Fleet Control using Coregionalized Gaussian Process Policy Iteration
This addresses the challenge of sample-efficient control in fleets like wind farms, where production errors or degradation prevent simple data aggregation, offering a novel approach to knowledge transfer over system specifications rather than tasks.
The paper tackles the problem of controlling fleets of similar machines with small discrepancies by proposing a reinforcement learning method that transfers knowledge between members using Gaussian processes, achieving significantly better median and variance results compared to individual and joint learning baselines.
In many settings, as for example wind farms, multiple machines are instantiated to perform the same task, which is called a fleet. The recent advances with respect to the Internet of Things allow control devices and/or machines to connect through cloud-based architectures in order to share information about their status and environment. Such an infrastructure allows seamless data sharing between fleet members, which could greatly improve the sample-efficiency of reinforcement learning techniques. However in practice, these machines, while almost identical in design, have small discrepancies due to production errors or degradation, preventing control algorithms to simply aggregate and employ all fleet data. We propose a novel reinforcement learning method that learns to transfer knowledge between similar fleet members and creates member-specific dynamics models for control. Our algorithm uses Gaussian processes to establish cross-member covariances. This is significantly different from standard transfer learning methods, as the focus is not on sharing information over tasks, but rather over system specifications. We demonstrate our approach on two benchmarks and a realistic wind farm setting. Our method significantly outperforms two baseline approaches, namely individual learning and joint learning where all samples are aggregated, in terms of the median and variance of the results.