船舶能效数据清洗方法研究

A Study on Data Cleaning for Energy Efficiency of Ships

船舶能效数据清洗对于建立准确的船舶能效模型,提高船舶能效计算和分析的准确度,指导船舶节能航行具有重要意义。对船舶能效数据的故障特征进行了分析,运用阈值理论、船舶航行关联理论对故障数据进行了识别,运用插值法和灰色关联理论方法对水深、对水航速以及主机油耗数据中的异常数据进行修正,并且根据数据错误特征和数据清洗方法制定了船舶能效数据清洗流程,以内河邮轮"凯娅号"船舶上安装的船舶能效数据采集系统采集到的能效数据为研究对象,对数据清洗结果进行了分析。结果表明,水深数据修正值与实测值平均误差为1.58%,油耗数据、对水航速数据修正值与实测值平均误差为2.8%,1.5%,误差较小,能够满足后续建模以及数据挖掘等工作。

Data cleaning for energy efficiency of ships is of great importance to develop an accurate energy efficiency modeling as well as to improve the accuracy of its calculation and analysis , which can be used to guide energy-saving navi-gations of ships. Fault features of the data are analyzed and recognized by threshold theory and co-relation analysis. The abnormal data of water depth, ship speed, and fuel consumption are modified by interpolation and grey relational theory. Based on this, a process for data cleaning of energy efficiency is developed. The data collected by the relevant systems on ships are chosen as a case study and cleaning results are analyzed. The results show that the mean error between the modi-fied and measured values of water depth, ship speed,and fuel consumption is 1.58%, 2.8%, 1.5%, respectively. They are satisfactory to meet the requirements of further modeling and data mining.