Water quality is a determining factor for the success of aquaculture operations. Traditional monitoring methods, based on manual and sporadic analyses, are inefficient in ensuring optimal conditions for aquatic life.
The need for constant monitoring and the ability to make quick decisions in response to changes in water quality in Recirculating Aquaculture Systems (RAS) have become pressing challenges for the sector. However, the growing adoption of advanced technologies is transforming this landscape.
In this regard, researchers from the National Institute of Technology Raipur (India) propose an improved Dilated Spatio-Temporal Convolutional Neural Network (DSTCNN) for water quality monitoring in aquaculture, using an Internet of Things (IoT) system configuration to capture real-time data from aquaculture tanks. Water quality data captured through IoT sensors are labeled according to Water Quality Index (WQI) standards for analysis.
Leveraging IoT and Water Quality Index for Better Management
IoT-enabled systems offer a comprehensive approach to water quality monitoring by collecting real-time data from multiple aquaculture ponds. However, this wealth of information requires sophisticated analysis to extract meaningful insights. The Water Quality Index (WQI) provides a valuable framework for assessing the overall quality of water based on various parameters, facilitating informed decision-making.
By combining IoT data with WQI calculations, aquaculture operators can gain a clearer picture of their pond conditions and identify potential issues early. However, the WQI alone cannot predict future water quality trends.
Deep Learning: Unleashing Predictive Power
Deep learning, a subset of artificial intelligence, has demonstrated immense potential in analyzing complex patterns within large datasets. While traditional deep learning models have been applied to water quality studies, they often fail to capture the intricate spatial and temporal relationships inherent in aquatic environments.
To overcome these limitations, researchers have developed the Dilated Spatio-Temporal Convolutional Neural Network (DSTCNN). This innovative model excels in analyzing water quality data by simultaneously considering spatial and temporal factors. By incorporating a hybrid activation function that combines ReLU and softmax, the DSTCNN improves accuracy and reliability.
An Innovative Solution: DSTCNN
Researchers have developed a new technique called the Dilated Spatio-Temporal Convolutional Neural Network (DSTCNN) to overcome these limitations. This advanced solution uses an IoT system to collect real-time data from aquaculture ponds. The collected data are labeled according to Water Quality Index (WQI) standards and classified by the DSTCNN model into two categories: suitable for fish growth or potentially lethal.
The key to DSTCNN’s success lies in its ability to handle the complexity of spatiotemporal data. Thanks to dilated convolutions, the model can capture essential patterns and features at multiple time points, enabling it to understand the intricate relationships between different factors affecting water quality.
To avoid overfitting and improve generalization, researchers have incorporated a hybrid activation function that combines the advantages of ReLU and sigmoid.
Impressive Results
The results obtained with DSTCNN are extraordinary. The model achieved an accuracy of 99.28% on real-time data and 99.02% on public data, significantly outperforming the benchmark PCR-GB model.
Enhancing Aquaculture with DSTCNN
The proposed DSTCNN model offers several advantages over existing approaches:
- Comprehensive Analysis: Captures the complex interaction of spatial and temporal factors influencing water quality.
- Higher Accuracy: Provides improved prediction accuracy through the use of a hybrid activation function.
- Early Warning System: Enables proactive measures to prevent water quality issues and protect aquatic organisms.
The Future of Aquaculture in RAS
The implementation of DSTCNN represents a significant advancement in aquaculture, allowing for more efficient and sustainable management of operations. By preventing water quality issues, fish mortality can be reduced, resource use optimized, and the profitability of aquaculture farms improved.
This technology has the potential to revolutionize the industry, providing aquaculturists with a powerful tool to make data-driven decisions and ensure the health of their crops.
Conclusion
In conclusion, DSTCNN represents a significant breakthrough in water quality monitoring for aquaculture. Its ability to analyze complex data and provide accurate predictions makes it an invaluable tool for producers and an important step towards more efficient and environmentally friendly aquaculture.
Contact
K. Jairam Naik
Department of Computer Science & Engineering, National Institute of Technology Raipur
Raipur, India
Email: Jnaik.cse@nitrr.ac.in
Reference (open access)
Arepalli, P. G., & Jairam Naik, K. (2024). An IoT based smart water quality assessment framework for aqua-ponds management using Dilated Spatial-temporal Convolution Neural Network (DSTCNN). Aquacultural Engineering, 104, 102373. https://doi.org/10.1016/j.aquaeng.2023.102373
