Session: 02-05-01 Data-driven and AI-based Models

Submission Number: 181815

Fatigue Assessment by Feeding Back the Results of Stress Measurement 

 In the spectral fatigue analysis using AIS and Hindcast data, the correction for the average deviation between the characteristic value of analyzed variable stresses (average power of the stress response spectrum) and the characteristic value of actual variable stresses (variance) in the short-term sea state and the definition of the stress range to calculate the short-term fatigue damage due to the variable stresses by Rayleigh distribution were examined based on the stress measurements and the fatigue damage by the rain-flow method. In this case, the cumulative fatigue damage can be accurately evaluated regardless of the short-term stochastic variation of sea state and operation condition by considering the representative values in a certain period. If the representative values in a certain period can be estimated by the statistical values of sea states and operation conditions in that period, the cumulative fatigue damage in the operation without measurements can be evaluated by obtaining AIS and Hindcast data. The models to estimate the correction coefficient for the characteristic value of analyzed variable stresses and the coefficient to define the stress ranges were established by machine learning from the data in the stress measurements period. It was confirmed that the cumulative fatigue damage can be accurately evaluated by estimating the values of these coefficients from the AIS and Hindcast data in the operation without measurements and in the operation of different routes. As a machine learning method, the comparison between multiple regression method and neural network method was examined from the viewpoints of learning effect and prediction accuracy.

Presenting Author: Norio Yamamoto Nippon Kaiji Kyokai

Presenting Author Biography: After graduating from university, started working at NK.
At NK, mainly involved in the research institute and in the rule development department.
Have been dealing with issues related to the ageing of structures, such as fatigue and corrosion, and structural safety assessment for many years, and has also developed fatigue rules and corrosion relating rules for hull structures. Have also been involved in IACS activities for many years and has been involved in the development and maintenance of CSR fatigue rules and corrosion addition rules.
Recently, have been investigating fatigue prediction as a topic related to Digital Twin.

Authors:

Norio Yamamoto Nippon Kaiji Kyokai