ADPAC: The Story So Far
Now that the ADPAC project has been running for around 18 months (we’re just over halfway!), we thought it was time to update everyone on the progress we’ve made so far. The project has been broken down into various work packages including: Advanced Sensors, Data Transmission, Analytics Toolbox and System Integration to name a few. In this article, we’ll be exploring the progress our partners have been making. Let’s take look…
Firstly, it’s important for us to validate that we can accurately measure the parameters that are necessary for modelling and important to the farm operators. Therefore, the first task in this work package was to select appropriate sensors to monitor these parameters and ensure that they were suitable for the application (e.g. sensitivity, power and communications). Chosen parameters include: Temperature, Conductivity, pH, Tryptophan, CDOM, Chlorophyll, Phycoerythrin, Phycocyanin, Turbidity, Absorbance and Nitrate/Nitrite.
Chelsea Technologies have recently developed a new multi-parameter sensor, VLux, that will be incorporated into the project and is able to satisfy 5 of these parameters: Tryptophan, CDOM, Chlorophyll, Turbidity and Absorbance – important parameters with regards to monitoring water quality (bacterial presence/activity and potentially waste feed). Activities are planned to demonstrate the benefits of monitoring these parameters including the deployment of Chelsea Tryptophan and CDOM sensors at a carp facility at Shuttleworth College, Bedford and trial deployments of cabinet systems further into the project.
A Chelsea Technologies TriLux sensor has already been deployed at a salmon farm site in Scotland and has been collecting data since March 2020. This deployment has
given us some understanding of what conditions we can expect in open cage systems, particularly over the high-risk period from spring-autumn when Harmful Algal Blooms (HABs) can be a nuisance. The data for this is currently being analysed by the University of Bedfordshire and some forecasting models have been developed.
We will also be adopting a novel Nitrate/Nitrite sensor for the project as it has: been proven to work effectively in both fresh and saltwater; and is relatively cheap compared with other optical systems currently available.
Data Transmission Network
The 5G data transmission capability is being developed by the University of Surrey and the initial task was to define the wireless transmission and storage specification based on the mobile communication system at Surrey’s 5G Innovation Centre. A Desk Egg system has been established and uses an existing hardware platform as its’ foundation to expedite its development. This hardware platform provides a considerable number of the microcontroller’s functional pins and is compatible with the Mbed online rapid development Software Development Kit for writing and compiling the microcontroller’s firmware. The microcontroller can be programmed directly via USB connection to a computer. A number of sensor modules and actuation components are integrated into the Desk Egg system to achieve different functions. For example, a user who is using a mobile device can collect temperature and humidity data from the device by installed WiFi Module. This Desk Egg system is an advanced telecommunication system which lays a solid foundation on sensor data and video information transmission for aquaculture system.
Data Analytics Toolbox
The Data Analytics Toolbox is being jointly developed between Perceptive Engineering, University of Surrey and University of Bedfordshire. PEL’s PerceptiveAPC software is being used and adapted for the project. The PerceptiveAPC software was originally developed for process control in the pharmaceutical industry and has since been adapted for other industries similar to recirculating aquaculture, such as wastewater treatment. The predictive models that are being developed between the partners will be incorporated into the software to assist farm operators with management/husbandry decisions.
University of Surrey has developed a preliminary dynamic model for a simplified aquaponics system. This model is verified and tested by comparing the existing research results, which has been transformed by PerceptiveAPC software. Based on the established model, for key but difficult-to-measure water quality parameters in aquaculture (e.g. nitrogen concentration), an adaptive filtering-based soft sensor technology is proposed to monitor water quality. Particle filter is used to eliminate the influence of measurement noise and obtain the estimation of unknown water quality parameter. To improve the performance of adaptive filtering-based soft sensor technology, moving horizon estimation method is applied to jointly estimate system states and parameters based on an optimization approach. In this process, two problems are focused and solved further. One is the result rectification strategy to increase the prediction accuracy of soft measurement results by using offline measurement results. The other one is the determination method of offline measurement frequency to balance offline sampling and laboratory analysis cost and soft sensor accuracy. Finally, the complete framework of adaptive filtering-based soft sensor method for an aquaponic system is established.
The next steps in the project are:
- To deploy Chelsea Tryptophan/CDOM sensors, along with dissolved oxygen, turbidity, ammonia and pH sensors from UoB, at Shuttleworth college for testing
- Flow-through sensor cabinet prototype to be built
- Further testing and validation of models
- Design and development of mobile app
- System integration of flow-through sensor cabinet, 5G data transmission and control software
- Pilot application of integrated system at Shuttleworth college
- To optimize fish growth model using different modelling methods
- To propose fish feeding control and optimization methods
- To test control and optimization methods using experimental data from aquaculture system
That’s all for now. Stay tuned for regular updates on the project or contact us for more information!