Open Access

ARTICLE

A Robust Optimisation Strategy for Active Distribution Networks Using VMD-LSO-LSTM Prediction

Zhongjin Shi^#, Zhe Zang^#, Chong Wang^#, Yue Yang^*
Key Laboratory of Modern Power System Simulation and Control & Renewable Energy Technology, Ministry of Education (Northeast Electric Power University), Jilin, 132012, China
* Corresponding Author: Yue Yang. Email:
# These authors contributed equally to this work

Energy Engineering https://doi.org/10.32604/ee.2026.074923

Received 21 October 2025; Accepted 01 December 2025; Published online 04 February 2026

Abstract

To address the critical challenges of power fluctuations and the imperative for efficient reactive power optimization in distribution networks with high photovoltaic (PV) penetration, this study proposes an innovative solution: a robust reactive power optimization approach that integrates VMD-LSTM-based PV power forecasting with the advanced Lion Swarm Optimization (LSO) algorithm. The methodology commences by employing Variational Mode Decomposition (VMD) to decompose the PV power sequence into distinct modal components in a seamless manner. Each modal component is subsequently modeled using a dedicated forecasting framework built on Long Short-Term Memory (LSTM) networks, with the LSO algorithm optimizing the LSTM hyperparameters to attain outstanding prediction precision. Building upon these precise forecasts, we establish a robust reactive power optimization model tailored for distribution networks, thoughtfully accommodating the inherent uncertainties of PV generation. This dual-layer model has an upper layer that addresses challenging PV scenarios and a lower layer that orchestrates the coordinated scheduling of energy storage systems, capacitor banks (CB), and Static Var Compensators (SVC). The primary objective is to minimise network losses while optimising reactive power compensation strategies that enhance overall performance. Case studies reveal that our innovative VMD-LSO-LSTM model significantly outperforms single LSTM and VMD-LSTM models in predictive accuracy. Additionally, the robust reactive power optimization strategy effectively dampens power fluctuations, decreases network losses, and substantially boosts distribution network operations’ economic efficiency and stability—ensuring a more resilient and sustainable energy future.

---

Keywords

Active distribution network (ADN); robust optimisation; photovoltaic (PV) forecasting; lion optimisation algorithm

---