Multi-Objective Optimization of Perishable Inventory Systems Using Adaptive Genetic Algorithm with Fuzzy Demand Forecasting

Managing inventory for perishable products presents significant challenges due to uncertain demand, limited shelf-life, and competing objectives involving cost minimization, service level maximization, and waste reduction. This study introduces a novel hybrid method-Adaptive Genetic Algorithm with Fuzzy Demand Forecasting that advances existing hybrid optimization frameworks by incorporating dynamic parameter tuning mechanisms. Unlike conventional GA-Fuzzy hybrids that rely on fixed evolutionary parameters, the proposed AGA-Fuzzy automatically adjusts its tournament size, crossover probability, and mutation strength in response to population diversity and convergence progress, thereby maintaining an optimal balance between global exploration and local exploitation without manual calibration. The fuzzy forecasting module models demand uncertainty using triangular fuzzy numbers, enabling more realistic handling of ambiguous and volatile demand patterns. Comprehensive simulation experiments across six distinct demand scenarios demonstrate that AGA-Fuzzy achieves an average cost reduction, waste reduction, and service levels outperforming traditional methods such as EOQ, (s,S), and standard GA. These findings validate the algorithm's adaptive and fuzzy synergy as a robust, intelligent, and computationally feasible approach for managing perishable inventory under uncertainty. The framework provides practical decision support for inventory managers seeking sustainable and data-driven optimization strategies.

کلیدواژه ها:

Genetic algorithm ، Perishable inventory ، Fuzzy logic ، Demand forecasting ، Multi-objective optimization ، Supply chain management