Research On The Method Of Building Incremental Fuzzy Associative Classifier And Its Application

Flight Operations Quality Assurance(FOQA)is an important part of civil aviation safety management.Current practice of FOQA does not consider the relationship between several QAR parameters and exceedance events,it also cannot explain the reason of exceedance events,and does not analysis massive QAR data deeply.Fuzzy Association Classifier(FAC)can reveal the laws between multiple variables and sample categories from a large amount of data,it has a good interpretability.Incremental learning is an effective way to build classifiers on mass data sets.This paper focuses on the construction of an efficient Incremental Fuzzy Associative Classifier and its application in FOQA,it includes the following three parts:(1)An incremental fuzzy associative classification method based on eVQ(evolving Vector Quantization)cluster algorithm is proposed.The eVQ cluster algorithm is adopted to generate and incrementally update the parameters of Gauss membership functions.UWEP(Update With Early Pruning)algorithm is extended to incrementally mine fuzzy frequent itemsets.FCORR(Fuzzy Correlation)of fuzzy associative classification rule(FACR)and the length of antecedent of FACR are regarded as measure to prune and update classification rule base.The experiment results show that compared with the method of the batch building the FAC,the proposed method can reduce the time of training the FAC in the premise of not decreasing the accuracy and interpretability.(2)On the basis of work(1),a fuzzy association classification rule base(FACRB)pruning method based on incremental genetic algorithm is proposed.FACRB is encoded by binary method,genetic algorithm is adopted to prune FACRB.During incremental learning phase,the new population is initialized with genetic information carried by the excellent individuals which retained in the previous genetic process.The results show that the proposed method can prune the FACRB quickly,and the classification accuracy of the model is higher.(3)Exceedance event "15m(50ft)to ground distance" of an airline is identified and analyzed by incremental fuzzy association classifier constructed in this paper.The experimental results show that the proposed method can identified the exceedance event effectively.The fuzzy association rules of classification model can be used to explain the causes of exceedance event.