Intelligent short-term forecasting for mud concentration in CSD dredging construction

Shuai Han, Heng Li, Mingchao Li, Huijing Tian, Liang Qin, Yi Yu, Jie Ma

Research output: Journal article publicationJournal articleAcademic researchpeer-review

4 Citations (Scopus)

Abstract

It has long been a challenge in the design of cutter suction dredgers (CSD) that the measurement of mud concentration lags behind other indicators like vacuum and flow rate. This phenomenon is named the time-lag effect. It significantly hinders the optimization and automation of CSD operation, for the mud concentration is a primary reference during CSD operation. As such, this study presented a data mining-based solution for short-term forecasting of mud concentrations. To accomplish this goal, first, an integration method based on hydraulics principles was proposed to align mud concentrations with other indicators over time. Second, normal construction identification, data filtering, and data smoothing were used to preprocess data. Third, a feature selection process, named "indicator class", was applied regarding the construction technology and the characteristics of the pump-pipeline system of CSDs. Last, a short-term forecasting algorithm was presented based on a hybrid ensemble strategy; this established the relationship between the actual mud concentration and the selected indicators. The validity of this algorithm was tested by a case study that collected a group of raw data logged by a CSD monitoring system. The result showed that the real-time prediction could reach a high accuracy where the coefficient of determination (R2) is 0.886 and the mean square error (MSE) is 4.708, and the short-time forecasting result within 12s is consistent with the actual trend of the mud concentration. The proposed approach has a significant deal of potential to enhance the quality and efficiency of operations by providing operators with essential references to real-time mud concentrations.

Original languageEnglish
Article number113151
JournalOcean Engineering
Volume266
DOIs
Publication statusPublished - 15 Dec 2022

Keywords

  • Dredging construction
  • Forecast
  • Machine learning
  • Mud concentration
  • Time-lag effect

ASJC Scopus subject areas

  • Environmental Engineering
  • Ocean Engineering

Fingerprint

Dive into the research topics of 'Intelligent short-term forecasting for mud concentration in CSD dredging construction'. Together they form a unique fingerprint.

Cite this