Robust Optimization of Dangerous Goods Delivery Path Based on Bi-level Programming

Abstract

Aiming at the optimization problem of dangerous goods distribution routes in multi-distribution centers with time windows under uncertain environment. Proposes a robust optimization method with robust control parameters comprehensively considers the transportation risks of dangerous goods. Transportation expenses and service time windows, Constructs a multi-objective double-layer robust optimization model of dangerous goods distribution routes, Upper layer model tracing for minimizing transportation risks and transportation costs, the lower model adopts the user equilibrium traffic allocation model; According to Bert Simas-Sim and Robust Optimization Theory. Robust Equivalence Transformation of Upper Model with Uncertain Parameters; Joint Enhancement, Pareto, Genetic Algorithm and Frank-Wolfe Algorithms, a multi-objective solution is constructed. The hybrid algorithm of standard double-layer robust optimization model, adopts 3 segment coding and decoding method allele matching crossover operation and inversion mutation genetic operation method to solve the upper layer model, adopts frank-Wolfe algorithm to solve the lower layer user equilibrium model; take classic sioux-falls and traffic network as an example XZ80 16 Case Analysis of Dangerous Goods Distribution Route Optimization Problem with 3 Distribution Centers 7 Demand Points, to Verify Rationality of Model and Its Algorithm. Research results show that the hybrid algorithm: when the robust control parameters are 0 30 and 60 and, can quickly obtain 32 and 3 groups of robust optimal solutions respectively, and all solutions are distribution schemes including specific transportation sections and departure times, instead of delivery order; compared with traditional two-stage heuristic algorithm saves 54 in operation time 74%. The hybrid algorithm is better than two-stage heuristic algorithm, in solving efficiency, and expression form of solution. It can better complete multi-objective and double-layer robust optimization task of dangerous goods distribution path in uncertain environment