Soft sensor development based on kernel dynamic time warping and a relevant vector machine for unequal-length batch processes
作者：
Highlights：
• A kernel DTW algorithm combining the kernel trick is designed.
• An adaptive selection strategy for the optimal parameter is proposed.
• A novel soft sensor is proposed for unequal-length batch processes.
• A penicillin fermentation process is used to verify the proposed soft sensor.
• The proposed soft sensor has better performance than other soft sensors.
摘要
•A kernel DTW algorithm combining the kernel trick is designed.•An adaptive selection strategy for the optimal parameter is proposed.•A novel soft sensor is proposed for unequal-length batch processes.•A penicillin fermentation process is used to verify the proposed soft sensor.•The proposed soft sensor has better performance than other soft sensors.
论文关键词：Batch processes,Unequal-length data,Soft sensor,Dynamic time warping,Trajectory synchronization,Relevant vector machine
论文评审过程：Received 20 December 2019, Revised 28 March 2021, Accepted 14 May 2021, Available online 19 May 2021, Version of Record 21 May 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.115223
