The fault diagnosis based on neural network of memory chip in underwater vehicle

The current research of fault diagnosis of underwater vehicle mainly focuses on actuator fault and sensor fault. However, with the increase of the amount of stored information, the failure of memory chip of underwater vehicle gradually becomes one of the important factors affecting the reliability of underwater vehicle. This paper aims to realize the typical fault diagnosis of the memory chip of underwater vehicle with neural network. Based on the idea of equivalence, the flash memory circuit of memory chip is established as the carrier of fault research. The typical faults are analyzed and modeled by the Technology Computer Aided Design (TCAD), and the corresponding simulation circuit models are established by Veriloga language. The power supply current is selected as the basis of fault diagnosis, and the fault is injected into the memory circuit for simulation and the power supply current data are collected to form the fault data set. A method using wavelet decomposition, Fast Fourier Transform (FFT) and Back Propagating (BP) neural network is proposed to process and analyze the power supply current data, and finally output fault classification. The results show that the accuracy of the proposed method can reach 97%, which proves the feasibility of the proposed method.