This paper proposes an approach to improve the performance of an autonomous underwater vehicle (AUV) simulator for inspection missions in the offshore renewable energies (ORE) sector. A complete vehicle with sensors and actuators was modeled and special care has been taken with the effects of proximity to submerged elements. Partial simulations of specific situations are used to generate reduced models of these dynamic behaviors to which artificial neural networks (ANN)-based learning systems are included to create more precise surrogate models that can be dynamically adapted from the tests carried out.
One of the main pillars of the ecological transition necessary to contain climate change is a very important increase in renewable energy generation and use. Offshore wind power has the highest growth potential of any renewable energy technology. If current policies continue, forecasts indicate that 235 GW of new offshore wind capacity will be installed over the next decade worldwide. That capacity is seven times the current market size and represents a 15% increase over the previous year's forecast. But offshore wind today is only 2% of what the world would need to achieve net-zero emissions by 2050.
All these offshore wind turbines installed in large numbers will need a maintenance program with adequate inspections to extend their life in service and guarantee the current and future operability of the platforms. For instance, in the offshore wind farms located in German waters, the support structures of each turbine are inspected at four-year intervals. Some other operators perform these inspections every two to three years to ensure acceptable reliability levels.
These high rates allow everyone to comprehend the high volume of inspection operations required for the near future. Also, the wind turbines are located far out at sea, they operate without humans on board, and the environmental conditions are often adverse.
