Reservoir optimization and Machine Learning methods

TL;DR

This paper explores the use of neural networks for reservoir optimization, demonstrating their superiority over traditional feedforward networks in estimating Bellman values and proposing a hybrid algorithm combining LP and neural network-based cuts for stochastic linear problems.

Contribution

The paper introduces novel neural network architectures tailored for reservoir problems and a hybrid optimization algorithm integrating LP and neural network predictions.

Findings

Neural networks outperform feedforward networks in reservoir Bellman value estimation.

Classical feedforward networks are ineffective for reservoir problems.

A new hybrid algorithm effectively solves stochastic linear problems using neural network-based cuts.

Abstract

After showing the efficiency of feedforward networks to estimate control in high dimension in the global optimization of some storages problems, we develop a modification of an algorithm based on some dynamic programming principle. We show that classical feedforward networks are not effective to estimate Bellman values for reservoir problems and we propose some neural networks giving far better results. At last, we develop a new algorithm mixing LP resolution and conditional cuts calculated by neural networks to solve some stochastic linear problems.