The balancing act: Biologically driven rapid-response automation of production conditions in recirculating aquaculture systems (RAS)

RAS 4.0 is a 4-year project with Nofima as a project owner and NORCE, UiT, the Arctic University of Norway, Searis, CreateView, Krüger Kaldnes, OxyGuard and Lerøy Seafood group as partners. The main objective of RAS 4.0 is to improve fish wellbeing and the production efficiency of RAS and reduce operational risks by developing integrated control systems for water quality, feeding and energy optimisation. The control systems will be based on novel sensors, data integration and smart algorithms that combine biological, environmental, and operational factors. The RAS 4.0 project will create highly valuable knowledge that can be used by aquaculture and industry suppliers working to maximise the sustainability of these operations. Realising the potential of RAS will reduce the pressure to increase production in the sea and secure increased investment in this important, potentially environmentally sustainable solution to fish production. Our work focused on the establishment of a digital infrastructure for the project that includes different water quality sensors, data standards, data platforms and their integration into one data platform for better data visualization, data capture and analysis. We have determined necessary water visibility in RAS for optimal operation of camera technology that has allowed us to successfully collect several months of video footage related to different feeding activity of Atlantic salmon in RAS. We are currently using machine learning to identify behavioural indicators of appetite and satiation in groups of salmon held in RAS conditions. Indicators are being analysed by the partners i) around meals and ii) in the build up to feeding as indicators of potential food anticipatory activity. In addition, we observed that tank soundscapes are affected by different feeding regimes in RAS. An experiment on energy use for carbon dioxide removal has been conducted and as a result a model of the energy use of degassing has been developed. The model will be tested against the experimental data and further utilized to study energy optimization. Machine learning models have been implemented for predicting multi-variant, multi-time step time-series data, with carbon dioxide as the target variable. The same approach can be applied to predict other parameters. These models were trained and tested using data from different time periods, achieving approximately 91% prediction accuracy on the test dataset.

The world is searching for sustainable solutions to produce healthy food. Aquaculture has been tasked to meet this demand. Recirculating aquaculture systems (RAS) have many sustainability advantages over open sea cages e.g. no interaction with wild fish, reduced water use and collection of waste. However, the potential of RAS to alleviate the demand for healthy fish is hindered by reoccurring unpredictable production losses, high costs and operational risk. While digitalisation and automation are on the forefront of priorities to improve production efficiency and reduce the environmental impacts of aquaculture, no production system has the same potential to exploit these enabling technologies as RAS. New automated approaches, making use of sensors and control algorithms, are required to realise the sustainability potential of RAS. The RAS 4.0 project brings together a multidisciplinary team to produce balanced and bio-inspired control loops on the most critical operations, namely water quality, feeding efficiency and energy optimisation. These loops will be integrated into a digital twin that will focus on optimising RAS performance and will be validated during RAS operations. The project utilises the expertise of industry partners established technologies and expertise from sensor, data standards and integration. Together with leading research partners in RAS technology, fish physiology, behaviour, welfare, machine learning, data analytics, smart cameras and computer vision they will work to develop smart digital approaches that seamlessly connect and optimise the physical, digital, and biological world. The RAS 4.0 project will create highly valuable knowledge that can be used by aquaculture and industry suppliers working to maximise the sustainability of these operations. Realising RAS potential will reduce the pressure to increase production in the sea and secure increased investment in this important environmentally sustainable solution to fish production.