Research On The Model Of Emergency Decision Making For Building Fire Based On CBR And Bayesion Network

With the rapid economic development,the increase of the number and kinds of building constantly,constantly enrich the types of building materials,which is associated with building emergency fire occurred frequently in recent years,has caused the attention of the countries all over the world,international scholars have carried on the thorough research.Due to the structure of the emergency fire has many characteristics,such as uncertainty,time urgency,when facing the building emergency fire,how decision makers in the shortest possible time to make effective emergency decision-making,and is the key point of this study.Effective emergency decision can reduce the tendency of fire and reduce casualties and economic loss.Bayesian network includes the knowledge of graph theory and probability theory knowledge,is a way to deal with uncertainty problems,due to emergency fire building has the characteristics of uncertainty,so this paper USES bayesian network to build the building's emergency fire emergency decision-making model.Due to the influence factors of building emergency fire is more complex,use traditional bayesian network modeling method is more complex,the consumption of emergency decision-making time is long,affects the effect of emergency decision.How to building emergency fire happened in the past by the emergency fire emergency decision is applied to the current buildings and emergency decision-making of the in order to save time,therefore,in this paper,the case-based reasoning technology as ancillary building emergency fire emergency decision-making of the construction of the bayesian network model,bayesian networks and case-based reasoning technology combine to build emergency decision-making model,is a focus in the study of this article,is also the difficulty.The main work of the thesis is as follows:First of all,the case database of building emergency fire emergency decision is built.In this paper,the case database of emergency fire emergency decision of buildings is divided into the database,text information database and the network structure of the bayesian network structure of emergency decision making.The historical building keyattribute of the emergency fire feature extraction,using framework representation,said and then analyzed the properties of the target case features,and putted forward the historical buildings of the emergency fire emergency decision-making case case retrieval and matching.Case retrieval divided into two steps,case index and case retrieval,case retrieval by using analytic hierarchy process(ahp)to determine the weights of case attributes,solving the hamming distance of the degree of similarity between cases,choose matching case similarity of cases and the ultimate goal of case matching is matched to the historical case emergency decision-making bayesian network model.In this paper,20 historical cases of sudden fire in buildings are used as the case library.Secondly,after a matching case retrieval and case study,might get several similar bayesian network structure of the history of the case,you need to this a few integrating useful part of the bayesian network structure,and get the target case of bayesian network structure.When the target case is matched to a similar historical case,the bayesian network model of the historical case needs to be modified according to the actual situation of the target case,that is,the case revision.It is necessary to adjust the network structure of bayesian network structure according to the actual situation of the target case and the change of the conditional probability of network node.Introduces several methods of bayesian network modification,after correction of bayesian network,get the target case bayesian network structure,the use of bayesian software Ge NIe2.0 on the parameter sensitivity analysis and studies.For high-rise building emergency fire together in the end,for example verification model,for its historical cases of emergency decision-making the specific process of the construction of the bayesian network model,for example for 20 cases of putted forward modeling,will build the bayesian network model is stored in the case of network structure in the library,in order to for case retrieval.The bayesian network of the successful cases was modified to obtain the bayesian network structure of the current case,and the parameter learning and sensitivity analysis were carried out.