Sundararajan Suma Christal Mary, Shankar Yamini Bhavani, Selvam Sinthia Panneer, Manogaran Nalini, Seerangan Koteeswaran, Natesan Deepa, Selvarajan Shitharth
Department of Information Technology, Panimalar Engineering College, Poonamalle, Chennai, Tamil Nadu, 600123, India.
Department of Networking and Communications, School of Computing, College of Engineering and Technology, SRM Institute of Science and Technology (SRMIST), Kattankulathur, Tamil Nadu, 603 203, India.
Sci Rep. 2025 Jan 14;15(1):1925. doi: 10.1038/s41598-024-84943-7.
The Internet of Things (IoT)-based smart solutions have been developed to predict water quality and they are becoming an increasingly important means of providing efficient solutions through communication technologies. IoT systems are used for enabling connection between various devices based on the ability to gather and collect information. Furthermore, IoT systems are designed to address the environment and the automation industry. The threats associated with aquaponics farming are managed through an IoT-based smart water monitoring framework, which has become increasingly relevant in recent days. Therefore, this approach is crucial for achieving a remarkable improvement in order to increase the productivity rate and yield. The quality of water directly affects the rate of growth, efficiency of feed, and the overall health rate of the fish, plants, and bacteria. Insufficient knowledge about species selection poses a significant challenge in aquaponics farming, as it heavily relies on the water quality parameters. To address the challenges of conventional models, we have developed an effective IoT-based water quality prediction model, more specifically designed for aquaponic fish ponds. The data needed to perform the developed water quality prediction model will be acquired from "a simple dataset of aquaponic fish pond IoT" database. After that, these data are forwarded to the feature extraction phase. The weighted features, DBN (Deep Belief Network) features, and the original features are achieved in the feature extraction stage. The weighted features are obtained using the Revamped Fitness-based Mother Optimization Algorithm (RF-MOA). Subsequently, these extracted features are fed into the Multi-Scale feature fusion-based Convolutional Autoencoder with a Gated Recurrent Unit (MS-CAGRU) network for predicting the water quality. Thus, the water quality predicted data is obtained. The proposed model integrates GRU networks with a convolutional autoencoder to improve water quality prediction by capturing trends and managing temporal dependencies. It enhances accuracy by analysing key parameters and employing techniques to reduce overfitting. The effectiveness of the proposed system is evaluated in comparison to the traditional models using some evaluation measures.
基于物联网（IoT）的智能解决方案已被开发用于预测水质，并且它们正日益成为通过通信技术提供高效解决方案的重要手段。物联网系统基于收集和采集信息的能力，用于实现各种设备之间的连接。此外，物联网系统旨在服务于环境和自动化行业。与鱼菜共生养殖相关的威胁通过基于物联网的智能水监测框架进行管理，该框架在最近变得越来越重要。因此，这种方法对于实现显著改进以提高生产率和产量至关重要。水的质量直接影响鱼、植物和细菌的生长速度、饲料效率以及整体健康状况。在鱼菜共生养殖中，对物种选择的了解不足构成了重大挑战，因为它严重依赖水质参数。为了应对传统模型的挑战，我们开发了一种有效的基于物联网的水质预测模型，该模型是专门为鱼菜共生鱼塘设计的。执行所开发的水质预测模型所需的数据将从“鱼菜共生鱼塘物联网简单数据集”数据库中获取。之后，这些数据被转发到特征提取阶段。在特征提取阶段获得加权特征、深度信念网络（DBN）特征和原始特征。加权特征是使用改进的基于适应度的母优化算法（RF-MOA）获得的。随后，这些提取的特征被输入到带有门控循环单元（MS-CAGRU）网络的基于多尺度特征融合的卷积自动编码器中，以预测水质。这样就获得了水质预测数据。所提出的模型将门控循环单元网络与卷积自动编码器集成，通过捕捉趋势和管理时间依赖性来改进水质预测。它通过分析关键参数并采用减少过拟合的技术来提高准确性。使用一些评估指标，将所提出系统的有效性与传统模型进行比较评估。
