Hybrid multilayer perceptron-firefly optimizer algorithm for modelling photosynthetic active solar radiation for biofuel energy exploration

Goundar, Harshna and Yaseen, Zaher Mundher and Deo, Ravinesh ORCID: https://orcid.org/0000-0002-2290-6749 (2021) Hybrid multilayer perceptron-firefly optimizer algorithm for modelling photosynthetic active solar radiation for biofuel energy exploration. In: Predictive modelling for energy management and power systems engineering. Elsevier, Amsterdam, Netherlands, pp. 191-232. ISBN 978-0-12-817772-3


Abstract

This study established the preciseness of the robust MLP-firefly optimizer (MLP-FFA) artificial intelligence algorithm coupled with satellite-derived photosynthetic active radiation (PAR) data in order to forecast PAR itself using historical values for a regional Queensland location (Toowoomba). To optimize the MLP-FFA model, (1) hidden and output transfer functions; (2) number of hidden neurons; (3) training, cross-validation, and test percentage splits; and (4) number of historical lags were trialled, such that the logarithmic sigmoid hidden function and tangent sigmoid output function, with 120 hidden neurons, nine lags as inputs and a 60% training, 20% cross-validation, and 20% testing set, was finally adopted for thre forecasting of PAR. To meet the objective (Section 7.1), the MLP-FFA objective model was benchmarked with the MLP, RF, and MLR models for PAR forecasting. The findings of this study are that the performance of the MLP-FFA model, according to forecasting error directly, as well as association and error performance metrics, outperformed the MLP and RF models. MLR generally outperformed MLP-FFA, but this is most likely a result of nonstochastic monthly PAR data, and it is recommended for future work to investigate lower temporal resolution data (which in turn is more stochastic), which a more robust model like MLP-FFA is known to handle, but a poor model like MLR cannot. A lower resolution such as daily or hourly was not found at the time of this study, and manual data collection was not appropriate due to the time and cost limits of this research. Albeit, the MLP-FFA model has still been very effective in modeling PAR with very high accuracy and low error, leading to a significant contribution to research which is confirming something unknown—that the MLP-FFA model is in fact very effective at satellite-based PAR modeling with historical data as inputs for learning for a regional Queensland location. The results of this study are a significant research contribution, which can be used to forecast PAR conditions, a vital requirement for the growth of algal biofuel; these algae can be farmed in the ideal location of the sunny, subtropical Toowoomba region.


Statistics for USQ ePrint 39835
Statistics for this ePrint Item
Item Type: Book Chapter (Commonwealth Reporting Category B)
Refereed: Yes
Item Status: Live Archive
Additional Information: Permanent restricted access to Published version + Front Matter in accordance with the copyright policy of the publisher.
Faculty/School / Institute/Centre: Current - Faculty of Health, Engineering and Sciences - School of Sciences (6 Sep 2019 -)
Faculty/School / Institute/Centre: Current - Faculty of Health, Engineering and Sciences - School of Sciences (6 Sep 2019 -)
Date Deposited: 06 Oct 2020 05:52
Last Modified: 06 Oct 2020 05:55
Uncontrolled Keywords: solar energy; forecasting
Fields of Research (2008): 05 Environmental Sciences > 0599 Other Environmental Sciences > 059999 Environmental Sciences not elsewhere classified
08 Information and Computing Sciences > 0801 Artificial Intelligence and Image Processing > 080108 Neural, Evolutionary and Fuzzy Computation
Socio-Economic Objectives (2008): E Expanding Knowledge > 97 Expanding Knowledge > 970108 Expanding Knowledge in the Information and Computing Sciences
URI: http://eprints.usq.edu.au/id/eprint/39835

Actions (login required)

View Item Archive Repository Staff Only