Neural networks and fuzzy systems: a dynamical systems approach to machine intelligence
Neural networks and fuzzy systems: a dynamical systems approach to machine intelligence
Neural Networks and Simulation Methods
Neural Networks and Simulation Methods
Time Series Analysis, Forecasting and Control
Time Series Analysis, Forecasting and Control
| Hi-index | 0.00 |
Continuous collection of wave data is being increasingly practiced since recent past all over the world. This is done in order to provide an essential input to carry out various operational civil engineering works in the oceans. Based on such continuous observations at the specific site, on-line prediction of wave heights can be made using different analytical schemes. This paper evaluates performance of several such schemes. The theoretical approaches involved are stochastic techniques of Kalman Filter, ARIMA, ARMA, AR and also that of neural networks. No single technique emerged as supreme when wave predictions over different short term periods were desired. Lower prediction intervals indicated superiority of neural networks, while higher ones were better tackled by the stochastic technique of Kalman Filter. The models with a very high degree of number of freedoms performed better than those with smaller flexibility of fitting.