Research On The Sound Fault-Detection Method Based On Laser Measurement For Loudspeakers

In the production of loudspeaker, the pure tone detection is the first product quality detecting procedure.It adopts sweep frequency signal to excite loudspeaker. At present the inspector uses their hearing to judge the quality of loudspeaker in the electro-acoustic industry. But this method is subjective, because the detection result is influenced by the working environment and experience of the staff.At the same time the inspector’auditory nervous system will be hurt for long-time listening.So looking for a kind of automatic detection method becomes an important subject in electro-acoustic industry.At present, the detection system uses sweep frequency signal to excite the loudspeaker and gets response voice signal through the microphone and A/D converter.Then the digital signal is transfered to computer, computer uses time-frequency analysis method to do feature extraction.This system has a good application in the laboratory and anechoic room. However the collecting signal is influenced by outside environmental noise, it can’t work normally in the manufacturing workshop.So how to realize the sound fault-detection in the big noise environment is the main task in this thesis.This thesis proposes a new sound fault-detection method based on laser sensor. This method can avoid the influence of environmental noise effectively.When the loudspeaker is excited, the laser sensor collects the vibration displacement signal in the center of paper cone.The system obtains a cycle of the vibration displacement signal, the feature is extracted through time-frequency analysis and image segmentation methods.Experiments prove that this method has a good accuracy for the speaker’s fault detection.The main task and research results including:(1) This thesis proposes a speaker paper cone geometry symmetry detection method based on laser sensor and mobile platform.(2) It proposes a method to search the centre of the paper cone.(3) This thesis uses the method of short-time Fourier transform for the analysis of the loudspeakers’ vibration displacement signal, and proposes an image segmentation method for the feature extraction.(4)It completes the design on hardware and software systerm including the modules of the paper cone geometry symmetry detection, centralized positioning and sound fault-detection.