Disaster management system based on Levenberg-Marquardt algorithm artificial neural network / W Ahmad Syafiq Hilmi Wan Abdull Hamid ...[et al.]

This paper presents Disaster Management System
Based on Levenberg-Marquardt Algorithm Artificial Neural
Network. Although Malaysia is located outside the “Pacific Rim
of Fire” and protected from severe ravages caused by natural
disasters, however, Malaysia do still experience other disasters.
In Malaysia, the disaster management is laid out under
integrated system called the Malaysia National Security Council
Directive No. 20 (MNSC No. 20). Unfortunately, the policy
introduced in the year 1997 is not enough to help the responders
managing disasters efficiently. Study shows, a computerized
system was identified as one of the best tools in supporting the
responders in Malaysia especially the lead responding agency to
manage disasters. Thus, the Disaster Management System Based
on Levenberg-Marquardt Algorithm Artificial Neural Network
was developed with the aim to help and assisting responders
(FRDM first responders) in Malaysia to manage disaster
particularly during early stage of response phase. The objective
of this paper is to analyse the system in terms of accuracy of
system (MLP model). Mean Square Error (MSE) value was used
to identify the suitable model for the ANN system. The analysis of
the results shows that the best model of ANN is at 15 neurons
with the MSE of 0.0159 which will be discussed thoroughly in this
paper.