Intelligent Prediction and Continuous Monitoring of Water Quality in Aquaculture: Integration of Machine Learning and Internet of Things for Sustainable Management

Water, Journal Year: 2025, Volume and Issue: 17(1), P. 82 - 82

Published: Jan. 1, 2025

Aquaculture is a vital contributor to global food security, yet maintaining optimal water quality remains persistent challenge, particularly in resource-limited rural settings. This study integrates Internet of Things (IoT) technology, Machine Learning (ML) models, and the Quantum Approximate Optimization Algorithm (QAOA) enhance monitoring prediction aquaculture. IoT sensors continuously measured parameters such as temperature, dissolved oxygen (DO), pH, turbidity, while ML models—including Random Forest—provided high accuracy predictions (R2 = 0.999, RMSE 0.0998 mg/L). The integration QAOA reduced model training time by 50%, enabling rapid, real-time responses changing conditions. Over 6000 corrective interventions were conducted during study, fish survival rates above 90% tropical aquaculture environments. adaptable system designed for both urban settings, using low-cost local data processing constrained environments or cloud-based systems analysis. results demonstrate potential IoT–ML–QAOA mitigate environmental risks, optimize health, support sustainable practices. By addressing technological infrastructural constraints, this advances management contributes security.

Language: Английский

---

Enhanced water quality prediction model using advanced hybridized resampling alternating tree-based and deep learning algorithms

Environmental Science and Pollution Research, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 24, 2025

Language: Английский

---

A novel machine learning-based framework for the water quality parameters prediction using hybrid long short-term memory and locally weighted scatterplot smoothing methods

Journal of Hydroinformatics, Journal Year: 2024, Volume and Issue: 26(5), P. 1059 - 1079

Published: April 12, 2024

ABSTRACT Water quality prediction is crucial for effective river stream management. Dissolved oxygen, conductivity and chemical oxygen demand are vital parameters water quality. Development of machine learning (ML) deep (DL) methods made them widely used in this domain. Sophisticated DL techniques, especially long short-term memory (LSTM) networks, required accurate, real-time multistep prediction. LSTM networks predicting due to their ability handle long-term dependencies sequential data. We propose a novel hybrid approach combining with data smoothing method. The Sava at the Jamena hydrological station serves as case study. Our workflow uses alongside LOcally WEighted Scatterplot Smoothing (LOWESS) technique filtering. For comparison, Support Vector Regressor (SVR) baseline Performance evaluated using Root Mean Squared Error (RMSE) Coefficient Determination R2 metrics. Results demonstrate that outperforms method, an up 0.9998 RMSE 0.0230 on test set dissolved oxygen. Over 5-day period, our achieves 0.9912 0.1610 confirming it reliable method

Language: Английский

---

Integrating machine learning models for optimizing ecosystem health assessments through prediction of nitrate–N concentrations in the lower stretch of Ganga River, India

Environmental Science and Pollution Research, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 30, 2025

Language: Английский

---

Toward Design of Internet of Things and Machine Learning-Enabled Frameworks for Analysis and Prediction of Water Quality

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 101055 - 101086

Published: Jan. 1, 2023

The degradation of water quality has become a critical concern worldwide, necessitating innovative approaches for monitoring and predicting quality. This paper proposes an integrated framework that combines the Internet Things (IoT) machine learning paradigms comprehensive analysis prediction. IoT-enabled comprises four modules: sensing, coordinator, data processing, decision. IoT is equipped with temperature, pH, turbidity, Total Dissolved Solids (TDS) sensors to collect from Rohri Canal, SBA, Pakistan. acquired preprocessed then analyzed using models predict Water Quality Index (WQI) Class (WQC). With this aim, we designed learning-enabled Preprocessing steps such as cleaning, normalization Z-score technique, correlation, splitting are performed before applying models. Regression models: LSTM (Long Short-Term Memory), SVR (Support Vector Regression), MLP (Multilayer Perceptron) NARNet (Nonlinear Autoregressive Network) employed WQI, classification SVM Machine), XGBoost (eXtreme Gradient Boosting), Decision Trees, Random Forest WQC. Before that, Dataset used evaluating split into two subsets: 1 2. 600 values each parameter, while 2 includes complete set 6000 parameter. division enables comparison evaluation models' performance. results indicate regression model strong predictive performance lowest Mean Absolute Error (MAE), Squared (MSE), Root (RMSE) values, along highest R-squared (0.93), indicating accurate precise predictions. In contrast, demonstrates weaker performance, evidenced by higher errors lower (0.73). Among algorithms, achieves metrics: accuracy (0.91), precision recall (0.92), F1-score (0.91). It also conceived perform better when applied datasets smaller numbers compared larger values. Moreover, comparisons existing studies reveal study's improved consistently For classification, outperforms others, exceptional accuracy, precision, recall, metrics.

