Integration Of Logistic Regression And Multi-Layer Perceptron For Single And Dual Axis Solar Tracking Systems
Loading...
Date
2018-06-01
Authors
A. Al-Rousan, Nadia
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Sains Malaysia
Abstract
Intelligent solar tracking systems to track the trajectory of the sun across the sky has
been actively studied and proposed nowadays. However, different solar tracking
variables have been employed to build those intelligent solar tracking systems without
considering the dominant and optimum ones. In addition, several low performance
intelligent solar tracking systems have been designed and implemented due to the
inappropriate combination of solar tracking variables and intelligent classifiers to drive
the solar trackers. Thus, this research aims to (i) investigate and evaluate the most
effective and dominant variables on solar tracking systems, (ii) investigate the
appropriate combination of solar variables and intelligent classifier for solar tracking
systems, (iii) propose new solar tracking systems by integrating supervised and
unsupervised intelligent classifiers. The results revealed that month, day, and time are
the most effective variables for single and dual axis solar tracking systems. By using
these variables, this study has successfully integrated between multi-layer perceptron
(MLP) or cascade multi-layer perceptron (CMLP) and trained logistic regression (LR)
models. The proposed MLP-LR system is able to increase the prediction rate of MLP
network to 99.13% for single axis tracking systems (i.e. which is 2.35% of
improvement). The system is also able to decrease the mean square error (MSE) rate
to 0.010 × 10−2 as compared to value of MSE for the conventional MLP. In addition,
the proposed CMLP-LR system is able to increase the prediction rate of CMLP
network to 99.19% for dual axis tracking system (i.e. 1.23% of improvement), while
the MSE rate is decreased to 6.250 × 10−5 as compared to value of MSE for the
conventional CMLP. The new developed models achieved less number of overall
connections (i.e. which are 77.58% and 86.16% of improvement for MLP and CMLP
respectively), less number of neurons (i.e. 63.51% of improvement for both MLP and
CMLP), and less time complexity (i.e. which are 70.40% and 99% of improvement for
MLP and CMLP respectively). The finding suggests that the proposed intelligent solar
tracking systems has a great potential to be applied for real-world applications (i.e.
solar heating systems, solar lightening systems, factories, and solar powered
ventilation).