Application of Long-Short Term Memory for Accurate Biochemical Oxygen Demand Prediction in Rivers through Water Quality Parameters

Abstract

Evaluating water quality is crucial for preserving the quality of river water. However, the typical technique of getting biochemical oxygen demand (BOD) values via laboratory testing might take several days, delaying the application of real-time measurement to improve water quality. This paper suggests using machine learning to predict BOD values from eight water quality measurements. The BOD rate in the Klang River, Selangor, Malaysia, was estimated using the long short-term memory (LSTM) method. The model was trained using historical data collected from eleven water collection points along the river. The predictive test results indicated that the LSTM model with 8 water parameters as input gave the most accurate predictions compared to the models with 5 and 3 water parameters. The results of this study indicate that machine learning methods can be used to predict BOD levels in real-time. It enables water quality managers to enhance water quality and safeguard human health proactively.

Keywords

• bod prediction
• deep neural network
• klang river
• lstm
• prediction