reflectance spectrum registered by the sensor. We propose an experimental calibration setup where the dense FBG sensor is located close to an array of sparse FBG sensors. The goal of DNNs is to predict the positions of the reflectance peaks of the reference sparse FBG sensors from the reflectance spectrum of the dense FBG sensor. We show that a convolution neural network is able to predict the positions of FBG reflectance peaks of sparse sensors with mean absolute error of 7.8 pm that is slightly higher than the hardware reused interrogator equal to 5 pm. We believe that dense FBG sensors assisted with DNNs have a high potential to increase spatial resolution and also extend the length of a fiber optical sensors.

Inverse design of mode-locked fiber laser by particle swarm optimization algorithm.

A wide variety of laser applications, that often require radiation with specific characteristics, and relative flexibility of laser configurations offer a prospect of designing systems with the parameters on demand. The inverse laser design problem is to find the system architecture that provides for the generation of the desired laser output. However, typically, such inverse problems for nonlinear systems are sensitive to the computation of the gradients of a target (fitness) function making direct back propagation approach challenging. We apply here particle swarm optimization algorithm that does not rely on the gradients of the fitness function to the design of a fiber 8-figure laser cavity. This technique allows us to determine the laser cavity architectures tailored to generating on demand pulses with duration in the range of 1.5-105 ps and spectral width in the interval 0.1-20.5 nm. The proposed design optimisation algorithm can be applied to a variety of laser applications, and, more generally, in a range of engineering systems with flexible adjustable configurations and the outputs on demand.

Ionic Liquid Gated Carbon Nanotube Saturable Absorber for Switchable Pulse Generation.

Materials with electrically tunable optical properties offer a wide range of opportunities for photonic applications. The optical properties of the single-walled carbon nanotubes (SWCNTs) can be significantly altered in the near-infrared region by means of electrochemical doping. The states' filling, which is responsible for the optical absorption suppression under doping, also alters the nonlinear optical response of the material. Here, for the first time we report that the electrochemical doping can tailor the nonlinear optical absorption of SWCNT films and demonstrate its application to control pulsed fiber laser generation. With a pump-probe technique, we show that under an applied voltage below 2 V the photobleaching of the material can be gradually reduced and even turned to photoinduced absorption. Furthermore, we integrated a carbon nanotube electrochemical cell on a side-polished fiber to tune the absorption saturation and implemented it into the fully polarization-maintaining fiber laser. We show that the pulse generation regime can be reversibly switched between femtosecond mode-locking and microsecond Q-switching using different gate voltages. This approach paves the road toward carbon nanotube optical devices with tunable nonlinearity.

Machine learning-based pulse characterization in figure-eight mode-locked lasers.

By combining machine learning methods and the dispersive Fourier transform we demonstrate, to the best of our knowledge, for the first time the possibility to determine the temporal duration of picosecond-scale laser pulses using a nanosecond photodetector. A fiber figure of eight lasers with two amplifiers in a resonator was used to generate pulses with durations varying from 28 to 160 ps and spectral widths varied in the range of 0.75-12 nm. The average power of the pulses was in the range from 40 to 300 mW. The trained artificial neural network makes it possible to predict the pulse duration with the mean agreement of 95%. The proposed technique paves the way to creating compact and low-cost feedback for complex laser systems.

Machine Learning Methods for Control of Fibre Lasers with Double Gain Nonlinear Loop Mirror.

Many types of modern lasers feature nonlinear properties, which makes controlling their operation a challenging engineering problem. In particular, fibre lasers present both high-performance devices that are already used for diverse industrial applications, but also interesting and not yet fully understood nonlinear systems. Fibre laser systems operating at high power often have multiple equilibrium states, and this produces complications with the reproducibility and management of such devices. Self-tuning and feedback-enabled machine learning approaches might define a new era in laser science and technology. The present study is the first to demonstrate experimentally the application of machine learning algorithms for control of the pulsed regimes in an all-normal dispersion, figure-eight fibre laser with two independent amplifying fibre loops. The ability to control the laser operation state by electronically varying two drive currents makes this scheme particularly attractive for implementing machine learning approaches. The self-tuning adjustment of two independent gain levels in the laser cavity enables generation-on-demand pulses with different duration, energy, spectral characteristics and time coherence. We introduce and evaluate the application of several objective functions related to selection of the pulse duration, energy and degree of temporal coherence of the radiation. Our results open up the possibility for new designs of pulsed fibre lasers with robust electronics-managed control.

Raman-converted high-energy double-scale pulses at 1270 nm in P2O5-doped silica fiber.

This work presents implementation of a new approach to single-cascade Raman conversion of laser pulses from the spectral range around 1.1 µm into the 1.3-µm wavelength region. The proposed conversion technique relies on double-scale pico-femtosecond pulses for synchronous pumping of an external cavity made of phosphosilicate fiber with high-precision adjustment of pulse repetition rate to the inter-mode frequency of the external cavity. This enabled generation of double-scale pulses centered at 1270 nm featuring a record energy of 63 nJ and the pulse envelope duration of 88-180 ps with the sub-pulse duration of 200 fs. The fraction of the radiation that was converted into the 1270 nm range amounted to 47 percent of the total Raman-converted radiation power. The generated results offer promising possibilities for new spectral ranges to be developed in the field of high-energy pulsed sources with unique double-scale temporal structure.

Layout of NALM fiber laser with adjustable peak power of generated pulses.

The Letter proposes a new layout of a passively mode-locked fiber laser based on a nonlinear amplifying loop mirror (NALM) with two stretches of active fiber and two independently controlled pump modules. In contrast with conventional NALM configurations using a single piece of active fiber that yields virtually constant peak power, the proposed novel laser features larger than a factor of 2 adjustment range of peak power of generated pulses. The proposed layout also provides independent adjustment of duration and peak power of generated pulses as well as power-independent control of generated pulse spectral width impossible in NALM lasers with a single piece of active fiber.

Ultrafast all-fibre laser mode-locked by polymer-free carbon nanotube film.

This work for the first time reports the results on study of a polymer-free carbon nanotube (CNT) films used as a saturable absorber in an all-fibre laser. It is demonstrated that free-standing single-walled CNT films fabricated by an aerosol method are able to ensure generation of transform-limited pulses in an Er all-fibre ring laser with duration of several picoseconds and high quality of mode locking. The optimal average output power levels are identified, amounting to 0.4-0.5 mW depending on the linear transmission of the studied samples (60% or 80%). Application of polymer-free CNT films solves problems related to degradation of conventional polymer matrices of CNT-based saturable absorbers and paves the way to longer-lasting and more reliable saturable absorbers compatible with all-fibre laser configurations.

Synchronously pumped picosecond all-fibre Raman laser based on phosphorus-doped silica fibre.

Reported for the first time is picosecond-range pulse generation in an all-fibre Raman laser based on P2O5-doped silica fibre. Employment of phosphor-silicate fibre made possible single-cascade spectral transformation of pumping pulses at 1084 nm into 270-ps long Raman laser pulses at 1270 nm. The highest observed fraction of the Stokes component radiation at 1270 nm in the total output of the Raman laser amounted to 30%. The identified optimal duration of the input pulses at which the amount of Stokes component radiation in a ~16-m long phosphorus-based Raman fibre converter reaches its maximum was 140-180 ps.