Fault diagnosis method of rare earth extraction production line based on wavelet packet and alexnet transfer learning

In rare earth production, improving the fault detection of extraction equipment has a key impact on improving extraction efficiency and product quality. The traditional method is based on the observation of the liquid level, the efficiency and accuracy can't be guaranteed. In this paper, a fault diagnosis and recognition method for rare earth extraction production line based on wavelet packet and alexnet transfer learning is proposed. Taking six fault states of rare earth extraction production line as the research object, the fault signal is extracted by wavelet packet decomposition, and the corresponding target data set of time-frequency diagram is generated. Then, the pre trained alexnet model is trained and fine tuned on the generated target data set, and finally applied to fault diagnosis of rare earth extraction production line. The results show that the proposed method is more accurate than the traditional convolution neural network, which verifies the effectiveness of the method.