Artificial Neural Network and Near Infrared Light in Water pH and Total Ammonia Nitrogen Prediction
Water quality plays an important role in aquaculture. The operation of a freshwater aquaculture fish farming is highly dependent on the ability to understand, monitor, and control the physical and chemical constituents of the water. pH and total ammonia nitrogen (TAN) levels are two critical water q...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Published: |
Penerbit UTHM
2022
|
| Online Access: | https://www.scopus.com/inward/record.uri?eid=2-s2.0-85132507118&doi=10.30880%2fijie.2022.14.04.017&partnerID=40&md5=57d308a3b81d569588f17f5f15687a65 http://eprints.utp.edu.my/33168/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Water quality plays an important role in aquaculture. The operation of a freshwater aquaculture fish farming is highly dependent on the ability to understand, monitor, and control the physical and chemical constituents of the water. pH and total ammonia nitrogen (TAN) levels are two critical water quality parameters that affect fish growth rate and health. However, pH and TAN levels are affected by uncontrollable factors e.g. weather, temperature, and biological processes occurring in the water. Therefore, it is important to monitor changes in pH and TAN levels frequently to maintain optimal conditions for freshwater habitats. Near infrared spectroscopy (NIR) has been extensively investigated as an alternative measurement approach for rapid quality control without sample preparation. Therefore, this research aims to evaluate the feasibility of machine learning combined with NIR light in predicting the water pH and TAN values of a fish farming system. The proposed system contains three main components i.e. a multi-wavelength light emitting diode (LED), a light sensing element, and a machine learning model i.e. artificial neural network (ANN). First, the transmitted NIR light with different wavelengths of water samples was measured using the proposed system. Then, the actual pH and TAN values of the water samples were quantified using conventional methods. Next, ANN was used to correlate the measured NIR transmittance with the pH and TAN values. The results show that ANN with four hidden neurons achieved the best prediction performance with a mean square error (MSE) of 0.1466 and 0.3136 and a correlation coefficient (R) of 0.8398 and 0.9560 for the pH and TAN predictions, respectively. These results show that ANN coupled with NIR light can be promisingly developed for in situ water quality prediction without sample preparation. © Universiti Tun Hussein Onn Malaysia Publisher�s Office |
|---|