Association Analysis Of Monitoring Points Value Of Airport Noise

With the development of civil aviation rapidly in recent year, the noise of airport is becoming atopic which attracts more and more attention. With the monitoring point data set of airport noise,airport noise distribution and trends can be mined in order to provide an important scientific evidencefor predicting the noise value effectively and taking effective measures to prevent airport noisepollution and developing the use of the airport and the surrounding area of the land. This paper studiesassociation rules among the noise value of monitoring points.Association rules is one of the main research of data mining and knowledge discovery, focusingon identifying the data link among the different attributes that meets the support level and theconfidence level. Since R.Agrawal and R.Srikant raised the association rule firstly in1993, there havebeen many association rule mining algorithms.This work is divided into the following areas:1) The current association rule mining algorithms are studied. This paper analyzes current majorresearch association rule mining algorithms and summarizes the advantages and disadvantages ofvarious algorithms.2) Characteristics of airport noise monitoring data set are studied. According to thesecharacteristics, this paper uses DENCLUE algorithm based on density of monitoring points to clusterdata sets, and finds a representative point in each cluster by the principles of climbing algorithm.3) Factors are analyzed which can disturb noise value of monitoring points. Airport noisemonitoring noise value is associated with many factors, such as weather, temperature and humidity,track, type and so on. This paper adapts gray correlation analysis method to analyze the effect ofdifferent types of the same flight and is verified by examples. The result shows that gray correlationanalysis method contributed to select conditions affecting the noise value is feasible.4) An initial set of association rules are established. Firstly, noise value stemmed from thenon-noise events is deleted; secondly, according to the flight track, points which are not on the flighttrack are deleted; thirdly, noise value can be converted into intervals.5) Association rules among noise value of monitoring points are mined. This paper uses Apriorialgorithm to mine association rules. At the same time, this paper also proposes a newalgorithm—ATOSOA-Apirori algorithm and methods are compared between each other. Theexperiment result shows that the proposed algorithm is much more efficient.6) The number of association rules is analyzed by regression analysis. Equations of a variety of rules between the number and parameters are designed. Multiple correlation coefficient method isused to test fitting results for each equation, finally significance test is used to verify the parameters ofthe coefficient is significant to zero. The biggest multiple correlation coefficient of regressionequation is regarded as the optimal fitting equation. The selected optimal equation can predict thenumber of association rules under given parameters very well.