Investigation On Experimental Data Denoising In Ocean Engineering Based On Discrete Wavelet Transformation

As a mathematical analysis tool, wavelet transform is now widely used in many research fields. Physical experiment is an important way in the field of ocean engineering. However, the data collected in experiments is always contaminated by noise. How to eliminate noise effectively is a key issue to guarantee the validation of the following-up data analysis. In this dissertation, the wavelet-based denoising method is adopted in the data analysis for an experiment of a single point soft yoke moored Floating Production Storage and Offloading (FPSO) system. By detailed analysis and comparison, the best combination of the choices of mother wavelet, threshold method and reset mode are determined for denoising data collected in physical experiments.The basic concept and theory of discrete wavelet transform (DWT) is introduced firstly. Then, the theory of denoising is presented in detail. Especially, the characteristics of the common used mother wavelet, the parameters used in noise reduction and the assessment criteria are discussed at great depth.Following the denoising theory of discrete wavelet transform, the experimental set-up for the single point soft yoke moored FPSO and the circumstance conditions are introduced. The data collected in a test for the free attenuation of the FPSO is used as the benchmark to assess the ability of the denoising method based on DWT. By adding known white Gaussion noise to the measured data and then denoise the contaminated single, the best combination for denoising the data of ship motion is db8 and sym10 wavelet functions, Stein unbiased risk threshold method, sln reset mode. Then the parameter combination is used in the reduce noise for the FPSO motion data collected in the experiments in different circumstance conditions, and the results indicate that the method can effectively reduce the noise.Lastly, the denoising method is applied in the analysis of the collected mooring forces. The results show that the DWT-based method not only can reduce the noise but retain the transient impact forces. However, the low-pass filter method erases the transient impact forces and hence the processed data cannot reflect the real phenomenon. The optimized choice of parameters in DWT denoising method for mooring forces is also obtained based on the experimental results.