Abstract
This project is about using the Levenberg Marquardt algorithm in Artificial Neural Network (ANN) to classified between clean and polluted water. By designing the ANN model system and then trained it, the ANN able to find the best-optimized model that can distinguish between clean and polluted water. 200 samples of clean and polluted water were collected in the process. The sample will then be divided into two cases, which are 100 clean water samples, and another 100 from polluted water samples. All samples were collected from Universiti Teknologi MARA (UiTM) Cawangan Pulau Pinang, and Sungai Derhaka, Pulau Pinang. The capacitance value calculation for both cases was carried out using LCR meter with a frequency range of 1kHz. To measure the normal distribution of the data, IBM SPSS Statistical Software were implemented. For both cases, the statistical analysis data show that the p-value is more than 0.05, which indicates that the data are normally distributed. These measurement inputs were then going through the process of classification in ANN to generate the optimized models by using LM algorithm. The model is being trained, tested, and validated to differentiate between clean water and polluted water. There were 1 optimized model selected from the classification process. The accuracy from the selected most optimized models were 100%. The selected most optimized models were then can be used to classify between clean water and polluted water based on capacitance input.
Metadata
Download
39866.pdf
Download (190kB)