Analysis Of QAR Data Trend Based On Cluster HMM Mode

In flight,flight safety is always a crucial issue.At present,the major domestic airlines use the quick access recorder(Quick Access Recorder,QAR)as the airborne flight data recording equipment to obtain the aircraft flight data of the daily operation.The parameters?recorded include the flight parameters of the aircraft,the performance parameters of the aircraft engine and the main components,the environmental parameters inside and outside the cabin,QAR monitoring as a scientific and effective technical method,is conducive to ensuring flight safety,improving operational efficiency.The results of the monitoring provide an important basis for flight inspection,safety assessment,safety incident investigation and maintenance.In order to make better use of QAR data to effectively provide support for the aircraft's fault detection and early warning decision.In the face of a large number of QAR data,the data mining method is used to analyze the QAR data,the trend of QAR data is obtained,the state model of QAR data is established by trend analysis,the state of airborne equipment is identified to carry on fault diagnosis and prediction,to provide decision support and security for safe flying.In this paper,for the traditional QAR data analysis method,which only focuses on the anomaly points and ignores the QAR state trend characteristics,a clustering hidden markov model(Hidden Markov Model,HMM)is proposed.It is applied to the QAR data,based on the trend analysis of the bump data in the QAR data and establish the hidden Markov model,the degree of the air bump fault is effectively predicted,and the effectiveness of the method is verified.The research work of this paper mainly completes the following work:1 ? The HMM model based on clustering is studied and the state trend analysis of QAR data is realized by taking the air bump fault as an example.For the application of the HMM model in QAR data to study.In view of that the QAR data is abnormal and will experience different states in the event of a flight failure.An HMM model can be established to describe the state change in the process of aircraft failure.2?For a feature of the QAR is also time series,and the relevant literature of time series is studied.The method of segmentation of time series is studied emphatically,and analyzes the use of QAR data in a variety of linear segmentation methods is analyzed.In addition,inview of the shortcomings of the traditional segmentation mean description segmentation.Using a new method of describing the absolute value of slope and endpoint difference is more conducive to QAR data clustering.3?To study the clustering algorithm of QAR data,in view of the limitations of a based on the breadth-first neighbor search clustering algorithm on the QAR data,it is improved.The main improvement method is to divide the data.The data of each region is segmented,and then the clustering is carried out.The results of clustering are obtained,and then clustering is carried out,that is,quadratic breadth first neighbor search clustering.