Predictive modelling of aquaculture water quality using IoT and advanced machine learning algorithms

100%

Predictive modelling of aquaculture water quality using IoT and advanced machine learning algorithms

One. Introduction

Aquaculture has become one of the world's most rapidly expanding food production sectors, providing a primary source of animal protein for millions of people. In particular, tilapia (Oreochromis niloticus) cultivation has been of significant importance in most nations due to its high growth rate, ability to tolerate extensive variations in environmental conditions, and huge market. Yet, productivity and sustainability in aquaculture systems greatly rely on the water quality in which aquatic organisms are produced. WQPs like temperature, DO, pH, and turbidity play pivotal roles in controlling the health and productivity of aquaculture systems. Therefore, monitoring and controlling these parameters around the clock is essential to enhance fish health, growth, and overall sustainability in aquaculture systems.

Aquaculture system water quality can vary significantly depending on various factors, including season, human activity, and ecological

ABSTRACT

Aquaculture plays a pivotal role in global food security, with tilapia (Oreochromis niloticus) being one of the most widely farmed species due to its resilience and productivity. However, maintaining optimal water quality remains a key challenge, particularly in rural aquaculture systems with limited access to real-time monitoring tools. This study presents a comprehensive six-month monitoring of key water quality parameters in tilapia ponds in Montería, Colombia, using a custom-built Internet of Things (IoT) system. The parameters monitored include pH, turbidity, temperature, and dissolved oxygen (DO)-critical indicators of aquatic health and fish productivity. Advanced machine learning models, including TensorFlow Neural Networks (TFN) and Aqua Enviro Index (AEI), were applied for predictive analysis. Results revealed a statistically significant regression model for temperature (P is less than point zero zero one) and a weak negative correlation between turbidity and temperature (r equals negative zero point zero nine three), highlighting the complex interactions within tropical aquaculture systems. The study offers valuable insights into temporal water quality dynamics and supports data-driven water quality management in resource-constrained areas. Future applications may involve developing mobile dashboards for real-time farmer alerts and decision support, alongside localized training to enhance data literacy. These initiatives can significantly improve aquaculture sustainability, foster technological adoption, and contribute to global food security by empowering rural fish farming communities.

changes. The alteration in water quality can affect the metabolic processes, growth rate, and disease resistance of fish. For instance, the metabolic rate and behavior of various fish species, such as tilapia are influenced by temperature, with tilapia having a preference for temperatures ranging from twenty-five degrees Celsius to thirty degrees Celsius. Exceeding this range can lead to stress, reduced growth rates, and disease susceptibility. Technological advancements have witnessed the installation of real-time monitoring systems that make continuous records of critical water quality parameters. IoT is one technology that has been used to monitor water quality in various aquaculture systems. IoT systems comprising sensors like digital thermometers, oxygen probes, portable pH meters, and turbidimeters enable continuous data acquisition. These systems facilitate on-the-spot water quality measurement, enabling instant intervention and controls to maintain optimal conditions for the growth of fish.

The information noted from IoT-enhanced sensors must undergo rigorous preprocessing for its veracity and quality for further processing. Gaps in data, outliers, as well as irregularity in measurements can corrupt results and lead to misleading findings. Some data preprocessing techniques, such as imputation, normalization, and outlier detection, have to be employed to address these issues. Once cleaned, the data can then be analyzed with statistical methods such as linear regression and correlation analysis to establish relationships between different water quality parameters and assess their effects on the health and productivity of fish. Linear regression is a common statistical tool that relates independent variables (e.g., temperature, pH) and a dependent variable (e.g., dissolved oxygen). Linear regression provides knowledge of the effect that the change in one parameter has on others and can be applied to predict future water quality patterns from previous records. Correlation analysis, such as Pearson's correlation, also supports this observation by quantifying the direction and magnitude of the relationship among several variables. Aside from these traditional methods, machine learning models have also emerged as effective tools for water quality prediction. Use of neural networks, namely TFN, facilitates the use of complex, multivariable data and helps enhance the accuracy of prediction. Machine learning models, such as the TFN model, facilitate the analysis of large, high-dimensional datasets by extracting spatial and temporal patterns in real-time. This approach is better at capturing interdependencies among water quality parameters since it can model non-linear interdependencies that are not detectable by conventional statistical methods. Using advanced machine learning models like TFN is effective in modelling water quality parameters in aquaculture systems. These models also help to assess the long-term sustainability of aquaculture systems by providing insight into how changes in environmental conditions can affect water quality and fish health in the long run.

