Array gain and phase error calibration method based on neural network

Abstract

The invention discloses an array gain and phase error calibration method based on a neural network. The method comprises two steps of using the neural network to carry out offline training and using a grouping calibration strategy to carry out online calibration; the neural network is composed of a front signal-to-noise ratio classification network and two rear error calibration networks, and the error calibration networks are composed of a gain calibration network and a phase calibration network; according to the grouping calibration strategy, an antenna array is divided into a plurality of sub-arrays, the number and size of array elements of the sub-arrays are matched with those of array elements of the trained neural network, and gain and phase errors are obtained by applying the neural network for many times. According to the method, the balance between the calibration precision and the calibration complexity can be well achieved, a grouping calibration algorithm is provided based on the data characteristics of the neural network input vector, the neural network can be applied to antenna arrays with different array element numbers and different shapes without being repeatedly trained, and the calibration accuracy is greatly improved. And the practical application cost of the algorithm is reduced.

Description

technical field [0001] The invention relates to the technical field of data filtering, in particular to a method for calibrating array gain and phase error based on a neural network. Background technique [0002] When the array works for a long time, the gain and phase of the sensors in the array will be inconsistent, which will lead to gain and phase errors in the array. Due to the existence of gain and phase errors, the array flow pattern matrix of the array will be known non-deterministically, which makes it impossible to apply traditional DOA estimation algorithms such as MUSIC. [0003] Literature 1 (Array gain / phase calibration techniques for adaptive beamforming and direction finding) proposes an active calibration algorithm for the existence of gain and phase errors in the array, which requires a source with a known angle of arrival in the far-field space, and then passes The array obtains the signal reception matrix and calculates the covariance matrix, thereby con...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the problem of gain and phase error in the array, the present invention proposes an array gain and phase error calibration method based on neural network, which can achieve a good balance between calibration accuracy and calibration complexity

Images