Please enable JavaScript.

Coggle requires JavaScript to display documents.

Sea State Estimation review , Strategies and series to improve the…

- - - - Response spectrum + RAOs = Wave spectrum
        
        Approaches for SSE
        
        Additional assumptions
        
        Ergodic Random Processes
        
        Stationarity and Stochastic Sense
        
        Constant speed and course
    - - Response +RAO = wave elevation
        
        Data driven
        
        Machine learning
        
        Deep learning
        
        Model-baesd
        
        Stepwise
        
        brute-force
        
        Kalman ﬁlterin
        
        the intrinsic information about the actual wave elevation can be included to the solution
- - - - Parametric approach: Optimize sea state parameters of a parameterized wave spectrum.
      - Bayesian approach: Solve using Bayesian modeling or by minimizing a discrete version of Eq. (1).
- - - - Stationarity and Stochastic Sense
  - - - Adding additional filters to the model : estimating inertia, damping and hydrostatic in real time to improve the accuracy
        
        experimental tests
        
        CFD
  - - - sensor fusion : IMU + GPS
  - - - indirect method : sensor fusion
        
        ship motion + Radar (lack of wave height)
        
        maybe adding wind data and take it into account in the model
- - - - Solutions
        
        Indirect estimation
        
        data from multiple ships Nielsen 2019
        
        challenge: negative impact of distance between vessels/pertinent hydrodynamic characteristics of
        every ship.
        
        Sensor fusion by adding wind data with an anemometer Montazeri 2016
        
        Challenge : sensor errors calibration
        
        Direct estimation
        
        Parametrization
        
        Identify HF tail by fitting standard wave spectrum.
        
        Challenges include data transformation to spectral domain and real-time capturing limitations.
        
        1 more item...
        
        Adding additional filters to the model to revive HF information ( kim2019)
        
        Challenge with Ship-Borne Sensor Data
        
        1 more item...
  - - - Effective Sampling: Using shuffled sampling (Mak, 2019) ensures accurate data distribution for training. must be before analysis step
  - - - Methods of estimating the hydrodynamic coefficients
        
        Panel method
        
        Strip theory
      - Problems and considerations
        
        Viscous impacts on the coef and seakeeping phenomena
        
        uncertainty in inertia terms M specifically in HF signals
        
        Viscosity affects added mass most significantly for heave motion due to the body's width cross-sectional area. the impact of viscosity on acceleration is very negligible,
        
        viscosity introduces nonlinearities
        
        Limitation of Analytical Techniques while adding viscosity terrn
        
        use CFD or experimental tests
        
        longer waves intensify viscosity effects on inertia, leading to nonlinearity and second-order wave loads
        
        Limitations in Linearity Assumption: linear relation between waves and ship responses
      - Solutions
        
        Modeling hydrodynamic coef terms
        
        defining and calibrating hyperparameters using experimental data (Pascoal, 2007; Soler et al., 2020).
  - - - Model based methods are stiff to modifications --> classical methods for calculating RAO have been employed almost in all studies,
      - Data driven is more flexible than model based
        
        Data-driven methods are
        data-hungry,
        
        requires many sensors
        
        demand the high expertise of the designer
  - - - time for computation SSE
        
        effect: Not a concern for most techniques
      - Stabilization the estimation results
        
        it is crucial during transition in seaway and ship conditions
      - Time for capturing sea data from sensors
        
        Effect: complicate estimation and require larger sliding window