Error Calibration Method of Angle-Dependent Complex Array Based on Deep Neural Network

Abstract

The invention discloses a method for calibrating angle-dependent complex array errors based on a deep neural network. Aiming at the difficulty in dealing with the calibration problem of angle-dependent complex array errors in traditional signal processing methods, the invention is based on darkroom measurement in order to deal with off-grid targets. For grid point data, the local array flow pattern interpolation method adopted is more suitable for angle-dependent array errors than the global array flow pattern interpolation method; moreover, the input feature of the deep neural network is the phase under complex conditions, not the actual The phase of the phase can avoid the problem of phase jumps at the ±π edge; finally, in order to adapt the neural network to noisy signals, we only need to generate training data on the data of a single SNR, instead of multiple SNR Generate training data on the network, thereby reducing the amount of training data and shortening the training time. Compared with the traditional signal processing method, the invention has the characteristics of smaller residual array error after calibration and better calibration performance.

Description

technical field [0001] The invention belongs to the field of array signal processing, and in particular relates to the calibration of receiver sensor array errors such as communication, sonar, radar, etc., and specifically relates to a calibration method suitable for angle-dependent complex array errors based on a deep neural network. Background technique [0002] Array signal processing is a technique widely used in civilian and military fields such as communications, sonar, radar, seismology, astronomy, and more. Its working method is to use multiple sensors placed in a specific form to detect and estimate the received signal at the same time. In an ideal situation, the response of each sensor is the same and independent, and the position of the sensor is known precisely, so the array steering vector has an exact analytical expression. At this time, the relevant array signal processing algorithm can be directly used to process the signal, such as direction finding and bea...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the difficulty of calibrating angle-dependent complex array errors in existing traditional signal processing methods, the present invention proposes a method for calibrating array errors based on deep neural networks. The specific technical solutions are as follows:

Images