报告题目：

1.Small Data Integration for Process Modeling by Using Deep Neural Network-based Word Embedding

2.Nondestructive Defect Detection in Polymer Composites using a Physics-Informed Neural Network for Thermographic Data Analysis

报告时间：2023年5月30日（周二）15：00

报告地点：C1-437

主讲人：姚远 教授

报告摘要：

1. Small Data Integration for Process Modeling by Using Deep Neural Network-based Word Embedding

In the big data era, small data problems still exist in many industrial sectors. Taking the high-value process industries as an example, a large number of materials and processing methods are often tested at the design stage. However, only a small amount of data can be collected for each material-process combination, which poses a serious challenge to data-driven process modeling. There is a great necessity to integrate the small data measured in different tasks and build the process model by sharing the information. In this work, a deep embedding neural network is proposed to extract the qualitative task information for process modeling. Specifically, an autoencoder is used to learn embeddings which are combined with the quantitative process conditions as the inputs of a feed-forward neural network to produce the final predictions. The feasibility, including interpretability and prediction accuracy, of the developed method is illustrated with an extrusion process.

2. Nondestructive Defect Detection in Polymer Composites using a Physics-Informed Neural Network for Thermographic Data Analysis

The non-destructive testing technique of pulsed thermography has been widely used in materials to detect defects due to its low cost, large detection area, and rapid implementation. After conducting experiments, data processing is often necessary to reduce the influences of noise and non-uniform backgrounds and highlight defect information. However, during the analysis, physical information is usually ignored. In addition to thermographic data analysis, numerical simulations are also popular for analytical studies based on physical information, but they don’t fully utilize the experimental data. To address this, a new method was proposed using a physics-informed neural network (PINN) for thermographic data processing. PINN combines the prediction capabilities of deep neural networks with physical laws presented as partial differential equations and boundary conditions, allowing for both experimental data and physics information to be utilized in modelling. In pulsed thermography, the heat transfer is governed by Fourier's law of heat conduction in a three-dimensional system. However, there is a lack of temperature measurements in the depth direction. The proposed method solves this problem by using collocation points generated from Latin hypercube sampling. The PINN model provides a good estimation of the backgrounds in the thermograms, and the features of surface/subsurface defects are highlighted by subtracting the estimated backgrounds from the original thermograms. In the case study, the performance of the proposed method was found to be effective.

主讲人简介：姚远，现任中国台湾清华大学化工系教授。研究领域包括过程数据分析、过程监控、软测量技术、无损检测技术数据分析以及过程控制等。硕士及本科毕业于浙江大学，控制科学与工程专业。博士毕业于香港科技大学。曾先后获中国台湾清华大学青年教师科研奖、最佳青年研究员称号（中国台湾科技部事务主管部门授予）、I&EC Research 2020最具影响力学者称号。至今，共发表SCI论文100余篇，出版著作两部，授权专利12项，主持（及参与）科研项目60余项。