Neural network angle measurement method applied to beam waveguide reflector antenna

Abstract

The invention relates to a neural network angle measurement method applied to a beam waveguide reflector antenna, and belongs to the technical field of microwaves. Building a neural network angle measurement system applied to the beam waveguide reflector antenna, wherein the neural network angle measurement system comprises a receiving antenna array and a neural network angle measurement model; acquiring two-dimensional directional diagram data; in a non-uniform radiation field, echo signal amplitude information received by a 16 horn antenna feed source is used as input of an angle measurement system, an azimuth angle and a pitch angle are used as output of the angle measurement system, and neural network training is carried out; and deploying the generated neural network angle measurement model in an echo receiving and processing system. According to the characteristic that different array antenna energy distributions correspond to different echo incident directions, angle measurement is carried out by using a neural network algorithm. According to the angle measurement system trained through the neural network algorithm, an accurate angle measurement result can be obtained.

Description

Technical Field

[0001] The present invention belongs to the field of microwave technology, and is specifically applied to a neural network angle measurement system and method for a beam waveguide reflector antenna. Background Art

[0002] For phased array or digital array radars, after the multi-channel signals are digitized, they directly enter the computer for processing, and two technical approaches can be adopted to achieve target angle measurement. (1) Sum and difference beams are formed in the computer, and then the target angle is measured by using amplitude sum and difference or phase sum and difference monopulse, that is, monopulse angle measurement technology; (2) Angle measurement of the target is carried out by using spatial spectrum estimation. Common methods include MUSIC and ESPRIT algorithms.

[0003] Monopulse angle measurement has the advantages of high measurement accuracy, strong anti-interference ability, and high data rate, and it is widely used in applications such as range precision tracking measurement, target tracking, and radio astronomy. This technology obtains accurate target angle information by comparing the received signals of two or more simultaneous antenna beams. Conventional monopulse angle measurement technology needs to place multiple horn feeds at the focus of the parabolic antenna to form sum and difference beams to achieve angle measurement. However, in order not to affect the microwave signal transmitted by the offset reflector antenna, the 16 horn feeds cannot be placed at the focus of the parabolic antenna, and only the 16 horn feeds can be installed on the outer circle of the beam waveguide. Therefore, conventional monopulse angle measurement technology cannot be used.

[0004] The spatial spectrum estimation technology based on array signal processing has the advantages of high precision, high resolution, and the ability to simultaneously separate and direction-find multi-target co-frequency signals, and has been widely used in radar, communication, exploration, radio astronomy, biomedical engineering and other fields. Spatial spectrum estimation direction-finding uses a multi-element antenna array and utilizes the characteristics that the amplitude and phase of the signals induced by the incoming wave on each antenna element are related to the incoming wave direction to achieve simultaneous direction-finding of multiple signals in space. Since the 16 horn feeds are installed on the outer circle of the beam waveguide, the signals received by the 16 horn feeds are reflected back by the reflector antenna, and the signal phase relationship cannot correspond to the incoming wave direction. Therefore, spatial spectrum estimation cannot be used to measure the target angle. Summary of the Invention

[0005] (I) Object of the Invention

[0006] The object of the present invention is to provide a neural network angle measurement system and method for a beam waveguide reflector antenna, and solve the technical problem that the phase relationship of the signals received by 16 channels is chaotic and cannot correspond to the incoming wave direction.

