Approximate policy iteration using neural networks for storage problems

TL;DR

This paper enhances approximate policy iteration for energy storage problems by integrating neural networks, especially effective when storage charging efficiency is low, using a novel aggregate variable model to reduce decision complexity.

Contribution

Introduces a new aggregate variable model enabling neural networks to improve policy evaluation in SNES, demonstrating significant performance gains, notably with low charging efficiency.

Findings

Neural networks improve API performance in SNES.

Aggregate variables reduce decision vector dimensionality.

Performance gains are more pronounced with low charging efficiency.

Abstract

We consider the stochastic single node energy storage problem (SNES) and revisit Approximate Policy Iteration (API) to solve SNES. We show that the performance of API can be boosted by using neural networks as an approximation architecture at the policy evaluation stage. To achieve this, we use a model different to that in literature with aggregate variables reducing the dimensionality of the decision vector, which in turn makes it viable to use neural network predictions in the policy improvement stage. We show that performance improvement by neural networks is even more significant in the case when charging efficiency of storage systems is low.