Direction-of-arrival estimation method based on fitting model under array model error

Abstract

The invention discloses a direction-of-arrival estimation method based on a fitting model under an array model error. The method comprises the steps of S1 building a steering vector model containing the array error; S2 generating a training set sample and a test set sample by setting different proportionality coefficients; S3 performing self-correlation operation on the training set sample and the test set sample to obtain a feature data vector as input data of a neural network, and generating a label corresponding to an incoming angle; S4 building a fitting neural network model, and initializing parameters; S5 inputting the feature data vector and the label corresponding to the incoming angle into the fitting neural network model for training; and S6 inputting the test set sample into the saved fitting neural network model for testing to obtain an estimation angle. According to the method, the neural network of the fitting model is realized by adopting a full connection layer, and the method can be better suitable for carrying out rapid and high-precision direction finding on interference signals received by the array in actual engineering.

Description

technical field [0001] The invention belongs to the technical field of direction of arrival, and in particular relates to a direction of arrival estimation method based on a fitting model under an array model error. Background technique [0002] A basic problem in array signal processing is the direction of arrival of space signals, and it is also one of the important tasks in the fields of radar, sonar and medicine. Traditional DOA estimation algorithms are based on ideal mathematical models. However, in practical engineering applications, there are various model errors in array received signals that affect the received data and thus affect the estimation results. The existing direction of arrival estimation algorithm based on machine learning needs too many training parameters in the training network model, the network is more complicated, and the training time is too long. Contents of the invention [0003] The purpose of the present invention is to solve the problem t...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to solve the problem that the existing direction of arrival estimation algorithm needs too many training parameters, the network is relatively complex and the training time is too long, and a method of direction of arrival estimation based on the fitting model under the error of the array model is proposed

Images