The Research On Detection Method Of Abnormal Response Signal To YDT318-9AU Loudspeaker

The abnormal sound detection of loudspeaker is very important for factory to ensure product quality.In the modern production line of the loudspeaker,the loudspeaker is usually detected by ears.What the sound was listened will decide whether the loudspeaker is fault or not.With the development of the loudspeaker industry and the rapid increase yield of the loudspeaker,the method of manual inspection gradually shows its more and more problems:First,as the main body,hearing sense has a lot of subjectivity,and it is very hard to maintain a stable state in long time,so it can not guarantee the accuracy of loudspeaker detection.Second,loudspeaker companies spend a lot cost on trainning a qualified employee and enterprises are faced the risk that is excellent employees to resign at any time,and lack of excellet employee who can distinguish between good loudspeaker and fault loudspeaker by listenning.Third,the long time and high intensity manual detection will have serious impact on the body and mind.Therefore,it is imperative to develop a set of loudspeaker system to meet the needs of the enterprises.According to the needs of the enterprises,this paper only analyzes the response signal of YDT318-9AU loudspeaker.Then the harmonic distortion of the response signal was analyzed and processed by short time Fourier transforms.The time-frequency map can be divided into different characteristic region,and the standard average energy curve of the acceptable range of the characteristic region is used as a standard to judge the abnormal sound defection of the loudspeaker.The detection process of the detection system:The sweep signal generator generate a signal as loudspeaker's input signal,then loudspeaker will generate a response signal and the sound sensor will receive the response signal and pass it to the data acquisition card,which transform the analogy signal to the digital signal.The digital signal was sent to the computer by acquisition channel.The computer will apply short time Fourier transforms to process the digital signal to get the time-frequency distribution map.Last,the time-frequency map is divided into different parts and calculated to the total energy of every part.According to the established model which contains the features of plenty of good loudspeakers,the appropriate threshold will be set by comparing qualified speakers and fault,and the threshold decides wheather the loudspeaker is good or not.The main research works include:1.Analysis and research on the principle of the abnormal sound fault in YDT318-9AU;2.According to the energy distribution feature of the YDT318-9AU loudspeaker,a new method that based on average energy curve is proposed to detect abnormal noise of loudspeaker;3.Save a periodic signal from response signal of time domain;4.Complete the integrated test system containing software and hardware modules.Experimental results show that the detection method that is based on average energy curve can identify the abnormal sound of YDT318-9AU loudspeaker.