Research On The Fault-Detection Method Based On The Dsp For The Lou Dspeakers

In the producing loudspeaker, detection of loudspeaker sound distortion is the very important to ensure quality of production. At present, the detection of loudspeaker sound distortion at home and abroad mainly adopts the sweep frequency signal to excite loudspeaker, relying on inspector’s experience to judge the loudspeaker is faulted or not. This inspection method has subjectivity, lack objective standard, and would be affected by working environment, people’s working status, can also harm the human auditory and nervous system. Therefore, detection of loudspeaker sound distortion is an important research topic in the electro-acoustic industry.With the development of science and technology, it’s an inevitable trend to turn the loudspeaker fault detection system into a small and intelligent embedded system. This thesis will apply DSP digital processing technology into the loudspeaker sound distortion detection; it has high processing speed and good real-time performance, etc. However, the response signal generated by the sweep signal to excite loudspeaker is very complex, fault feature extraction by traditional Fourier transform is restricted, so this thesis use time-frequency analysis method to do feature extraction.The research findings include:(1) It researched the problem of window function and bandwidth selection of the short time Fourier transform, and given the results that when selecting the window function, the main lobe should be as narrow as possible to improve the frequency resolution, next to the flap as small as possible to reduce the leakage of the conclusions.(2) It demonstrated hardware solutions feasibility of the loudspeaker fault detection system based on DSP, proposed the concept of flexible sub-band frequency sweep, and accordingly designed the loudspeaker exciting system based on AD9850. In order to take full advantage of DSP digital signal processing capabilities, using TMS320VC540and AVR128formed dual-core system, DSP is responsible for processing the response signal of the loudspeaker collected, and the microcontroller is responsible for interface control and display functions.(3) It studied the fault diagnosis method based on high frequency characteristic curve and energy methods. According to the actual problems encountered during the test,"intermediate state" concept was proposed. Through analysis of the loudspeaker time-frequency diagrams characteristics, it is concluded that fault features are mainly concentrated in higher-order frequency area, and the signal energy mainly concentrated in low order frequency area. (4) It compared two kinds of fault diagnosis methods systematically, and got the conclusion that in the environment of the less demanding of production process and relatively harsh on-site environment, it’s more appropriate to use higher-order frequency overall for energy coefficient in the condition of high demanding of production process and relatively friendly on-site environmental, adopting energy characteristic curve method can greatly improve recognition accuracy.(5) It solved the electromagnetic interference problem of the loudspeaker abnormal noise fault detection inflicted in production workshop site, solved the problem of signal source urge loudspeaker and synchronous collect response signal.