Emergency Vehicle Routing Problem Applying Variable Neighborhood Search Algorithm

Vehicle routing problem is not only the key problem of normal logistics and supplychain optimization decisions, but also means a lot to improve the rescue efficiency inemergency rescue. Compared with the normal logistics, emergency logistics has manyspecialties such as uncertainty, time urgent and weak economy. Most of the currentresearch of vehicle routing problem is combinatorial optimization models based ondeterministic demand and Euclidean distance assumption, and the focus is usually ondifferent constraints, including time window constraint, multi-depot constraint, multi-vehicle and open vehicle route, as well as few extensions of hybrid constraints. Thealgorithm is mostly heuristics, and the high quality of solution and improvement ofconvergence are achieved by modification of operators. Most of the emergency vehiclerouting research is concentrated on demand uncertainty with a minimum total time, whichis somehow not enough to simulate the emergency situation.In this paper, the author tries to study the vehicle routing problem of a single materialfrom multi-rescue centers to multi-disaster locations with mixed vehicles, and eachdisaster point has a latest served time, which belongs to a fuzzy set. During the routeplanning and optimizing procedure, not only the capacity constraints of vehicles have tobe satisfied, but also the time window constraints and maximum route time constraints.Firstly, the author introduces the background and current research situation of emergencyvehicle routing problem. Secondly, the articles are reviewed by the difference of constraintand algorithm. Considering the real life rescue under emergency circumstance, weformulate the vehicle routing model for this paper, the time window constraints arepresented in object function by penalty of violation. According to some specific rules, weconstruct an initial solution and design a series of neighborhood and local optimizeoperations, by which we apply the variable neighborhood search algorithm to solve themodel. At last, we verify the efficiency of the algorithm by experiment tests.