Improved merit order and augmented Lagrange Hopfield network for unit commitment

This paper proposes an improved merit order (IMO) and augmented Lagrange Hopfield network (ALHN) for unit commitment (UQ). IMO is a merit-order method which is based on average production cost of generating units improved by heuristic search algorithms, whereas ALHN is a continuous Hopfield neural network with its energy function based on augmented Lagrange relaxation. The proposed IMO-ALHN solves UC problem in three stages. In the first stage, IMO is applied for unit scheduling. In the second stage, ALHN is used to solve ramp rate constrained economic dispatch (RED) based on the obtained unit schedule, and a strategy for repairing ramp rate constraint violation is performed if a feasible solution is not found. In the last stage, a heuristic search for unit decommitment is applied on the obtained solution from RED for further improvement and ALHN is again applied to solve RED if there is any change in the unit schedule. The proposed method is tested on systems up to 1000 generating units with schedule time horizon up to 168 h. Test results indicate that the proposed method is very attractive and favourable over many other methods due to substantial production cost savings and faster computational times.