Hydrogen Flow Rate Estimation in Fuel Cells using Machine Learning

Abstract

This paper proposes a method for hydrogen flow rate estimation in fuel cells using time-series prediction with machine learning based on existing measurement values such as fuel cell output power, output current, and purge timing. In the proposed method, the appropriate combinations of feature quantities and sequence length are selected using a genetic algorithm. As a result, the proposed method reduces the mean absolute error (MAE) by approximately 50% compared to the linear approximation between the current and hydrogen flow rate.