Structural health monitoring: Frequency prediction for offshore platform by polynomial regression model

Nowadays, the data-based method in Structural Health Monitoring (SHM) is widely used in engineering to provide rich information about damage detection. However, implementation of the data-based SHM method for offshore platforms still faced several challenges, such as higher costing due to equipment and workforce, lengthy time assessment, the complexity of equipment and difficulty in knowing the future state of the platform due to periodic monitoring. Therefore, one approach to solving this problem is to develop models based on data from current data-based methods for predicting structural health monitoring for offshore platforms. The development of polynomial models based on the backward method from the fifth-order to first-order used the relationship between average maximum wave height and natural frequency in both directions. The performance of the model was evaluated with overall model significance, statistical analysis and regression parameter significance. As a result, the second-order equation is achieved by parameters in the east-west direction. In contrast, outputs in the north-south direction use the first-order polynomial equation to predict the data. Therefore, to verify the effectiveness of the suggested technique, the predictive value was compared with the current structural health monitoring output data. The result indicates that the predictive value holds higher accuracy with an overall accuracy of 99%. Thus, these results demonstrate that the selected produced equation from the regression method can be the most efficient model for predicting the offshore platform's natural frequency in both directions.