数据驱动的软测量方法研究: 用于车身加速度和垂向轮心力估计
日期：2022/10/18 - 2022/10/18
博士论文答辩：数据驱动的软测量方法研究: 用于车身加速度和垂向轮心力估计
主讲人：Xueke Zheng, Ph.D. candidate at UM-SJTU Joint Institute
时间：2022年10月18日（周二）下午2:00
讲座摘要
When designing a new vehicle, durability performance can be assessed by analyzing the information of vehicle body accelerations and vertical wheel forces, which are measured by a large number of expensive sensors. With constraints of limited time and cost, such instrumentation is typically affordable only for a very limited set of vehicles or testing activities. To improve the quality of products and accelerate the vehicle design phase, it is appealing to resort to alternative ways of obtaining vehicle body accelerations and vertical wheel forces without directly measuring them by sensors.
The objectives of this work is to develop simple and effective data-driven soft sensors for the estimation of vehicle body accelerations and the vertical wheel force. Four research topics are proposed in this dissertation: a) Sensor-to-sensor transmissibilities and discrete mode observability of switched linear systems are investigated to provide the theoretical foundations of the proposed data-driven soft sensors; b) a frequency-domain soft sensor based on piecewise linear frequency-domain identification is proposed to estimate vehicle body accelerations; c) a time-domain soft sensor based on a primary-auxiliary model scheduling procedure is proposed to estimate vertical wheel forces; d) and finally, a sensor selection method based on convex optimization is proposed to optimally allocate the sensor resource in the time-domain soft sensor. It is believed that the present results give new insights of soft sensors, and make an important step for building a theoretical framework of soft sensors in complex engineering systems.
主讲人简介
Xueke Zheng received his B.S. degree from Northeastern University (Shenyang), in 2014. Then, he received M.E. from School of Aeronautics and Astronautics, SJTU, in 2017. He is currently a PhD candidate at UM-JI Joint Institute, SJTU, supervised by Professor Mian Li. His current research focuses on system identification of complex engineering systems.