top of page

Navigating Clean Waters: Predictive Nitrogen Monitoring for Sustainable Water Quality in Malaysia

Most of our daily household water needs are supplied by water treatment plants, but constant pollution at vital water sources have resulted in many abrupt closures and thus water shortages. As much of this pollution is comprised of nitrogen compounds, predicting the rise in nitrogen concentration levels will serve to enhance our water monitoring. The ACO-ENN hybrid model has been developed as a new AI-based prediction model capable of autonomously gathering daily data to create high-accuracy data models. This hybrid model has the potential to become a highly-efficient and cost- and time-effective prediction method used not just in Malaysia but around the world as well.



Water treatment plants supply a major portion of our daily household water needs in Malaysia. Unfortunately, however, these plants have been facing a serious threat in the form of rising pollution levels (i.e. nitrogenous compounds) in streams and rivers, leading to the closure of these plants (see Figure 1). These abrupt closures in turn affect the water supply to consumers, resulting in water shortages, and thus adding additional pressure on the government for arranging an alternate source of water supply. In order combat this problem, we need to develop better methods of predicting nitrogen content.



Figure 1: News articles covering water treatment plant closures

The prediction of nitrogen will not only assist in monitoring the nitrogen concentration in streams but will also help in optimizing the usage of fertilizers in agricultural fields, as agricultural fields are considered as primary source of nitrogenous compounds. A precise prediction model guarantees the delivery of better-quality water for human use, especially since the operation of various water treatment plants in Malaysia depend on the concentration of nitrogen in streams. This project proposes the Artificial Intelligence integrated hybrid model-based Ant colony optimization (ACO) and Elman neural network (ENN) as a new means of predicting nitrogen concentrations. This model is based on two decades’ worth of data concerning nitrate-nitrogen and ammonia-nitrogen levels at two different measuring stations (LUI and KAJANG) in the Langat River basin in Malaysia. The development of the model has been carried out for almost 30 months to achieve the most appropriate structure.


The development of the new ACO-ENN hybrid model will provide enhanced prediction accuracy with regression values of all the models greater than 0.96. Data plots show that the predicted nitrate-nitrogen and ammonia-nitrogen almost overlap with the actual values, thus representing the prediction accuracy of the model (See Figure 2).



Figure 2: Charts showing prediction of nitrate-nitrogen and ammonia-nitrogen at Kajang Station

The main outcome from this project is that the development of a user-friendly Graphical User Interface (GUI) software package based on the ACO-ENN hybrid model could provide accurate nitrogen concentration readings for the two monitoring stations in the Langat River basin (see Figure 3). These models can autonomously provide the daily data of nitrogen pollutants thus saving the daily effort of quantifying such data in the laboratory. Furthermore, these models can help in creating an alert for nitrogen surges in rivers before it actually happens, thus granting the government ample time to optimize various nitrogen inputs in the rivers. This model and software has the potential to be expanded to cover several locations on the river streams in Malaysia as well as becoming a valuable alternative cost- and time-effective method for predicting nitrogen concentration at different river monitoring stations worldwide.



Figure 3: Nitrogen prediction software
 

Authors and researchers featured:

Associate Prof. Dr. Lai Sai Hin
Department of Civil Engineering, Faculty of Engineering

Prof. Dr. Ahmed Hussein Kamel Ahmed Elshafie
Department of Civil Engineering, Faculty of Engineering

Pavitra Kumar



Copyedit: Michael Hoe Guang Jian (michaelhoe.hoe@gmail.com)
46 views0 comments

Comments


bottom of page