كنفرانس بين المللی مهندسی صنايع و سيستم ها

صفحه اصلی / دوازدهمین كنفرانس بين المللی مهندسی صنايع و سيستم ها

Lexicographic Industrial Freight-Wagon Loading: An Adaptive Multi-Start Heuristic with Exact Repartitioning

نویسندگان :

Amirreza Taraghi¹ Farzad Fathalizadeh Sohrabi² Majid Mardani Shahri³

1- گروه مهندسی صنایع، دانشکده مهندسی، دانشگاه فردوسی، مشهد، ایران 2- گروه مهندسی صنایع، دانشکده مهندسی، دانشگاه فردوسی، مشهد، ایران 3- گروه مهندسی صنایع، دانشکده مهندسی، دانشگاه فردوسی، مشهد، ایران

کلمات کلیدی :

Freight-wagon Loading،Lexicographic Optimization،Set Partitioning،Exact-Neighborhood Search،Industrial Heuristics،Uncertainty Analysis

چکیده :

This paper addresses an industrial freight-wagon loading problem where indivisible steel consignments are assigned to product-homogeneous wagons with exactly three items each. The task is modeled as fixed-cardinality set partitioning with eight lexicographic priorities covering capacity violations, overload, target fulfillment, reserve load, and dispersion. We propose a solver-free heuristic combining multi-start construction, exact two- and three-wagon local search, reserve-aware intensification, and product-wise decomposition. A Pyomo model solved with CPLEX V12.10.0 provides eight-stage exact validation. The case includes 151 records (3,451.180 ton) forming 51 wagons, with all items assigned once, no product mixing, and no hard-capacity violations. Across the six industrial products, the heuristic matched the certified P_1-P_8 vectors in all 60 product-seed comparisons generated by 10 seeds. Median workflow time was 0.4642 s versus 20.3010 s for CPLEX, corresponding to an observed case-specific runtime ratio of 43.731. Across 108 synthetic heuristic runs, all runtimes remained below 3.8 s; among the 45 comparisons with fully proven CPLEX references, 26 matched the complete objective vector. The method is practical and auditable but remains heuristic without global optimality guarantees.