[0007] (II) Technical Solution

[0008] In order to achieve the above object and solve the above technical problem, the technical solution of the present invention is as follows:

[0009] A neural network angle measurement method applied to a beam waveguide reflector antenna specifically includes the following steps:

[0010] Step 1, build a neural network angle measurement system applied to a beam waveguide reflector antenna

[0011] The neural network angle measurement system includes two parts: a receiving antenna array and a neural network angle measurement model;

[0012] The receiving antenna array contains 16 horn antennas. The 16 horn antennas are deployed on the outer circle of the waveguide outlet of the microwave system. 8 antennas are spaced on each of the upper and lower sides and are symmetrically arranged at an interval of 10° to receive the microwave echo signal;

[0013] The neural network angle measurement model is established using a BP neural network. The input layer has 16 input nodes, the hidden layer has 64 nodes, the output layer has 2 nodes, and the nonlinear function used is the Sigmod function;

[0014] In a non-uniform radiation field, the neural network angle measurement system uses the amplitude information of the echo signal received by the 16-horn antenna feed as the input and the azimuth angle and elevation angle as the output;

[0015] The neural angle measurement system establishes a non-linear relationship between the 16-channel signal amplitude information and the target azimuth angle and elevation angle through a neural network algorithm to obtain a neural network model for subsequent prediction of the angle information of the microwave echo signal;

[0016] Step 2, two-dimensional pattern data acquisition

[0017] Install a signal source on a high tower that meets the far-field conditions in the experimental field to radiate to the array antenna installed in the microwave system. The antenna servo mechanism of the microwave system controls the antenna to align with the radiation antenna of the signal source on the tower, finds the angle with the maximum received signal energy, and takes this angle as the center to scan in two dimensions of azimuth and elevation to obtain two-dimensional pattern data of azimuth and elevation;

[0018] Step 3, neural network training

[0019] Use the Levenberg-Marquardt algorithm to train the neural network, and use the 16-channel signal amplitude information collected in the experiment to train the BP neural network to generate a neural network angle measurement model;

[0020] Step 4, deploy the generated neural network angle measurement model in the echo receiving and processing system

[0021] The neural network angle measurement module is used for processing the echo received signal. The trained neural network model is deployed in the echo receiving and processing system to collect the signal received by the antenna, perform data preprocessing, and implement the angle measurement of the neural network algorithm.

[0022] Furthermore, when using the neural network algorithm to train the data, it is necessary to normalize the data to reduce the angle measurement time.

[0023] Furthermore, in step 3, the neural network training does not require phase calibration of the channels.

[0024] (III) Effective benefits

[0025] In view of the situation where monopulse angle measurement cannot be used, the present invention uses a neural network to train the amplitude information of the measured 16-channel signal to form a neural network angle measurement model. Placing this model in the angle measurement system can achieve relatively accurate angle measurement results. Specifically, it is reflected in the following aspects:

[0026] (1) Through the neural network algorithm, the present invention trains and learns the amplitude information of the microwave echo signal, and can obtain the angle information of the target from the echo signal in a non-uniform radiation field, solving the problem that it is difficult to obtain the angle information of the target in a non-uniform radiation field by conventional methods.

[0027] (2) The neural network angle measurement method of the present invention applied to the beam waveguide reflector antenna uses the angle error information of the microwave echo signal to achieve target tracking.

[0028] (3) The neural network angle measurement system of the present invention needs to normalize the amplitude information of the 16-channel signal. The normalized data enables the trained neural network angle measurement system to meet the requirements for use under different signal intensities, and the angle measurement time is greatly reduced. Description of the drawings

[0029] Figure 1 It is the simulation diagram of the receiving antenna array of the present invention;

[0030] 1(a) Radiation pattern of channel 1;

[0031] 1(b) Radiation pattern of channel 2;

[0032] 1(c) Radiation pattern of channel 3;

[0033] 1(d) Radiation pattern of channel 4;

[0034] 1(e) Radiation pattern of channel 5;

[0035] 1(f) Radiation pattern of channel 6;

[0036] 1(g) Radiation pattern of channel 7;

[0037] 1(h) Reception pattern of channel 8;

[0038] 1(i) Reception pattern of channel 9;

[0039] 1(j) Reception pattern of channel 10;

[0040] 1(k) Reception pattern of channel 11;

[0041] 1(l) Reception pattern of channel 12;

[0042] 1(m) Reception pattern of channel 13;

[0043] 1(n) Reception pattern of channel 14;

