Research On Emergency Resources Scheduling Methods Based On Memetic And Multi-agent Genetic Algorithms

The frequent occurrence of large-scale natural disasters has exerted great influence on the people's life.In order to minimize the impact of natural disasters and protect the lives and property of the public,it is very important to establish an effective disaster emergency resources scheduling system.A good emergency resources scheduling system can ensure that emergency resources are delivered to the disaster points in a timely and efficient manner and mitigate the impact of disaster in the maximum extent.This thesis mainly studies the emergency resources scheduling problem from the aspects of models and algorithms.The main research work is summarized as follows:(1)For the problem of minimizing transportation costs during the process of emergency scheduling after disaster,we designed a corresponding single-objective model,which is to minimize the total transport costs of emergency resources scheduling.The model consists of two parts.The first part is the transportation cost caused by emergency resources scheduling without delaying.The second part is the transportation cost of emergency resources scheduling with delaying.For the proposed model,we design a memetic algorithm based on emergency resources scheduling(MA-ERS).MA-ERS includes selection operators,crossover operators,mutation operators,and local search operator.Experimental results showed that MA-ERS has achieved good performance compared with Genetic algorithm(GA).(2)For the problem of minimizing transportation costs and transportation time during the process of emergency resources scheduling,we design a multi-objective model to minimize the transportation costs and transportation time simultaneously.For the proposed model,we design a multi-agent genetic algorithm based on emergency resources scheduling(MAGA-ERS).In order to improve the optimization efficiency of MAGA-ERS,we design the neighborhood crossover operator and mutation operator.By comparing MAGA-ERS with GA and MA,experimental results illustrate that MAGA-ERS is superior to the corresponding GA and MA.(3)For the ambulance scheduling problem during the process of disaster relief,we design a multi-objective model,which is to minimize the latest rescue time among the last marked red patient in all marked red patients and the last marked green patient in all marked green patients.For the proposed model,we design a multi-agent genetic algorithm for ambulance emergency resources scheduling(MAGA-AERS).By comparing MAGA-AERS with GA and MA,experimental results show that MAGA-AERS is much better than GA and MA.