Language: Английский

---

Water Quality Monitoring and Assessment for Efficient Water Resource Management through Internet of Things and Machine Learning Approaches for Agricultural Irrigation

Water Resources Management, Journal Year: 2024, Volume and Issue: unknown

Published: June 3, 2024

Language: Английский

---

Comparison of Water Quality Prediction for Red Tilapia Aquaculture in an Outdoor Recirculation System Using Deep Learning and a Hybrid Model

Water, Journal Year: 2024, Volume and Issue: 16(6), P. 907 - 907

Published: March 21, 2024

In modern aquaculture, the focus is on optimizing production and minimizing environmental impact through use of recirculating water systems, particularly in outdoor setups. such maintaining quality crucial for sustaining a healthy environment aquatic life, challenges arise from instrumentation limitations delays laboratory measurements that can animal production. This study aimed to predict key parameters an recirculation aquaculture system (RAS) red tilapia including dissolved oxygen (DO), pH, total ammonia nitrogen (TAN), nitrite (NO2–N), alkalinity (ALK). Initially, random forest (RF) model was employed identify significant factors predicting each parameter, selecting top three features routinely measured farm: DO, temperature (Temp), TAN, NO2–N, transparency (Trans). approach streamline analysis by reducing variables computation time. The selected were then used prediction, comparing performance convolutional neural network (CNN), long short-term memory (LSTM), CNN–LSTM models across different epochs (1000, 3000, 5000). results indicated at 5000 effective ALK, with high R2 values (0.815, 0.826, 0.831, 0.780, respectively). However, pH prediction showed lower efficiency value 0.377.

Language: Английский

---

Predicting water quality in municipal water management systems using a hybrid deep learning model

Engineering Applications of Artificial Intelligence, Journal Year: 2024, Volume and Issue: 133, P. 108420 - 108420

Published: April 23, 2024

Language: Английский

---

Deep learning for water quality multivariate assessment in inland water across China

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2024, Volume and Issue: 133, P. 104078 - 104078

Published: Aug. 16, 2024

• Demonstration of limited but representative training dataset for efficient modeling. Robust DNN models independent and simultaneous retrieval Chl-a, TSS SDD. Better performance over XGBoost, RF, SVM. Applicability on heterogeneous lakes. Challenges significant water quality degradation trends in Chinese Remote sensing optically complex inland waterbodies is challenging due to the nonlinear correlation between parameters optical properties. However, integration deep learning techniques datasets offers potential address these challenges effectively. This study aims develop robust models, utilizing highly in-situ radiometrically corrected hyperspectral remote reflectance (R rs ) measurements collected from diverse lakes China, Chlorophyll-a (Chl-a), Secchi Disk Depth (SDD), Total Suspended Solids (TSS) using Sentinel-2 analysis ready products. The GLObal Reflectance community Imaging Aquatic environments (GLORIA) provides 400 such lakes, which are simulated R with its spectral response function build a dataset. Using this dataset, Multilayer Perceptron (MLP) based Deep Neural Network (DNN) developed compared eXtreme Gradient Boosting (XGB), Random Forest (RF), Support Vector Machine (SVM) algorithms. outperformed effective evaluation Chl-a (Root Mean Squared Error (RMSE) = 14.18 mg/m 3 ), (RMSE=7.23 g/m SDD (RMSE=0.12 m) test (RMSE=14.42 (RMSE=0.07 against Sentinel-2A validation Liangzi lake. Mixed Density (MDN) model showed less accuracy (RMSE=16.76 same Impact different atmospheric correction processors also assessed achieved their Atmospheric Correction (Sen2Cor) processor. Finally, maps various China produced showing realistic ranges. These results show trained practical applications spatial temporal quality.

Language: Английский

---

A Water Quality Prediction Model Based on Modal Decomposition and Hybrid Deep Learning Models

Water, Journal Year: 2025, Volume and Issue: 17(2), P. 184 - 184

Published: Jan. 10, 2025

When the total nitrogen content in water sources exceeds standard, it can promote rapid proliferation of algae and other plankton, leading to eutrophication body also causing damage ecological environment source area. Therefore, making timely accurate predictions quality at is vital importance. Since data exhibit non-stationary characteristics, predicting them quite challenging. This study proposes a novel hybrid deep learning model based on modal decomposition, ERSCB (EMD-RBMO-SVMD-CNN-BiGRU), enhance accuracy forecasting. The first employs Empirical Mode Decomposition (EMD) technology decompose original data. Subsequently, quantifies complexity subsequences obtained from EMD using Sample Entropy (SE) further decomposes most complex Sequential Variational (SVMD). To address matter selecting balanced parameters SVMD, this introduces Red-Billed Blue Magpie Optimization (RBMO) algorithm optimize hyperparameters SVMD. On basis, forecasting constructed by integrating Convolutional Neural Networks (CNN) Bidirectional Gated Recurrent Unit (BiGRU) networks. experimental results show that, compared existing prediction models, has an improved 4.0% 3.1% for KaShi River GongNaiSi areas, respectively.

Language: Английский

---