Research On The Optimization And Decision Models Of Emergency Material Collecting Toward Earthquake Disaster

The frequent occurrence of earthquakes in recently years illustrates that anoutbreak of earthquake usually triggers a rapid expansion of demand for emergencymaterials in the disaster area, which poses great challenges to emergency materialcollecting. The low efficiency and high cost of this activity might cause moredisaster casualties and property damage, and induce secondary disasters. Therefore,it has become a hotspot being continuously studied in the field of emergencylogistics how to collect emergency materials with high efficiency and low cost afterthe outbreak of earthquake.However, the current researches on emergency material collecting mainlyfocused on some certain parts of emergency material collecting system or on thestructure of emergency material collecting network, with no systematic researchedon the theories or decision-making methods involved, so that it is impossible toapply and commercialize most of the research results. Furthermore, according to theresearch methods, most of these researches were descriptive and macroscopic, whilefew were quantitative and operable. In order to improve the efficiency and benefit ofemergency material collecting for earthquake disasters, and to secure rapid andstable provision of emergency materials, it is necessary to perform systematicresearch on the critical management in emergency material collecting system,especially on the collecting procedure in the context of changing emergencymaterials along with time and space, as well as its optimization. Currently, however,there was little literature concerning such subject.Based on emergency material mobilization theory, earthquake emergencyresponse phasing theory, emergency state termination theory and hub-and-spokenetwork theory, this dissertation employed different research methods includingmulti-objective optimization for constrained problem, optimal stopping decision,stochastic process, intelligent optimization technique and combinedqualitative-quantitative research method, to dig into some frontier topics, such asprediction of emergency material demand, dynamics optimization and collectingdecision-making for hub-and-spoke emergency material network, randomizedtermination model of emergency material collecting and adaptability assessment ofemergency material collecting network.Firstly, this dissertation put forward the concept to establish a hub-and-spokenetwork of emergency material collecting for earthquake disasters, analyzed themethod to choose the stations of this network, and presented the structuralcharacteristics and the topological structure of hub-and-spoke network adapting to the different stages of emergency material collecting.Secondly, in the initial stage of earthquake disaster emergency, based on taskshifts in emergency material collecting and the target of emergency management forearthquake disasters, this dissertation discussed the three categories key optimaldiscision problems about emergency material collecting. First of all, in order topredict emergency material demand without complete information under the workenvironment of hub-and-spoke emergency material collecting network, thisdissertation analyzed the method to decompose the indicators of emergency materialdemand in the early stage of earthquake emergency response which was positivelycorrelated to population size, and build a robust wavelet v SVM (Support VectorMachine, SVM) model which could effectively solve the prediction of emergencymaterial demand with small sample size, non-linearity and high dimensionality.Then this dissertation build emergency material collecting models with no constraintof time limit when the demand was a fuzzy interval, hub-node was single-hub andbuild emergency material collecting models with no constraint of time limit whenthe demand was fuzzy number, hub nodes were with limited, while the type of hubnodes was multi-hub.Thirdly, for the intermediate stage of earthquake disaster emergency, with theconstraint of time limit, taking collecting time as the priority target to be optimized,this dissertation mainly explored the emergency material collecting strategies ofemergency material collecting system under the restraints of collecting amount andcollecting cost and this dissertation explored the decision problems of emergencymaterial collecting under the constraint of time limit, the fuzzy interval collectingtime and multi-hub. this dissertation build the decision-making model of collectingamount and collecting cost based on two different starting times t0=0and t0≠0,and build double-deck multi-hub emergency material collecting model with fuzzycollecting time. for the late stage, according to the random change of the state ofemergency material collecting, this dissertation employed Markov stochasticdecision making method and optimal termination criteria to explore the maximumtermination time and optimal stopping time of emergency material collecting forearthquake disasters, and build the randomized stopping model of emergencymaterial collecting based on Markov method and optimal terminationdecision-making method.At last, in order to perform adaptability assessment for the operation ofemergency material collecting network, this dissertation developed an adaptabilityassessment system based on improved Compensatory Fuzzy Neural Network(CFNN). Through extracting indicators and building models, this dissertationeffectively solved the fuzziness and non-linearity of collecting time, collecting costand network stability indicator in assessing emergency material collecting network; through improving the dynamic global optimization algorithm, this dissertationallowed the CFNN to be trained starting form a fuzzy rule with correct or incorrectinitial definition, which effectively covered the disadvantage of typical fuzzy neuralnetwork as was likely to be trapped into partial optimization, and improved the faulttolerance and robustness of the model.This dissertation introduced hub-and-spoke network in building emergencymaterial collecting network, and developed the theories and decision-makingmethods of emergency material collecting as a system, aiming at addressing theweakness in making decisions for current emergency material collectingmanagement, which was promising to develop the theory of emergency logisticsmanagement, to avail rapid response of emergency material collecting forearthquakes, and to promote the efficiency and benefit of this activity.