The ocean is one of the greatest untapped renewable energy sources on earth, where a wave energy converter (WEC) may be used to extract a substantial amount of energy and convert it into electricity. However, for WECs to be economically competitive with other renewable resources a number of challenging engineering problems must be dealt with. Power maximizing is one way to reduce its Levelized Cost of Energy (LCOE) compared to established energy resources. This can be achieved by tuning a real-time control system of the WEC according to the instantaneous incident wave, for which the future knowledge of excitation force or wave elevation is required in order to make the control problem causal. This research investigates the methodologies to estimate and predict the excitation to be used in CorPower Ocean (CPO) WEC control system.
An Auto-Regressive (AR) model is used to predict the excitation force for a required prediction horizon based on only its past history. The required horizon length is computed by the effect of suboptimal velocities, i.e. replacing the full noncausal reference transform function by truncated versions in the convolution with excitation force. To create the history for the AR model, a Kalman Filter with Harmonic Oscillator (KFHO) estimates the excitation force based on a simple WEC model and measured values for position and velocity of the WEC.
The accuracy of the predicted signal compared to the measured signal is presented for two different time horizons. The results display that the performance of the predictor/estimator is highly dependent on the predictor parameters and the optimum number of frequencies which identifies the dynamic of the excitation force. The proposed method of combined prediction and estimation has shown around 96.6% and 99.4% accuracy compared to filtered estimated force as well as 79.9% and 77.6% accuracy compared to measured excitation force for prediction horizons of 5 seconds and 2.5 second respectively.