Research On Aircraft Type Recognition Method Based On Call Background Sound

Currently,for intercepting aircraft short-wave radio communication audio,the task of identifying the type of aircraft is mainly to identify the sound of the aircraft engine by means of manual interception to infer the type of aircraft.However,this method of identifying by people’s ears is often very inaccurate.There are two main reasons,first,the intercepted sound signals are very short,they may be tens of milliseconds;the other is the intercepted sound signals of aircraft mixed with a variety of noise,making the identification difficult.It can be seen that it is of great significance to study the classification and recognition of aircraft cabin sound and background sound of the pilot’s call.At present,there are few literatures on aircraft type recognition through aircraft cabin background sound.This paper mainly studies two methods for feature extraction of sound signals,and the simulation experiments were carried out respectively.Through the analysis,it was concluded that the signal characteristics calculated by Mel frequency cepstrum coefficient algorithm are more suitable for the sound signals used in this paper.According to the obtained characteristics of signals,the BP neural network,naive Bayes classifier and support vector machine were used to classify and recognize the sound signals of eight types of aircraft.Compared with BP neural network and naive Bayesian classifier,the support vector machine is more suitable for classifying small-scale data and has higher classification accuracy.However,the effect of the support vector machine is greatly affected by the parameters.To find the optimal parameters,the particle swarm optimization algorithm and harmony search algorithm are used to optimize the parameters of the support vector machine.In this paper,we use BP neural network,naive Bayes classifier,support vector machine,particle swarm optimization support vector machine and harmonic search algorithm optimized support vector machine to simulate the sound signal recognition experiments.Simulation experiments show that when using the Mel frequency cepstral coefficents algorithm to extract the characteristics of the eight aircraft sound signals for classification experiments,the harmonic search algorithm optimized support vector machine has the highest classification accuracy and is more suitable for aircraft sound signals studied in this paper.