Owing to their compactness, robustness, low cost, high stability, and diffraction-limited beam quality, mode-locked fiber lasers play an indispensable role in micro/nanomanufacturing, precision metrology, laser spectroscopy, LiDAR, biomedical imaging, optical communication, and soliton physics. Mode-locked fiber lasers are a highly complex nonlinear optical system, and understanding the underlying physical mechanisms or the flexible manipulation of ultrafast laser output is challenging. The traditional research paradigm often relies on known physical models, sophisticated numerical calculations, and exploratory experimental attempts. However, when dealing with several complex issues, these traditional approaches often face limitations and struggles in finding effective solutions. As an emerging data-driven analysis and processing technology, artificial intelligence (AI) has brought new insights into the development of mode-locked fiber lasers. This review highlights the areas where AI exhibits potential in accelerating the development of mode-locked fiber lasers, including nonlinear dynamics prediction, ultrashort pulse characterization, inverse design, and automatic control of mode-locked fiber lasers. Furthermore, the challenges and potential future development are discussed.
Infrared thermal imager can provide two-dimensional temperature information of the measured object on-line without contact, which is one of the most commonly used temperature measurement methods. However, the measuring distance and viewing angle between the thermal imager and the measured object will cause the infrared temperature measurement error. In order to automatically compensate for the temperature measurement error caused by measuring distance and viewing angle when the infrared thermal imager is applied in actual measurement scenarios, this paper proposes an adaptive compensation method for the infrared temperature measurement error based on 3D thermal imaging. Firstly, based on the polynomial regression and infrared temperature measurement principle, a compensation model is established to compensate for the temperature measurement error caused by measuring distance and viewing angle. Secondly, the parameters of the compensation model are determined by using a blackbody furnace and a reference cylinder. Then, a 3D thermal imaging system is constructed with a depth camera and an infrared thermal imager to quantify the measuring distances and viewing angles between all temperature pixels and the thermal imager. Finally, the surface temperature of the measured object can be adaptively compensated using the established compensation model. The experimental results of the blackbody furnace and resin cylinder show that the proposed method can adaptively compensate for the infrared temperature measurement error caused by measuring distance and viewing angle.
Currently, soft sensors have been extensively used in complex industrial processes. How to learn effective features from nonlinear process data is the core task of building precise soft sensors. Deep learning, a type of feature extraction approach, has shown unique advantages in many fields, as well as in soft sensor modeling. In this article, stacked autoencoder (SAE) is introduced for soft sensor modeling. However, most SAE-based methods focus on learning feature representations from raw input data, as well as ignore the impact of redundant feature information on model performance during the hierarchical pre-training stage. To this end, a novel target-guided related feature learning (TRFL) method is proposed to address the aforementioned two issues. First, an improved grey relational analysis (GRA) method is utilized to evaluate the importance of input variable for each autoencoder (AE), and a threshold determination mechanism is developed to remove the unrelated features. Later, in the decoding stage of each AE, the target variable and input layer variables are simultaneously reconstructed at the output layer, and a dynamic trade-off factor is defined to balance the dimension ratio change of the input layer variable with the target variable. In this way, target-related features can be sufficiently extracted by TRFL. Finally, the effectiveness and superiority of the proposed TRFL are validated with a numerical case and the prediction of two key parameters in the blast furnace ironmaking process.
Dual-comb spectroscopy has been an infusive spectroscopic tool for gas detection due to its high resolution, high sensitivity, and fast acquisition speed over a broad spectral range without any mechanical scanning components. However, the complexity and cost of high-performance dual-comb spectroscopy are still high for field-deployed applications. To solve this problem, we propose a simple frequency domain post-processing method by extracting the accurate position of a specific absorption line frame by frame. After aligning real-time spectra and averaging for one second, the absorbance spectrum of H13C14N gas in the near-infrared is obtained over 1.1 THz spectral range. By using this method, the standard deviation of residual error is only ~0.002, showing great agreement with the conventional correction method. In addition, the spectral resolution is improved from 13.4 GHz to 4.3 GHz compared to direct spectrum averaging. Our method does not require a specially designed common-mode suppression comb, rigorous frequency control system, or complicated computational algorithm, providing a cost-effective scheme for field-deployed Doppler-limited spectroscopy applications.