Optimum Vehicle Evacuation Plan for Nuclear Emergency Using Fuzzy Credibility Theory and Improved Genetic Algorithm

The occurrence of nuclear accidents is unpredictable and destructive, and it is necessary to develop a reasonable evacuation plan in advance to evacuate affected people to safe places. Aiming at the optimization of vehicle evacuation path in a nuclear emergency with fuzzy demand, an uncertainty vehicle evacuation model is modeled with the objective of minimizing the cumulative radiation dose, and a fuzzy-improved genetic algorithm (F_IGA) is proposed. To reduce the influence of parameters on the algorithm, the crossover probability is dynamically adjusted according to the fitness value. Step size is also introduced to improve the position update formula of the elephant herding optimization, and an adaptive crossover approach is proposed, which is used to improve the algorithm's local optimization ability. To improve the global search capability of the algorithm, the roulette wheel strategy and large neighborhood search algorithm are introduced to locally adjusted individuals, and a simulated degradation operator is introduced in the mutation operation. Further, to avoid planning paths that exceed vehicle capacity, different vehicle paths are independently adjusted according to the current environment to reduce the radiation harm to people. Finally, public test cases and road network structures are selected to verify the performance of the algorithm. The results demonstrate that F_IGA outperforms the other algorithms in terms of convergence accuracy, and the number of iteration convergence, and it has more obvious advantages with the increase of the evacuation scale.