<![CDATA[Pond water quality is a pivotal factor that influences the productivity and health of biota in aquaculture systems. The monitoring and prediction of water quality parameters, including temperature, pH, and dissolved oxygen (DO) levels, are imperative for maintaining optimal environmental conditions. The objective of this research is to develop the Euclidean Distance algorithm as an optimization method in adaptive neuro-fuzzy inference system (ANFIS) modeling to enhance the accuracy of internet of things (IoT)-based pond water quality prediction. Water quality parameter data is collected in real-time using IoT sensors connected to an ESP32 microcontroller and transmitted to a cloud storage platform for analysis. Subsequently, the data undergoes a series of processing steps, including min-max normalization and feature selection based on Euclidean distance. This process aims to generate a more representative and relevant subset of data for the subsequent model training process. The ANFIS model was trained using the optimized data and evaluated using MSE, MAD, MRSE and MAPE metrics. The training process involving four data sharing scenarios demonstrated a reduction in error when compared to the model that lacked optimization, specifically: The following proportions were determined: 50% versus 50% (0.11824 versus 0.15536), 70% versus 30% (0.18666 versus 0.19454), 80% versus 20% (0.17843 versus 0.18833), and 90% versus 10% (0.22477 versus 0.22859). The findings indicate that the incorporation of the Weighted Euclidean Distance algorithm within the IoT-based prediction system can markedly enhance the efficiency and precision of the ANFIS model.]]>
Copyrights © 2025
