个人简介：
严文裔教授予北京航空航天大学获得学士和硕士学位，并于中国清华大学获得博士学位。目前，严教授在澳大利亚莫纳什大学机械与航空航天系工作。他当前的研究兴趣包括铁路结构力学、结构优化、增材制造计算力学。他已指导完成28名博士和6名硕士。严教授总共发表了181篇期刊论文。根据Google Scholar，他的H指数为46。
讲座内容：
Railway in-train forces are an essential element in assessing railway many aspects of rolling stocks. Conventional methods for obtaining the forces, such as field measurements and longitudinal train dynamics simulations (LTSs), can be time-consuming and require significant investment in manpower and domain expertise, while only allowing for the data collection on a single specified service condition at a time. However, automatic train operation (ATO) systems can measure real-time information of trains and tracks by on-board and trackside sensors, which could provide an opportunity for predicting in-train forces. This paper presents a data-driven approach that uses ATO-measured data to predict in-train forces under service conditions. To develop this approach, LTSs for a heavy haul train were conducted to establish the relationship between ATO measurements and specific in-train forces, which was embedded in a large amount of training data. After that, a specially developed self-attention based causal convolutional neural network (SA-CNN) was employed to learn the underlying relationship and predict the in-train force histories by considering the dependencies of current and past time steps. The performance of SA-CNN was compared with four different neural network models, and the predicted results demonstrated that all the well-trained models can accurately predict in-train forces. Furthermore, the generalisation ability of the well-trained SA-CNN model was verified with LTS under four different service conditions. The results showed that the proposed data-driven approach has superior compatibility for any arbitrarily combined inputs with significantly reduced computational time compared to LTSs. This approach has the potential to realise quick and reliable in-situ monitoring of railway in-train forces, which is beneficial to both in-train force related research and industrial applications.
