Research On Data Mining Technology Of Ship Intelligent Energy Efficiency Management

Intelligent energy efficiency management is one of the six functional modules of intelligent ships.Big data analysis,numerical analysis and optimization technology are important means to realize intelligent energy efficiency management.The harsh and changeable real ship environment results in the distortion or abnormality of some ship energy efficiency data collected;The relationship between energy efficiency data is critical to intelligent energy efficiency management and evaluation.Therefore,it is of great significance to study the preprocessing and data mining technology of real ship energy efficiency data,and the massive data accumulated by ships over a long period of time provides the possibility for data mining.Firstly,by consulting relevant literatures on data preprocessing and data mining and combining the data characteristics of ship energy efficiency,this paper determines to use Python as the main tool for research.Ship data were obtained from ship energy efficiency database,and the main engine,meteorological and other parameters affecting ship energy efficiency were analyzed,and the correlation between them was established.The time interval of energy efficiency data is unified by time interpolation,and the vacancy value is processed.Then,according to the characteristics of energy efficiency data,an anomaly detection model based on principal component analysis is established.After the data are standardized,the parameters are transformed into the main components,and the data larger than the detection threshold are eliminated by using the PauTa Criterion.Then,taking the energy efficiency data of one voyage of the target ship as the input,the number of parameters of the data set to be tested is controlled,and the influence of the number of principal components on the detection accuracy is studied.One year’s data(7 parameters)of the target ship was used for verification,and the verification results were as follows:average accuracy P = 93.95%,recall rate R = 86.68%,F = 90.17%,and the detection effect was relatively ideal.It shows that this method can obtain clean and available data,which can provide guarantee for the following data mining.Finally,the data of a certain voyage of the target ship is selected for data mining.Gaussian mixture model and EM algorithm were used to cluster the parameters in each working condition,and the clustering center of each parameter was obtained,which represented the energy efficiency status in each working condition.The change rule of the clustering center of the ship in a certain working condition in 3 years is analyzed,and the energy efficiency change in this working condition is obtained.Using the energy efficiency data of three working conditions of the same voyage with the corresponding sea state information for correlation analysis,three groups of correlation coefficient matrices were obtained,which represent the degree of correlation between every two parameters.The relationship between fuel consumption index and energy efficiency parameters and the influence of sea state information on ship energy efficiency can be obtained by comparison.In this paper,the research on data mining technology of big data of ship energy efficiency is universal,and the relationship law of each parameter obtained can provide auxiliary decision-making significance for intelligent energy efficiency management and evaluation.