English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Model predictive control (MPC) is increasingly being considered for controlof fast systems and embedded applications. However, the MPC has somesignificant challenges for such systems. Its high computational complexityresults in high power consumption from the control algorithm, which couldaccount for a significant share of the energy resources in battery-poweredembedded systems. The MPC parameters must be tuned, which is largely atrial-and-error process that affects the control performance, the robustnessand the computational complexity of the controller to a high degree. In thispaper, we propose a novel framework in which any parameter of the controlalgorithm can be jointly tuned using reinforcement learning(RL), with the goalof simultaneously optimizing the control performance and the power usage of thecontrol algorithm. We propose the novel idea of optimizing the meta-parametersof MPCwith RL, i.e. parameters affecting the structure of the MPCproblem asopposed to the solution to a given problem. Our control algorithm is based onan event-triggered MPC where we learn when the MPC should be re-computed, and adual mode MPC and linear state feedback control law applied in between MPCcomputations. We formulate a novel mixture-distribution policy and show thatwith joint optimization we achieve improvements that do not present themselveswhen optimizing the same parameters in isolation. We demonstrate our frameworkon the inverted pendulum control task, reducing the total computation time ofthe control system by 36% while also improving the control performance by 18.4%over the best-performing MPC baseline.

Optimization of the Model Predictive Control Meta-Parameters Through Reinforcement Learning