Improvement of Statistical Estimation of Ship Performance in Actual Seas by Normalization of Data Unevenness Using Cluster Analysis

Information of real phenomena in actual seas are all appeared in on-board monitoring data. However, this data is unevenly distributed according to the operation policy, and there is no guarantee that information completely covers real phenomena. In order to solve this problem, the authors proposed a statistical model with added constraints based on physical consideration not based on monitoring data, and tried to estimate the shaft power component in each sea condition, but the result were not enough. Therefore, by using cluster analysis, the uneven distribution of monitoring data was clarified, and adding the influence to statistical model is proposed. In this paper, after clarifying the problem of previous model, the monitoring data objectively separated by cluster analysis is presented. As a result, it was confirmed that the data has uneven distribution, and by redefining a statistical model considering the influence, it was showed that uneven distribution on the model was normalized.