<p>This study proposes a novel approach to demand forecasting in the boiler manufacturing industry by combining Convolutional Long Short-Term Memory (ConvLSTM) networks with Enhanced Osprey Optimization (EOO). Demand forecasting plays a critical role in improving production planning, inventory management, and resource allocation. However, the inherent complexity and variability of demand patterns make accurate predictions challenging. In this research, we introduce a hybrid forecasting model that leverages the strengths of ConvLSTM networks to capture both spatial and temporal dependencies within demand data, leading to more accurate predictions compared to traditional methods. To further enhance the model’s performance, EOO is applied to optimize the network’s parameters, improving its global search ability and avoiding local minima. The resulting approach offers more reliable demand forecasts, enabling manufacturers to make data-driven decisions for better planning, inventory control, and resource management, ultimately boosting operational efficiency and customer satisfaction.</p> Graphical abstract <p></p>