We analyzed the reasons why ZnTe generates terahertz echo. Based on the Gaussian beam, the terahertz signal with echo is produced by simulation. In this paper, the deconvolution algorithm is further improved. The linear assumption that sample absorption is proportional to frequency is abandoned. We studied the method of echo based on Echo State Network. Set 200 neural units in each layer of the network. We specify 250 rounds of training. To prevent the gradient from exploding, set the gradient threshold to 1. Specify an initial learning rate of 0.005. After 125 rounds of training, multiply it by a factor of 0.2 to reduce the learning rate. This method largely removes echo. But it has not been completely removed, and there is still less than 5% of the echo signal remaining. We consider the reason may be that the number of neurons, the number of iterations, weight setting and other parameters have yet to be optimized. Although this method does not completely remove the echo, but provides a new method for removing the echo. By optimizing parameters such as the number of neurons, the number of iterations, weight setting and increasing the amount of training data, this method is expected to become a new method for echo removal.
|