Renewable energy has a constantly increasing role in modern infrastructure. Distributed systems allow for lower transmission losses and in situ generation. However, integrating renewable energy production into existing infrastructure poses several challenges. The intermittent nature of renewable energy production and the high cost of storage make it difficult to rely solely on these sources. A robust method for forecasting production could significantly enhance integration into existing networks. Artificial intelligence (AI) techniques have shown promising results in tackling time series forecasting. Nevertheless, AI models require careful tuning to achieve the desired outcomes, a challenge that is considered NP-hard. This work proposes a modified metaheuristic algorithm based on the Particle Swarm Optimization (PSO) algorithm to optimize the parameters of long short-term memory (LSTM) networks and improve forecasting accuracy. The proposed approach is evaluated on a real-world solar power production dataset, yielding promising results.