Publication: Aquaculture parameters based on iot smart platform analysed using intelligent networks
Loading...
Date
2021-07-01
Authors
Sivanesan, Subramanian
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Aquaculture industry is growing rapidly and it has become a source of income. One of the premium products of aquaculture is the growth of shrimp. Like any other industry, aquaculture also needs to catch up with the current technology, so as to get rich yields. In order for the shrimps to grow and thrive, the water quality should be in a good
state. An aqua farmer’s daily work involves keeping track of water parameters that influence the growth of shrimps. Monitoring of water quality parameters in shrimp farms
is nowadays mainly carried out through manual processes which are tedious. For example, some of them use handheld meters for each parameter, others give their pond
water sample to a laboratory for testing. Upon researching the traditional ways of tracking shrimp pond parameters, it was found to be time-consuming, extra labour and less
efficient. This research proposes a smart monitoring system for the shrimp pond. Under the proposed system, the parameters of the shrimp ponds will be monitored by utilizing
smartphones. The monitoring system is based on an artificial neural network. The shrimp pond uses sensors that are placed in the pond to collect the reading of each parameter.
The data is then stored with the capabilities of IoT. The stored data can then be retrieved from the cloud and fed into a prediction model. A mobile application is then created using
MATLAB. An alert sound to notify the aqua farmers is created if the water parameters exceed the threshold values. A number of techniques were compared to achieve a good
classification and prediction model. The Generalized Regression Neural Network (GRNN) technique was chosen as the classification model since it provides 100% accuracy with no errors. To achieve a prediction model, two artificial neural networks were selected and compared: Multilayer Perceptron and Long Short-Term Memory (LSTM). From the comparison results, the water quality parameters were quite accurately predicted through the multilayer perceptron model. In conclusion, the overall developed system can be implemented in industries or farms related to aquaculture.