Design And Implementation Of Data Acquisition And Processing System For Helicopter Sound Detection

This thesis describes the algorithm and software system design and implementation of detecting,classifying and locating the noise signal based on the aerodynamic rotor of helicopter.At present,the detection and identification of helicopters mainly relies on infrared,radar,video and other detection methods.As a detection method,audio signal has the advantages of all-weather and all-time monitoring,and can be used as a complementary means of the above methods.There are few software systems for detection and recognition of helicopter acoustic targets.Based on the technical research project of Aerodynamic Noise Control Key Laboratory of China Aerodynamics Research and Development Center,this thesis designs relevant algorithms and software systems.The main work is as follows:Firstly,an integrated detection and recognition method for helicopter acoustic targets is proposed,and the training of deep neural network is optimized based on the model transfer method in transfer learning.There are some problems in helicopter detection and recognition,such as poor continuous recognition effect,separation of detection and recognition methods.We design the neural network algorithm that combines detection and recognition,and at the same time perform murmur suppression on the original audio data to solve the above problems.In order to solve the problem of small data samples and difficulty in obtaining helicopter audio signals,we transfer the network model used for environmental sound recognition to the helicopter acoustic target recognition problem through model fine-tuning based on source domain training,and significantly improve the recognition accuracy of the algorithm under small sample data sets.Then,relying on the hardware equipment,secondary development interface and key algorithm model provided by the cooperative unit,the helicopter acoustic detection data acquisition and processing software was designed and developed.The function and performance requirements are established through requirement analysis,and the outline design is carried out based on MVC architecture.Finally,the detailed design and development are carried out according to the interface interaction design and specific function realization of each module,and the unit test,integration test and system test are carried out according to the standard software development process.Finally,the actual flight test is carried out under the cooperation organization,which verifies that the algorithm model and software system proposed in this thesis can be practically used for helicopter target detection in the real environment.In this thesis,the detection and recognition method of helicopter acoustic target is studied,and the helicopter acoustic detection data acquisition and processing system is designed and implemented.It can be used as a supporting means for outside helicopter target monitoring and early warning,and has certain practical significance and practical value for national defense construction and security early warning of our country.