Research On Identification Method Of Activated Remainder For Electronic Equipments In Satellites

Electronic equipments in satellites are sealed electronic modules or devices used in aerospace, which accomplish important scientific experimental tasks in satellite systems. Compared with common electronic equipments, their special working environment makes higher reliability demands of them and the internal moveable remainders significantly affect their reliability. Presently, there is no report about identification methods of remainder for systems both at home and abroad. On the basis of identification method of remainder for electronic components, this dissertation extensively studied existence identification method and material identification method of remainder for electronic equipments in satellite based on PIND experiment.In this dissertation, the composition of PIND test signals of electronic equipments in satellites has been analyzed and the signal is divided into remainder signal, component signals and noise and impact signal. The key of existence identification of remainder is how to extract remainder signal and remove the rest from the PIND tests signal. Thus, existence identification method of remainder for electronic equipments in satellite with and without movable components has been studied respectively.As for the electronic equipments in satellites without moveable components, the key elements of wavelet de-noising process,including wavelet function, threshold function, wavelet decomposition levels and threshold algorithm, have been extensively studied, and a wavelet threshold de-noising method that is suitable for removing noise from PIND test signal has been proposed. Moreover, a pulse extraction method was put forward. Experiments indicated that this method can effectively remove impact signals and thus the existence identification of remainders has been accomplished.As for the electronic equipments in satellite with moveable component, the wavelet threshold de-noising method and the pulse extraction method have been adopt to processing the PIND test signal. Furthermore, this dissertation proposed a method of identification of movable components based on short-time Fourier transform to single pulses. Experiments showed that this method can effectively classify remainder signal and component signal and accomplish the existence identification of remainders。At the same time, a confidence evaluation method of existence identification of remainders was put forward as for the electronic equipments in satellite with moveable component. Experiments indicated that the confidence of the identification method was more than 80%. This dissertation used method of time-domain features, frequency-domain feature and BP neural network accomplished identification of material of remainders respectively. Furthermore, method of confidence evaluation has been put forward respectively. Results of experiments showed that confidence level of material identification of remainders based on BP neural network was much higher, which was more than 99%.