Analysis And Research On Fuel Consumption Of Civil Aviation Aircraft Landing Process Based On The Improved K-means Algorithm

The aircraft fuel consumption is one of the most important parts of airline operating costs,and exhaust emissions generated by aircraft gradually have become an essential aspect of urban air pollution.Reasonable and effective control of aircraft fuel consumption is of great significance to reduce airline operating costs and environmental pollution.Compared to the aircraft cruise stage,the aircraft landing is one of the most complicated stages in the flight process.Through scientific plan and effective design for the aircraft landing,the aircraft fuel consumption can be greatly reduced.To investigate the potential relationship between the aircraft fuel consumption and the aircraft height,weight,speed,and airport environment during landing,it is necessary to divide aircraft fuel consumption data into different clusters and conduct classification research,so as to provide reasonable suggestions to reduce aircraft fuel consumption during landing.Therefore,in this paper,the aircraft fuel consumption during landing is studied based on the controllable factors both in airlines and airports using clustering analysis.As a kind of important method for data grouping and data division in the field of data mining,cluster method relies on the data correlation to find the similar data type or data sets,and extract the similarity of data without prior information,so as to classify data sets from the view of attributions.K-means is a widely applied cluster method and it is suitable for clustering analysis of massive data.However,there are some problems in traditional K-means such as the number of clusters is hard to determine,the algorithm is easy to trap into the local optimum,and the cluster boundary is difficult to decide.To solve these problems,this paper proposes an improved K-means algorithm to analyze the aircraft fuel consumption data during landing and conducts a case study based on the real flight records of Boeing 737 s operated by China Eastern Airlines.The main research contents and contributions of this paper are summarized as follows:(1)A new cluster effectiveness evaluation index is proposed from three aspects:the sum of similarity,separability,and degree of overlapping.We adopt the landingdata of civil aviation aircraft for clustering analysis which is characterized by strong continuity,small difference between data,and difficulty in determining category size.The traditional clustering effectiveness evaluation index usually only adopt the sum of similarity or the sum of separability,so when it is applied to the analysis of landing data of civil aviation aircraft,it cannot identify the boundary between different classes in the clustering process,the range size of different classes and the optimal number of clustering cannot be determined.On the basis of previous studies,this paper considers the sum of similarity,separability,and degree of overlapping at the same time,and proposes a CPC clustering effectiveness index that comprehensively measures the effectiveness of clustering results.Four sets of artificial numerical experiments and four sets of real value experiments verify the advantages of the proposed CPC clustering effectiveness index in determining the clustering quality.(2)In order to deal with data which has strong continuity and small differences,an improved K-means clustering algorithm is proposed.To reveal the relationship between the real distribution of in the data set and the data distribution generated by clustering division,this paper studied the uniform effect of clustering from the perspective of data distribution and proposed an improved K-means clustering algorithm.The clustering performance of the proposed improved K-means algorithm using different types of data sets is discussed according to the change of the coefficient of variation CV,and the cross-validation is conducted by the artificial data set and the CPC clustering effectiveness index.The experimental results prove that the improved K-means algorithm can guarantee a relatively good clustering quality using the data sets which have differences in the total number of samples,number of categories,and number of attributes.In addition to verifying by manual data set,this paper also uses real data sets to verify the improved K-means algorithm.The experimental results prove the wide applicability of the improved K-means algorithm.(3)From the perspective of the controllable factors of airlines,the improved K-means algorithm is applied to the clustering analysis of Boeing 737 fuel consumption data during landing.At present,the fuel consumption prediction model of civil aviation aircraft is mainly based on the flight manual data,and the modelanalysis process is mainly dependent on the query of the flight manual.To satisfy general applicability and safety,flight manual has relatively broad requirements for the flight process,which means it also has low requirements for fuel-saving flight.For these reasons,the improved K-means algorithm is used to cluster two Boeing 737 aircraft fuel consumption data during landing at different airports and provides more refined landing suggestions for pilots so as to further supplement the requirements in the flight manual considering energy conservation and emission reduction.In addition,this paper conducts sensitivity analysis for the improved K-means algorithm from four aspects: optimal k value selection,algorithm robustness,CPC clustering effectiveness,and different airports,the results verified the reliability of clustering results generated by the improved K-means algorithm.(4)From the perspective of airport environment,the improved K-means algorithm is applied to the clustering analysis of Boeing 737 aircraft fuel consumption data during landing on different airports.Firstly,the specific factors that may affect the fuel consumption during landing are discussed through basic geographical environment,airport construction,and airport climate.Then the relevance between the various factors are analyzed,six factors which have large influences on civil aviation aircraft fuel consumption during landing and have low relevance with each other are selected as variables to analyze the impact of different airport environmental factors on civil aviation aircraft fuel consumption during landing.After that,21 major domestic airports are selected as the analysis targets,the improved K-means algorithm is applied to the clustering analysis process of airport environmental factors and fuel consumption.The fuel consumption of civil aviation aircraft landing in different airports is clustered to explore the impact of different airports on civil aviation aircraft fuel consumption during landing.In particular,the influence of airport environmental factors on civil aviation aircraft fuel consumption during landing is studied.Based on the results of clustering analysis,some suggestions for the construction of hub airports are proposed,which can reduce the fuel consumption of civil aviation aircraft during landing.This paper puts forward an improved K-means clustering algorithm and applies itto the clustering analysis of aviation fuel consumption in the landing process of civil aviation aircraft.From the perspective of energy conservation and emission reduction,it provides management suggestions for airline pilots' driving behaviors and airport construction.Also,some innovative research results are obtained,which has important theoretical and practical significance.