Towards Deep Learning for Predicting Microbial Fuel Cell Energy Output

Adam Hess-Dunlop,Harshitha Kakani,Colleen Josephson

2024-06-18

Abstract:Soil microbial fuel cells (SMFCs) are an emerging technology which offer clean and renewable energy in environments where more traditional power sources, such as chemical batteries or solar, are not suitable. With further development, SMFCs show great promise for use in robust and affordable outdoor sensor networks, particularly for farmers. One of the greatest challenges in the development of this technology is understanding and predicting the fluctuations of SMFC energy generation, as the electro-generative process is not yet fully understood. Very little work currently exists attempting to model and predict the relationship between soil conditions and SMFC energy generation, and we are the first to use machine learning to do so. In this paper, we train Long Short Term Memory (LSTM) models to predict the future energy generation of SMFCs across timescales ranging from 3 minutes to 1 hour, with results ranging from 2.33% to 5.71% MAPE for median voltage prediction. For each timescale, we use quantile regression to obtain point estimates and to establish bounds on the uncertainty of these estimates. When comparing the median predicted vs. actual values for the total energy generated during the testing period, the magnitude of prediction errors ranged from 2.29% to 16.05%. To demonstrate the real-world utility of this research, we also simulate how the models could be used in an automated environment where SMFC-powered devices shut down and activate intermittently to preserve charge, with promising initial results. Our deep learning-based prediction and simulation framework would allow a fully automated SMFC-powered device to achieve a median 100+% increase in successful operations, compared to a naive model that schedules operations based on the average voltage generated in the past.

Signal Processing,Emerging Technologies

What problem does this paper attempt to address?

This paper aims to address the energy output prediction problem of microbial fuel cells (SMFCs). SMFCs are an emerging technology that provides clean and renewable energy for environments that are not suitable for traditional power sources such as chemical batteries or solar energy. However, due to the incomplete understanding of the electricity generation process, SMFCs exhibit significant fluctuations in energy production, limiting their use in applications such as outdoor sensor networks. This paper introduces the use of deep learning methods, particularly the Long Short-Term Memory (LSTM) model, to predict the energy output of SMFCs at different time scales ranging from 3 minutes to 1 hour. Through magnitude regression, the paper not only provides point estimates but also quantifies the uncertainty of these estimates. Experimental results show prediction errors ranging from 2.29% to 16.05%. Furthermore, the study simulates how these models can be utilized in an automated environment to control the intermittent switching of SMFC power supply devices, thereby conserving energy and enhancing the efficiency of successful operations. The proposed deep learning prediction and simulation framework significantly improves the success rate of SMFC power supply device operations, especially in low-power applications.

Towards Deep Learning for Predicting Microbial Fuel Cell Energy Output

Modeling of microbial fuel cell power generation using machine learning-based super learner algorithms

Evaluation of Deep Learning-based prediction models in Microgrids

Microbial Community Predicts Functional Stability of Microbial Fuel Cells.

Data-driven Proton Exchange Membrane Fuel Cell Degradation Predication Through Deep Learning Method

Prediction of Fuel Cell Performance Degradation Using a Combined Approach of Machine Learning and Impedance Spectroscopy

A Multi-Scale Fuel Cell Degradation Prediction Method Based on Isometric Convolution Block and Long Short-Term Memory Networks

Transfer learning-based deep learning models for proton exchange membrane fuel remaining useful life prediction

A Scientific Machine Learning Approach for Predicting and Forecasting Battery Degradation in Electric Vehicles

Enhancing Microgrid Performance Prediction with Attention-based Deep Learning Models

A performance degradation prediction model for PEMFC based on bi-directional long short-term memory and multi-head self-attention mechanism

Degradation Prediction of Proton Exchange Membrane Fuel Cell Based on Bi-LSTM-GRU and ESN Fusion Prognostic Framework

Data-Driven Voltage Prognostic for Solid Oxide Fuel Cell System Based on Deep Learning

Optimal Data-Driven Modelling of a Microbial Fuel Cell

Deep Learning-Based State-of-Health Estimation of Proton-Exchange Membrane Fuel Cells under Dynamic Operation Conditions

A Novel Deep Belief Network and Extreme Learning Machine Based Performance Degradation Prediction Method for Proton Exchange Membrane Fuel Cell

Data-driven Prognostics for PEM Fuel Cell Degradation by Long Short-term Memory Network

Incorporating Microbial Community Data with Machine Learning Techniques to Predict Feed Substrates in Microbial Fuel Cells.

Enhancing Prognostics of PEM Fuel Cells with a Dual-Attention LSTM Network for Remaining Useful Life Estimation: A Deep Learning Model

Short-term performance degradation prediction of a commercial vehicle fuel cell system based on CNN and LSTM hybrid neural network

Interpretable Machine Learning for Predicting Biomethane Production in Industrial-Scale Anaerobic Co-Digestion