Systems of the present day are getting increasingly complex with a higher degree of autonomy in performing safety-critical tasks. Digital twins, that are capable of representing these systems digitally throughout their life cycle, are slowly replacing traditional simulation models. Technologies like IoT, ML, AI and Cloud Computing are making this possible faster than ever. Digital twins are already a reality in many industries showing their technical and commercial potential. On the other hand, Unmanned Surface Vehicles(USV) - a type of autonomous marine system are getting popular in the robotics community for their ability to operate in complex and remote environments. This makes them suitable candidates for digital twin development. However, to the author's knowledge, the power of digital twins has not yet been fully realised in the case of USVs. The current thesis addresses this gap with an aim to develop a digital twin for Otter - an Unmanned Surface Vehicle from Maritime Robotics, Norway. In doing so, two objectives are considered. Firstly, ‘Probabilistic Graphical Models' are used as a mathematical framework for the digital twin due to their capability of handling the complexity and uncertainty of systems. Also, these models provide a general framework that can accommodate different modelling techniques making them suitable for several digital twin applications. Secondly, the above framework is used to perform ‘Actuator Fault Diagnosis’ on Otter's actuators as a demonstrative application of the digital twin. The necessary theoretical treatment is given to both the above concepts. Faults were introduced into the actuator subsystem by means of a broken port side propeller and experiments were conducted on both the faulty and faultless Otter. Subsequently, the faults were identified using an Adaptive Extended Kalman Filter algorithm with promising results. The results are thoroughly discussed with regard to both their meaning and the value they add in the context of digital twins. Finally, recommendations for future work are made.