Real-Time Decision Making for Anaerobic Digestion of Food Wastes Using Machine Learning Approach 

Investigator: Oregon State University 
Start Date: November 2023 
Award Amount: $152,000 

Food waste makes up approximately 22% of municipal solid waste generated in the United States, and over 90% of food waste is ultimately landfilled. Anaerobic digestion (AD) of food waste can effectively decrease the volume of waste and produce an alternative energy source. However, due to the diversity and variability of food waste and complexity of the AD microbial process, adopting AD for food waste management still faces many technical and economic challenges. The current decision-making schema for AD is still based on a reactive approach. Determining the best course of action is difficult due to the multitude of dynamic factors that must be considered. Data-driven models such as machine learning (ML) models have the potential to provide forecasts of projected AD performance and digester instability and real-time control strategies that operators can use to respond proactively based on the current state of the digester. 

The overarching goal of project is to develop an ML-based real-time decision making tool to help operators manage the AD of food waste. More specifically, the following objectives will be achieved: 

1. Develop datasets through monitoring waste properties, operational conditions, microbial communities, and AD performance. 

2. Develop real-time prediction models for forecasting AD performance and process instability. 

3. Develop a real-time control model, based on historical and current data, for decision support of human operators in optimizing digester performance.