An Ensemble Forecasting Method based on optimized LSTM and GRU for Temperature and Humidity Forecasting

Temperature and humidity predictions play a crucial role in various sectors such as energy management, agriculture, and climate science. Accurate forecasting of these meteorological parameters is essential for optimizing crop yields, managing energy consumption, and effectively mitigating the impact of climate change. In this context, this paper proposes an enhanced ensemble forecasting method for day-ahead temperature and humidity predictions. The proposed method integrates a Long Short-Term Memory (LSTM) network, a Gated Recurrent Unit (GRU), Particle Swarm Optimization (PSO) and Bayesian Model Averaging (BMA). PSO is employed to optimize the parameters of the LSTM and GRU, thereby improving forecasting accuracy. The method is implemented using Python 3.10 with TensorFlow. Additionally, the proposed approach is compared with ensemble-1, LSTM, and GRU models to demonstrate its effectiveness. The simulation results confirm the superior performance of the proposed method over existing competitive approaches.