Hybrid Deep Learning and Metaheuristic Optimization Approaches for Short- and Long-Term Urban Water Consumption Forecasting: A Systematic Review (۲۰۱۰–۲۰۲۵)

Urban water consumption forecasting represents a critical component in sustainable resource management, particularly amid growing populations, climate variability, and infrastructural demands. This systematic literature review (SLR) examines hybrid approaches integrating deep learning architectures with metaheuristic optimization techniques for short- and long-term predictions from ۲۰۱۰ to ۲۰۲۵. By synthesizing ۳۰ key studies, we categorize deep learning models such as recurrent neural networks (RNNs), long short-term memory (LSTM) units, and transformers, alongside metaheuristics like genetic algorithms (GA) and particle swarm optimization (PSO). The review highlights integration strategies that enhance predictive accuracy, reduce computational overhead, and address uncertainties in urban water data. Key findings reveal that hybrid frameworks improve short-term hourly forecasts by up to ۲۰% in mean squared error (MSE) and long-term seasonal projections through hyperparameter tuning. We also discuss performance metrics, open challenges such as data scarcity and real-time adaptability, and a roadmap for future research emphasizing multi-modal data fusion and edge computing. This SLR provides actionable insights for water utilities and policymakers to optimize distribution systems and mitigate shortages.

کلیدواژه ها:

Water consumption forecasting ، Deep learning ، Metaheuristic optimization ، Hybrid models ، Urban water management ، Systematic review