Design of a Femtosecond Laser Percussion Drilling Process for Ni-Based Superalloys Based on Machine Learning and the Genetic Algorithm.

Femtosecond laser drilling is extensively used to create film-cooling holes in aero-engine turbine blade processing. Investigating and exploring the impact of laser processing parameters on achieving high-quality holes is crucial. The traditional trial-and-error approach, which relies on experiments, is time-consuming and has limited optimization capabilities for drilling holes. To address this issue, this paper proposes a process design method using machine learning and a genetic algorithm. A dataset of percussion drilling using a femtosecond laser was primarily established to train the models. An optimal method for building a prediction model was determined by comparing and analyzing different machine learning algorithms. Subsequently, the Gaussian support vector regression model and genetic algorithm were combined to optimize the taper and material removal rate within and outside the original data ranges. Ultimately, comprehensive optimization of drilling quality and efficiency was achieved relative to the original data. The proposed framework in this study offers a highly efficient and cost-effective solution for optimizing the femtosecond laser percussion drilling process.

Multimodal Deep-Learning Framework for Accurate Prediction of Wettability Evolution of Laser-Textured Surfaces.

The lengthy process through which laser-textured surfaces transform from hydrophilic to hydrophobic severely restricts their practical applications. Accurately predicting the wettability evolution curve is crucial; however, developing a reliable prediction model remains challenging. Herein, a data-driven multimodal deep-learning framework was developed, in which multimodal data of micro/nanostructure morphology images, composition distribution images, and time information are effectively coupled and fed into a convolutional neural network (CNN). Rich data input and in-depth data mining make the framework more robust, achieving accurate prediction of the wettability evolution curves of various typical micro/nanostructures. Additionally, accurate prediction of input images with varying magnifications and untrained laser-textured surfaces demonstrates the generalizability of the multimodal CNN framework. The visualization results of the convolution layer confirmed the rationality of the information learned by the model. Additionally, the proposed multimodal CNN framework was successfully utilized to investigate the optimization process. Further, a laser-textured surface with a shorter evolution period and a larger final contact angle was realized. The proposed multimodal CNN framework offers an efficient and cost-effective method for predicting the wettability evolution curves and exploring the optimization processes, enhancing the application potential of laser micro/nanofabrication of superhydrophobic surfaces.

Martensite Start Temperature Prediction through a Deep Learning Strategy Using Both Microstructure Images and Composition Data.

In recent decades, various previous research has established empirical formulae or thermodynamic models for martensite start temperature (Ms) prediction. However, most of this research has mainly considered the effect of composition and ignored complex microstructural factors, such as morphology, that significantly affect Ms. The main limitation is that most microstructures cannot be digitized into numerical data. In order to solve this problem, a convolutional neural network model that can use both composition information and microstructure images as input was established for Ms prediction in a medium-Mn steel system in this research. Firstly, the database was established through experimenting. Then, the model was built and trained with the database. Finally, the performance of the model was systematically evaluated based on comparison with other, traditional AI models. It was proven that the new model provided in this research is more rational and accurate because it considers both composition and microstructural factors. In addition, because of the use of microstructure images for data augmentation, the deep learning had a low risk of overfitting. When the deep-learning strategy is used to deal with data that contains both numerical and image data types, obtaining the value matrix that contains interaction information of both numerical and image data through data preprocessing is probably a better approach than direct linking of the numerical data vector to the fully connected layer.

Prediction of Deformation-Induced Martensite Start Temperature by Convolutional Neural Network with Dual Mode Features.

Various models were established for deformation-induced martensite start temperature prediction over decades. However, most of them are empirical or considering limited factors. In this research, a dual mode database for medium Mn steels was established and a convolutional neural network model, which considered all composition, critical processing information and microstructure images as inputs, was built for Msσ prediction. By comprehensively considering composition, processing and microstructure factors, this model was more rational and much more accurate than traditional thermodynamic models. Also, by the full use of images information, this model has stronger ability to overcome overfitting compared with various traditional machine learning models. This framework provides inspiration for the similar data analysis issues with small sample datasets but different data modes in the field of materials science.

Austenite layer and precipitation in high Co-Ni maraging steel.

In high Co-Ni maraging steel, austenite has a great effect on the fracture toughness of the steel and the precipitated carbides are the main strengthening phase. In this study, both austenite layers and precipitation were observed and their formation theory was analyzed by Thermo-Calc simulation and several reported results. TEM and HRTEM observation results showed that the thickness of the austenite layers was about 5-10 nm and the length of the needle-like precipitated carbides was less than 10nm. The carbides maintained coherent or semi-coherent relation with the matrix.