基于数据驱动方法的船舶CII预测模型研究

Review of the Research on Vessel’s Carbon Intensity.Indicator Prediction Models Based on Data-driven Approaches

相比通过物理模型和数值计算进行碳强度指数(Carbon Intensity Indicator, CII)预报,采用数据驱动方法构建模型进行CII预报能避免消耗大量计算资源。在分析已有研究的基础上,总结机器学习、时间序列等数据驱动方法在构建此类预测模型中应用的优势和劣势,以及运用这些方法构建的船舶CII预测模型。对比分析各种算法的优势和劣势可知,运用决策树算法和整合自回归移动平均算法构建船舶CII预测模型的研究目前虽然较少,但其颇具研究和创新空间,开展此项研究有助于在未来构建更准确、高效的船舶CII预测模型。

Building models based on data-driven approach for Carbon Intensity Index(CII)prediction can avoid consuming large amount of computing resources compared with CII forecasting using physical models and numerical calculations.On the basis of literature research,this paper summarizes the advantages and disadvantages of data-driven approaches such as machine learning and time series analysis,and also the existing vessel’s fuel consumption and CII prediction models built based on them.The summary indicates that the research on building vessel’s CII prediction models based on decision tree(DT)and autoregressive integrated moving average(ARIMA)algorithms is far from sufficient,given the unique advantages of DT&ARIMA algorithms,which means plenty of research and innovation work can be carried out in this area,and such work will be conducive to the construction of more accurate and efficient ship CII prediction model in the future.