TY - JOUR
TI - Short term wind speed prediction using search and rescue optimization with deep belief network on scatterometry data
AU - Almulhim Fatimah
AU - Ishak Ben Anis
AU - Elhag Azhari A
AU - Aljohani Hassan M
AU - Abdel-Khalek Sayed
JN - Thermal Science
PY - 2024
VL - 28
IS - 6
SP - 5097
EP - 5111
PT - Article
AB - Scatterometry is a technique used to transmit radio or microwaves to examine  different geophysical properties, wind speed, and direction. Precise and  rapid weather predictions become essential in several applications in  assisting planning and management in response to weather conditions. At the  same time, timely wind speed prediction gains considerable attention in  several economical, business, and management areas. With the consideration  of wind speed as an arbitrary variable, precise wind speed prediction using  machine learning and deep learning models can be established. With this  motivation, this study develops a short-term wind speed prediction using  search and rescue optimization with deep belief network (STWSP-SRDBN) model.  To accomplish accurate wind speed prediction, the STWSP-SRDBN method  initially pre-processes the weather data using min-max normalization.  Additionally, the STWSP-SRDBN model utilizes DBN model to predict the  weather data. Moreover, the SRO algorithm is utilized to fine tune the  hyperparameters related to the DBN approach. The presented STWSP-SRDBN  method makes use of Spatio-temporal multivariate multi-dimensional  historical weather data to learn new representations utilized for wind  forecasting. The experimental validation of the STWSP-SRDBN method is tested  using a set of weather data and the outcomes are investigated under numerous  aspects. The experimental results indicated the enhanced outcomes of the  STWSP-SRDBN method over recent state of art methods.