Detection of Low Frequency Pulse Signal in LSI Based on Deep Self Coding Algorithm

Abstract

In order to detect the pulse of LSI with high accuracy and efficiency, a new detection method of low frequency pulse signal of LSI based on deep self coding algorithm is proposed in this paper. Firstly, Radon transform is used to reduce the dimension of the time-frequency diagram of the pulse signal of the integrated circuit, and the short-time Fourier transform of the pulse signal is completed according to the Hamming window. Secondly, the principle of the depth self encoder is analyzed, and the pulse signal after the short time Fourier transform processing is input into the depth self encoder. Finally, according to the distance function, the gradient descent algorithm is used to update the parameters. Through iterative calculation, the detection of low frequency pulse signal of LSI is completed. The experimental results show that, compared with the traditional low-frequency pulse signal detection method, the detection accuracy of this method is greatly improved, up to 99.79%, and can shorten the overall detection time.