Adaptive baseline model for autonomous marine equipment and systems. (June 2021)

• Record Type: Journal Article
• Authors: Zhang, Peng; Sun, Peiting; Zhang, Yuewen; Jiang, Xingjia
• Is Part Of: ISA transactions. Volume 112(2021)
• Page Start: 326
• Page End: 336
• Publication Date: 2021-06
• Subjects: Autonomous shipping -- Baseline -- Reference-site method -- Dynamic kernel model -- Sliding window; Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2020.12.010
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: With the rapid development of the Internet of Things (IoT) and the Fourth Industrial Revolution, marine equipment and systems are becoming increasingly automated and autonomous. Judging the status of equipment and systems for autonomous shipping assumes that the benchmark of status evaluation is not easily obtained, and the performance baseline for the benchmark is usually static and cannot be accurately adapted under dynamic operating conditions. This paper deals with the issue of establishing a baseline for marine equipment and systems by using a data-driven method. Considering the working conditions of marine equipment and systems, a reference-site (R-S) model was first proposed to determine the initial baseline. This method could solve the problem of inadequate parameters in the initial state very well. Then, a dynamic kernel (D-K) model was used to increase the number of reference sites and update the reference points. This method reduced the amount of data calculation in the process of a dynamic update of the baseline. Continuously fitting the reference points enabled the dynamically updated performance baseline to cope with the working conditions. To implement the proposed method, the index parameters in the R-S and D-K models were processed, and the sliding window capacity was determined using the Kolmogorov–Smirnov method. Finally, the proposed baseline model was applied to a practical case of the exhaust temperature of a marine diesel engine. The result revealed that the proposed method yielded a more accurate baseline and better adaptive performance. Highlights: An adaptive baseline model based on reference-site and dynamic kernel is proposed. Reference-site method to establish initial baseline under the condition of insufficient data is proposed. Dynamic kernel model considering fully the impact of new data is applied to update baseline.