Research And Implementation Of Water Transport Accidents Cause Analysis And Prediction Model

China has abundant water resources and water transport is a very important mode of transport.Most of China's foreign trade transportation is done by water transport.After the reform and opening up,along with the progress of international tradition,the state has invested heavily on the marine industry,so that the coastal cities become more active and water industry and broad development prospects.As a means of transportation,the water transport accidents are inevitable.Because of the mode of transport,the consequences of the accident is often more serious.Therefore,it can use data mining,machine learning,pattern recognition and other advanced technology to help water transport administrative department analysis and explore the deep-seated reasons of water transport accident.Then it will be of great significance for analysis and prevention of accidents within a certain range.Through understanding water transport accident research information of domestic and overseas,combined with existing research methods and results,learning related algorithm,this paper use Bayesian Network technology as a core method to establish cause analysis and prediction models.According to characteristics of water transport accidents data and purpose of the study,this paper comes up with a new Bayesian Network structure learning method based on feature clustering and feature mapping.Then get some useful rules on the prevention of accidents,In this paper,the research work carried out as follows:Firstly,this paper analyzes the impact factors of water transport accidents,and selects the appropriate features according to the data characteristics.The data is preprocessed by missing value fill technology,data discretization technique and experts' advice.Secondly,this paper research the data mining algorithms such as association rules algorithms,Bayesian network structure learning algorithms,probabilistic inference algorithms,clustering algorithms,neural network algorithm and so on.According to the research purposes,the algorithms mentioned above are combined to become a new Bayesian network structure learning algorithm based on feature fusion and this new algorithm is used to build model.Lastly,this paper use the new algorithm to realize the water transport accidents cause analysis and prediction model.And analysis causes of the accident by Bayesian network structure,using the network structure and probabilistic reasoning algorithm to do predictive analysis.By the research and realization of water transport accidents cause analysis and prediction model,combined with real marine data,the paper analyzes how the human factor,location factor,ships factors,weather factors affect the accident,which has a certain reality guiding signification.And use prediction of the model to predict the severity of the accident with a high accuracy.