Development of wwm system using mlnn-ga for ph prediction of water quality in catchment area

Muhammad Aznil, Ab Aziz (2021) Development of wwm system using mlnn-ga for ph prediction of water quality in catchment area. Masters thesis, Universiti Malaysia Pahang (Contributors, UNSPECIFIED: UNSPECIFIED).

Development of wwm system using mlnn-ga.pdf - Accepted Version

Download (616kB) | Preview


Water quality is crucial to maintain as water is a necessity in our daily life. In aquatic ecosystem, good water quality allows aquatic life to have good health and improving their productivity leads to significant benefits to the environment. However, the method of collecting data is still in manual way. Data samples need to be collected and evaluated in the laboratory. It leads the results of water quality took a long time to retrieve and cannot be obtained continuously. Hence, a system is required to monitor the water quality that capable to obtain the results fast and continuously. In this research studies, Wireless Passive Water Quality Catchment Monitoring System or WMM System is introduced to collect the water quality parameters. Five parameters are measured in the system, which are potential hydrogen (pH), temperature, light intensity of water, coordinate of collecting data, and wave velocity. Primary lake at University Malaysia Pahang is selected as the experimental area for the WWM System. However, coverage area of the lake is large and require many WWM Systems to be developed. This method is expensive in terms of budget expenditure. Thus, prediction of water quality is applied to solve this problem with a number of WWM Systems used for collecting data will be reduced. Artificial Neural Network or ANN is one of the methods that commonly used to predict the quality of water. The structure and function of ANN is based on biological neural network. ANN process consists of three layer which are input layer, hidden layer, and output layer and weight of the network usually adjusted by using back-propagation (BP) technique. This technique has several advantages which are fast, simple, and easy to program. However, there are some weaknesses identified in BP such as being often trapped in local minima, sensitive to noisy data, and the performance of BP on certain problems is dependent on the input data. In this research study, Multi-Layer Neural Network optimized by Genetic Algorithm (MLNN-GA) is introduced to predict the water quality. This model will predict the pH value obtained by WWM System based on the value of pH surrounding within 100m2 area. MLNN-GA contain three hidden layers and Genetic Algorithm (GA) will optimize the weight of neural network during the training process. GA has several advantages, which are less often trapped in local minima and can operate in a complex process such as extended neural network. In training process of MLNN-GA, number of neurons in a hidden layer will be tested to obtain the best accuracy. The parameter of pH is chosen for predicting water quality as pH is one important parameter that can influence the condition of water. MLNN-GA is a new method proposed in this research study for predicting the water quality in terms of pH value. The results of prediction will be compared with BP technique and the average value of pH within the 100m2 area. The results of prediction obtained by MLNN-GA model with 99.64% accuracy has a significant potential to be used in the future and WWM System has a great potential to be used in monitoring water quality.

Item Type: Thesis (Masters)
Additional Information: Thesis (Master of Science) -- Universiti Malaysia Pahang – 2021, SV: DR. MOHAMMAD FADHIL BIN ABAS, NO. CD: 12934
Uncontrolled Keywords: Potential Hydrogen (pH); Genetic Algorithm (GA)
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering
Faculty/Division: Faculty of Electrical and Electronic Engineering Technology
Depositing User: Mrs. Neng Sury Sulaiman
Date Deposited: 13 Jun 2022 04:24
Last Modified: 13 Jun 2022 04:24
Download Statistic: View Download Statistics

Actions (login required)

View Item View Item