Skip to content
SDGs Center UDINUS

Evidence record · UDINUS · SDGs

Prediction of hourly solar radiation in Indonesia using LSTM

Rektorat UDINUS 2022 Published 2026-10-30

Evidence record · UDINUS · SDGs

Unit
Rektorat UDINUS
Period
2022
Goals
Record
1e0fca1e-32a9-4506-83e6-b7f6e120b6ac
Published
2026-10-30

Universitas Dian Nuswantoro conducted a journal activity on predicting hourly solar radiation using Long Short-Term Memory (LSTM). The activity participants were Sari, Dian Puspa; Zainul Fanani, Ahmad; and Shidik, Guruh Fajar. The activity addressed the value of solar radiation in supporting the optimal utilization of solar energy for building design, solar power plants, and agricultural systems. It also described solar radiation as an element that affects other weather parameters.

The activity discussed forecasting data at hourly, daily, monthly, and yearly scales and stated that forecasting processes use different multivariate meteorological parameters. It identified data errors and fast-moving clouds as possible causes of fluctuations in solar radiation. The meteorological parameters used in the activity were solar radiation, sun duration, relative humidity, dew point, air temperature, and sky cover. These parameters were used to predict hourly solar radiation.

LSTM was proposed to identify the best activation function among ReLU, sigmoid, and tanh. ReLU was reported as the best activation because it had the smallest Mean Square Error (MSE), with a value of 0.1885. The MSE values for sigmoid and tanh were 0.2359 and 0.3270. Using ReLU, the wall time was 1min 51s, and the calculation process stopped in the 18th epoch. The optimum learning rate was 0.00249, with a wall time of 1min 38s. The output prediction was hourly solar radiation one-day ahead, which could be used as solar radiation information. In relation to SDG 7, the activity concerns solar-energy utilization and clean energy research through solar-radiation prediction.

Reference https://www.scopus.com/pages/publications/85141508354?origin=resultslist