Applying reinforcement learning and tree search to the unit commitment problem

Recent advances in artificial intelligence have demonstrated the capability of reinforcement learning (RL) methods to outperform state art decision-making problems under uncertainty. Day-ahead unit commitment (UC), scheduling power generation based on forecasts, is a complex systems task that becoming more challenging light increasing While RL promising framework for solving UC problem, space possible actions from given exponential number generators and it infeasible apply existing larger than few generators. Here we present novel algorithm, guided tree search, which does not suffer an explosion action with The method augments search algorithm policy intelligently reduces branching factor. Using data GB system, demonstrate outperforms unguided terms computational complexity, while producing solutions show no performance loss operating costs. We compare against mixed-integer linear programming (MILP) find solution using reserve constraints, current industry approach. exhibit behaviours differ qualitatively MILP, demonstrating its potential as decision support tool human operators.