Bulletin of Electrical Engineering and Informatics
Vol 15, No 5: October 2026

Hybrid multi-task deep learning framework for joint forecasting of solar and wind power

Assemgul Tynykulova (Astana International University)
Raushan Moldasheva (Kh. Dosmukhamedov Atyrau University)
Elmira Eldarova (Kh. Dosmukhamedov Atyrau University)
Yerlan Izbassarov (Kh. Dosmukhamedov Atyrau University)
Shynar Kodanova (Atyrau University of Oil and Gas named after S. Utebayev)
Lunara Diyarova (Non-Profit Joint-Stock Company «Zhangir Khan West Kazakhstan Agrarian-Technical University»)
Saya Baigubenova (Non-Profit Joint-Stock Company «Zhangir Khan West Kazakhstan Agrarian-Technical University»)
Zhanar Azhibekova (Non-Profit Joint-Stock Company «Zhangir Khan West Kazakhstan Agrarian-Technical University»)



Article Info

Publish Date
01 Oct 2026

Abstract

This study presents the adaptive three-expert ensemble (A3E), a reproducible framework for joint one-hour-ahead solar and wind power forecasting. A3E combines temporal, physically informed, and high-generation Extra Trees experts through a causal local-error gate and is evaluated under a strictly chronological, leakage-free protocol. After timestamp reconciliation and causal feature construction, the primary source yielded 8,700 model-ready observations: 5,775 for training, 1,463 for validation, 719 for internal testing, and 743 for a same-source December holdout. The primary scale-neutral comparison used Macro train-SD normalized root mean square error (NRMSE) on 672 common forecast origins. A3E achieved 0.1978, compared with 0.2314 for the official photovoltaic (PV)-Client 2026 comparator and 0.2340 for TimeMixer. However, validation-optimized static weights achieved 0.2017 on the complete 719-origin internal test, compared with 0.2025 for adaptive A3E; the difference was not statistically significant under dependence-aware analysis. On the independent CISO and ERCO systems, TimeMixer outperformed A3E, demonstrating source-dependent model rankings. Event evaluation used 98 CISO and 320 ERCO high-generation events. The results support the value of heterogeneous expert integration, but not a universal accuracy advantage of adaptive weighting.

Copyrights © 2026






Journal Info

Abbrev

EEI

Publisher

Subject

Electrical & Electronics Engineering

Description

Bulletin of Electrical Engineering and Informatics (Buletin Teknik Elektro dan Informatika) ISSN: 2089-3191, e-ISSN: 2302-9285 is open to submission from scholars and experts in the wide areas of electrical, electronics, instrumentation, control, telecommunication and computer engineering from the ...