Signal Conditioning Optimization And Component Signal Recognition For Redundant Particles Detection Of Sealing Electronic Components

Sealed electronic components are used as an important part of high-reliability equipment systems,and the redundant items often pose a threat to their reliability.The existing redundant detection equipment generally adopts the collision detection mechanism of fine particles,and the advanced one is the DZJC-III type PIND redundant detection system,but it has certain problems in the application process.In this paper,the signal conditioning circuit of DNJC-III type PIND loose particle detection system is studied and improved for the above problems.At the same time,the pulse extraction method in the detection metho d is optimized in the classification algorithm to improve the reliability of the judgment result in the use of the device.For the signal acquisition and conditioning process,the specification of the sensor crystal is standardized and the FPCB cable is optimized to improve its anti-interference ability.The signal amplification scheme was adjusted,the filtering performance of the filter circuit was optimized,the signal-to-noise ratio was improved,and the reliability of the conditioning circuit was ensur ed.For pulse extraction,the corresponding detection speed and precision are improved by optimizing the original analysis thread;By designing a two-level threshold method based on the two characteristics of the short-term average energy and the short-term average zero-crossing rate,the completeness of the pulse signal is improved.The misclassification of the signal classification result is due to the inconspicuous difference between the single feature of the redundant signal and the component signal.It is difficult to classify accurately based on a single feature.Therefore,the signal pulse is analyzed in the time-frequency domain,and then a variety of features in the time-frequency domain are extracted.In the frequency domain,short-time Fourier transform is performed on the signal to obtain its frequency domain distribution,and a frequency domain characteristic capable of reflecting the difference between the two signals in the frequency domain is selected to describe the signal.In the time domain,the time-domain characteristics of the discrete sequence of the signal are described in terms of time,distribution,and energy,and the differences in the timing of the expression of the relevant features are extracted.Finally,the characteristics of signal differentiation are selected as the basis for signal classification.According to the characteristics of the unwanted signal and the component signal,a self-learning signal pedigree clustering algorithm with the signal as the main body and a single pulse categorization algorithm with the signal single pulse as the main body were designed.For the first algorithm,experiments show that the correct recognition rate of the algorithm is relatively high for pure component signals and pure excess signals,which can reach more than 80%.However,the correct recognition rate of the mixed signal of the component signal and the unwanted signal is low.Therefore,a component signal identification method for a single pulse signal is further designed.For a certain type of relay testing,the accuracy rate for a single pulse classification can reach about 90%,and from the results,the corresponding confidence level is also low for wrong judgments.