Authors: Vikas Kumar Vidyarthi; Pragya Mukherjee; Shikha Chourasiya
Addresses: Department of Civil Engineering, National Institute of Technology, Raipur, Chhattishgarh 492010, India ' Department of Civil Engineering, SRM IST, Modinagar, Ghaziabad, UP 201204, India ' Central University of Jharkhand, Ranchi, 835205, India
Abstract: The forecasting of relative humidity (RH) is very important in planning various industrial activities and in designing future climate control systems. However, research on forecasting of RH is very few and far. In this study, a novel technique is proposed for forecasting one-day ahead RH using artificial neural network (ANN) and multiple linear regression (MLR) techniques by reducing the number of variables in input space gradually for an India region. The results show that both ANN and MLR models forecasted one-day ahead RH equally well. The ANN and MLR models which even use only lagged RH values performed equally well with nearly similar values of R (0.969 and 0.966), and RMSE (0.055 and 0.057), but MLR model has an advantage of being simpler and hence the present study recommends the use of MLR technique for RH forecasting. Also, the lagged RH values are sufficient for forecasting one-day ahead RH.
Keywords: relative humidity; artificial neural network; ANN; multivariate linear regression; MLR; hydrology; building and environment; climate control systems.
International Journal of Hydrology Science and Technology, 2023 Vol.15 No.3, pp.219 - 236
Received: 06 Jul 2021
Accepted: 02 Oct 2021
Published online: 05 Apr 2023 *