[0044] 1(o) Reception pattern of channel 15;

[0045] 1(p) Reception pattern of channel 16;

[0046] Figure 2 Schematic diagram of the receiving antenna array structure of the present invention;

[0047] Figure 3 Experimental deployment diagram of the present invention;

[0048] Figure 4 Schematic diagram of the angle measurement model of the neural network of the present invention;

[0049] Figure 5 Schematic diagram of the probability distribution of the angle measurement error of the present invention;

[0050] 5(a) Angle measurement error of pitch angle;

[0051] 5(b) Angle measurement error of azimuth angle. Detailed implementation manners

[0052] In order to understand the above objects, features and advantages of the present invention more clearly, the present invention will be further described in detail below with reference to the accompanying drawings and specific implementation manners. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments may be combined with each other.

[0053] The present invention is an angle measurement method proposed to utilize the angle error information of microwave echo signals to achieve target tracking.

[0054] The scientific basis or design principle of the present invention is as follows: Simulate the reception patterns of 16 horn antennas, as Figure 1As shown, it can be seen that there are angular differences in the reception pattern of each horn antenna. Therefore, the echo signal energies received by horn antennas at different positions are different, and the energy received by each antenna varies with the incident wave angle. The core idea of the present invention is to use the neural network algorithm for angle measurement based on the characteristic of different energy distributions of array antennas corresponding to different echo incident directions.

[0055] The present invention provides a neural network angle measurement method applied to a beam waveguide reflector antenna, which specifically includes the following steps:

[0056] Step 1: Build a neural network angle measurement system applied to a beam waveguide reflector antenna

[0057] The neural network angle measurement system includes two parts: a receiving antenna array and a neural network angle measurement model;

[0058] For the angle measurement method of the beam waveguide reflector antenna based on neural network provided by the present invention, an antenna array structure needs to be installed beside the beam waveguide of the microwave system. The tooling structure of the antenna array combines the design structure of the microwave system. The tooling structure uses 16-channel horn antennas, and each horn is evenly arranged up and down on the outer circle of the waveguide outlet at an interval of about 10°. There are eight horns on each of the upper and lower sides, as Figure 1 shown.

[0059] The receiving antenna array includes 16 horn antennas. The 16 horn antennas are deployed on the outer circle of the waveguide outlet of the microwave system. Eight antennas are symmetrically arranged at intervals of 10° on each of the upper and lower sides to receive microwave echo signals, as Figure 2 shown;

[0060] The present invention uses a BP neural network to establish a neural network angle measurement model. The BP neural network is a multi-layer feedforward neural network algorithm. Figure 4 This is the network structure diagram of the BP neural network of the present invention. Among them, the input layer has 16 input nodes, the hidden layer has 64 nodes, and the output layer has 2 nodes. The non-linear function used is the Sigmod function. The network training method uses the Levenberg-Marquardt algorithm. When using the neural network algorithm to train data, the data needs to be normalized. The normalized data can greatly reduce the neural network training amount, and can also make the generated neural network angle measurement model meet the requirements for use under different signal intensities, and the angle measurement time is greatly reduced. Since the data for neural network training is the amplitude information of the signal, channel phase calibration is not required.

[0061] The angle measurement system built by the present invention can use the amplitude information of the echo signals received by a 16-horn antenna feed as the input and the azimuth angle and elevation angle as the output in a non-uniform radiation field. The angle measurement system establishes a non-linear relationship between the 16-channel signal amplitude information and the target azimuth angle and elevation angle through a neural network algorithm to obtain a neural network model for subsequent prediction of the angle information of microwave echo signals.

[0062] Step 2: Two-dimensional pattern data acquisition

[0063] In the experimental field, the azimuth and elevation pattern data are collected. The collected data are used for the training of the neural network in Step 4. The experimental deployment in the experimental field is as Figure 3 shown. On a high tower that meets the far-field conditions in the experimental field, a signal source is erected to radiate to the array antenna installed in the microwave system. The antenna servo mechanism of the microwave system controls the antenna to align with the radiation antenna of the signal source on the tower, finds the angle with the maximum received signal energy, and takes this angle as the center to scan in both the azimuth and elevation dimensions to obtain the two-dimensional azimuth and elevation pattern data.

