Wave Energy Converters (WEC) are designed first to absorb the energy from ocean waves and second, to convert the absorbed energy into electricity toward the grid. Associated challenges arise from the high cost of electricity generated by WECs compared to more mature technologies such as wind and solar. There are two main ways of increasing this cost: the first is to increase the survivability of the device and associated maintenance costs, and the second is to increase the energy absorption of the device. The latter requires a state-of-the-art control algorithm designed towards energy absorption maximization. Nowadays, such algorithms involve model-based reinforcement learning and model-based predictive control where the accuracy and computational speed of the WEC model are keys in the performance of the control policy. However, WECs are deployed in a highly stochastic environment where non-linear forces and dynamic occurs. Most of the models used for control involve BEM solvers for the body/wave interaction which include strong assumptions such as small waves and small displacement from the equilibrium position. In particular, in a controlled environment, the WEC displacement around equilibrium is typically amplified and hence the above assumption doesn't hold. On the other hand, higher fidelity models are based on CFD solvers which have extremely heavy computation time and their use in real-time for control cannot be considered. This creates a need for models that are both fast and accurate. To this extent, data-driven models present an interesting potential given enough data. This project aims to develop a data-driven model that would fulfill the requirements described above, speed and accuracy, for real-time model-based control. The dataset will be generated from a low-fidelity but fast model based on linear potential flow theory, in order to provide a large amount of data. To this extent, it is worth noting that the dataset itself doesn't represent the WEC dynamics in a high-fidelity manner, hence, rather than trying to describe with accuracy the WEC dynamics, the scope of the project is to derive a data-driven model able to predict accurately a given dynamics, given the wave excitation force and the actuation force, regardless of the data quality compare to real WEC dynamic.