Wave predictor

further information on wave predictor and HP collaboration. Points of interest
double digit gains
can be used in many other applications

Multi Agent Reinforcement learning MARL controller


The proposed MARL controller yielded double-digit gains over the entire spectrum of waves boosting higher energy production. At the same time, it helped reduce mechanical stress, which impacts maintenance and operating costs, and actively mitigated adverse effects of high waves character- istic of disaster events, helping to preserve ctal invest- ment and cost of manufacturing.This RL controller can be extended in its scope from an in- dividual wave energy converter to a cluster of wave energy converters to address the “cross-talk” issue that is a well- established problem in wind farms. MARL architectures, re- finements, and optimizations described in this paper, with multiple objectives mentioned above, are applicable to many other clean energy problems like wind energy. Further, the proposed MARL architecture and the PPO re- finements to stabilize training for global optima described in this paper can be used in many other complex control appli- cations with multiple actors or entities to control.

link to paper:
hpe.to/6015KkjhT