[0064] Step 3: Neural network training

[0065] The Levenberg-Marquardt algorithm is used for the network training method. The 16-channel signal amplitude information collected in the experiment is used to train the BP neural network to generate a neural network angle measurement model.

[0066] Step 4: Deploy the generated neural network angle measurement model in the echo receiving and processing system

[0067] The neural network angle measurement module is used for the processing of the echo receiving signals. The trained neural network model is deployed in the echo receiving and processing system to collect the signals received by the antenna, perform data preprocessing, and implement the neural network algorithm for angle measurement.

[0068] Embodiment 1

[0069] The present invention builds an angle measurement system based on a neural network, uses the amplitude information of 16-channel signals as the input of the neural network, uses the azimuth angle and elevation angle as the network output, trains the neural network to generate a neural network angle measurement model, and places the model in the angle measurement system to implement the neural network angle measurement function. Figure 5 For the probability distribution of the angle measurement error of the neural network, it can be seen that the angle measurement effect of the neural network is relatively accurate, as Figure 5 (a), 5(b) shown.

[0070] The above content is a further detailed description of the present invention in combination with specific embodiments, and it cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the concept of the present invention, several simple deductions or substitutions can still be made, and all should be regarded as belonging to the protection scope of the present invention.

Claims

1. A neural network angle measurement method applied to a beam waveguide reflector antenna, characterized in that: The specific steps include: Step 1: Build a neural network angle measurement system for beam waveguide reflector antennas The neural network angle measurement system includes two parts: a receiving antenna array and a neural network angle measurement model; The receiving antenna array includes 16 horn antennas, which are deployed on the outer circle of the microwave system waveguide outlet, and 8 antennas are arranged symmetrically at intervals of 10 degrees on the upper and lower sides to receive microwave echo signals; The neural network angle measurement model is established by using BP neural network, the input layer has 16 input nodes, the hidden layer has 64 nodes, the output layer has 2 nodes, and the nonlinear function used is Sigmod function; The neural network angle measurement system uses the amplitude information of the echo signal received by the 16-horn antenna feed source as input in a non-uniform radiation field, and uses the azimuth and elevation angles as output; The neural angle measurement system establishes a nonlinear relationship between the 16-channel signal amplitude information and the target azimuth and elevation angle through a neural network algorithm to obtain a neural network model for subsequent prediction of the angle information of the microwave echo signal; Step 2: 2D pattern data collection A signal source is set up on a high tower that meets the far-field conditions in the experimental field to radiate to the array antenna installed in the microwave system. The antenna servo mechanism of the microwave system controls the antenna to align with the radiating antenna of the signal source on the tower, finds the angle of maximum energy of the received signal, and scans in azimuth and elevation with this angle as the center to obtain azimuth and elevation two-dimensional directional pattern data; Step 3: Neural network training The Levenberg-Marquardt algorithm is used to train the neural network, and the 16-channel signal amplitude information collected in the experiment is used to train the BP neural network to generate a neural network angle measurement model. Step 4: Deploy the generated neural network angle measurement model in the echo receiving and processing system The neural network angle measurement module is used to process the echo reception signal. The trained neural network model is deployed in the echo reception and processing system to collect the signal received by the antenna, preprocess the data, and implement the neural network algorithm angle measurement.

2. The neural network angle measurement method for a beam waveguide reflector antenna according to claim 1, characterized in that: When using a neural network algorithm to train data, the data needs to be normalized to reduce the angle measurement time.

3. The neural network angle measurement method applied to the beam waveguide reflector antenna according to claim 1, wherein: In step 3, the neural network training does not require phase calibration of the channel.

Images