FlexBioNeuro
Flexibilisation of Biogas production via machine learning algorithms – FlexBioNeuro
The main goal of the project is the demand-oriented production of biogas based on an intelligent control system. By feeding various substrate mixtures into the main fermenter the gas production is adjusted in a targeted manner. The control is based on a sensor-supported real-time measuring system. The data is taken in a scale of minutes.
The control system is designed to enable biogas plant operators to maximize revenues on the power exchange (e.g. EPEX Spot SE – European Power Exchange Spot SE) in the day-ahead market and possibly in the intra-day market while minimizing the use of substrate.
For example, a load profile is transferred to the control model, which is derived from participation in the day-ahead market for the following day. From this load profile, feeding quantities for the substrate (corn, liquid manure, dung) are calculated, taking into account the process limits of the biogas plant by fuzzy logic.
Target for the control model:
- Specification of the required load profile (from day-ahead auction or intraday auction)
- Calculation of the feeding intervals and the mass flows for feeding in order to be able to run the desired load profile with as little material input as possible
- Design a feeding schedule recommendation for the operator of the plant
Control system based on machine learning:
The control model is designed to map the complex, biological processes of this plant concept on the basis of machine learning algorithms. For this purpose a data set of around 100 trials will be generated.
Data generation via online measurements at a real plant:
In order to be able to specify boundary conditions for the control model, which are important for process stability in the hydrolysis reactor and in the fermenter, the following parameters are measured online in both vessels:
- pH value (pH electrode
- dry matter content (NIR sensor)
- Acetic acid concentration(NIR sensor)
- FOS/TAC
- Temperature (pt 100)
The research project aims to provide practical solutions in the field of measurement and control technology. The concept will be transferable to other plants, but the creation of a certain amount of training data will be necessary for each plant. These solutions serve to make biogas production more flexible and thus improve the utilization of existing gas storage capacities.
Funding:
We are very grateful for the support and promotion of the project by the Agency for Nutrition and Agriculture.
Project duration: 01.11.2020 - 31.10.2023
Funding reference: 2220NR046A
Contact persons:
Bernhard Huber, M.Sc.
Lingga Aksara Putra, M.Sc.