Real-time monitoring and predictive modelling of water quality parameters not only support aquaculture management practices but also enhance sustainability. By maintaining optimal conditions for fish growth, aquaculture producers can maximize productivity with minimum environmental impact. Furthermore, real-time monitoring can help in early detection of likely issues such as oxygen deficiency, anomalous turbidity, or toxic pH fluctuations, enabling intervention promptly. In response, developing more sophisticated models for water quality estimation and prediction is necessary to advance with aquaculture system sustainability, particularly in tropical nations such as Montería, Colombia, where tilapia aquaculture is a key driver of local economies.

In modern aquaculture, maintaining optimal water quality is crucial for ensuring healthy fish growth and sustainable production. Environmental chemistry plays a vital role in understanding and managing the complex interactions of chemical parameters such as dissolved oxygen, pH, ammonia, nitrites, and heavy metals in aquaculture ponds. Integrating Internet of Things (IoT) technology into pond management offers a powerful solution for real-time monitoring and control of these water quality parameters. Using IoT-enabled sensors and wireless communication systems, farmers can continuously track environmental conditions, receive instant alerts about potential hazards, and automate interventions such as aeration or water exchange. This smart monitoring approach not only improves fish health and productivity but also enhances environmental sustainability by reducing chemical imbalances and minimizing resource waste in aquaculture ecosystems. Consequently, the combination of environmental chemistry and IoT technologies is transforming traditional aquaculture into an intelligent, data-driven practice.

The objective of this research is to track water quality parameters in tilapia ponds in Montería, Colombia, through real-time monitoring supported by IoT sensors. Moreover, the paper addresses the application of machine learning techniques for predicting and modelling water quality parameter correlations to provide a better and improved tool for aquaculture management. Using linear regression, correlation analysis,

and TFN models, this research adds to the rising body of knowledge on aquaculture sustainability improvement through enhanced water quality monitoring and predictive modelling.

Two. Materials and methods

Two point one. Data collection and study area

Two point two. Data preprocessing and cleaning

Two point three. Software and tools

Two point four. Statistical methods

Two point five. Adaptive ensemble integration and three D multivariable prediction modeling

Two point six. Model evaluation and index construction

Three. Results

Three point one. Aqua-Enviro index (AEI)

Three point two. TFN-Based classification model for aquaculture

Three point three. TFN Model architecture design, training, and evaluation

Four. Discussion

Dissolved Oxygen Prediction Model

Five. Conclusion

Ethics and Consent to Participate Declarations

Software Usage

Third-party involvement

CRediT authorship contribution statement

Funding statement

Declaration of competing interest

Overview

The study investigates the monitoring of water quality in tilapia aquaculture using IoT technology and machine learning models. It highlights the importance of parameters like pH, turbidity, and dissolved oxygen in ensuring fish health and productivity, aiming to foster sustainable practices in resource-constrained aquaculture systems.

Key Points

1IoT technology enables real-time monitoring of critical water quality parameters for aquaculture
2Advanced machine learning models improve the accuracy of predictive analysis for water quality
3Maintaining optimal water quality is crucial for the health and productivity of tilapia
4The study offers insights into enhancing aquaculture sustainability through data-driven management practices
5Challenges in data quality and processing must be addressed for effective predictive modelling.

Details

Authors: Md. Abdullah Al Mamun Hridoya, Chiara Bordin, Andleeb Masood, Khalid Masood
Category: Technology and Engineering

PDF
Proficiency of records practitioners using artificial intelligence to enhance records management practices at the Gauteng Department of Education in South Africa
This study examines the proficiency of records practitioners at the Gauteng Department of Education in using artificial intelligence (AI) to improve records management practices. It employs a mixed-methods research approach to evaluate the existing skill levels and proposes a framework for upskilling these practitioners in AI utilization.
PDF
Intelligent Archive Management Based on Deep Learning Technology Driven by Artificial Intelligence
This research article presents a study on an intelligent archive management system (AMS) that employs deep learning (DL) technology to enhance the efficiency and accuracy of processing multimodal data. The proposed model addresses traditional AMS shortcomings such as data classification accuracy and system responsiveness.
PDF
Discharging Records Management Activities Using Artificial Intelligence at the Council for Scientific and Industrial Research, South Africa
This study investigates the application of artificial intelligence (AI) in enhancing records management activities at the Council for Scientific and Industrial Research (CSIR) in South Africa. It provides a framework for utilizing AI to improve efficiency in managing records throughout their life cycle.
PDF
Human-like systematic generalization through a meta-learning neural network
This article explores the capacity of neural networks to achieve human-like systematicity through a meta-learning approach that enhances compositional skills. It presents experimental findings comparing human and machine performance on systematic generalization tasks.
PDF
TBI-603 Network Security & Cyber Law Unit Notes
This document serves as comprehensive unit notes for a course on Network Security and Cyber Law, covering topics like network security practices, types of security attacks, and legal aspects of cybersecurity.