CN-122017830-A - Vertical measurement ionization diagram O/X wave separation method based on polarization matched filtering

CN122017830ACN 122017830 ACN122017830 ACN 122017830ACN-122017830-A

Abstract

The invention discloses a vertical measurement ionization map O/X wave separation method based on polarization matched filtering, which comprises the steps of obtaining a double-channel complex signal, carrying out incoherent accumulation to obtain echo intensity, generating a mask by using an OS-CFAR detector and morphological processing, calculating a phase difference based on the mask, obtaining O wave and X wave characteristic phases by fitting a double Gaussian mixture model, estimating a complex polarization ratio, constructing a weight vector of a polarization matched filter, multiplying the double-channel signal, separating to obtain an O/X wave complex signal matrix, and generating a separated ionization map by amplitude calculation and detection. The method does not depend on circular polarization assumption, can directly estimate actual polarization parameters, realizes high-precision separation through polarization matched filtering, and improves robustness and accuracy.

Inventors

LIU WEN
ZHOU BING
DENG ZHONGXIN
NING YUN
XIA JINYU

Assignees

湘潭大学

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (10)

1. The vertical measurement ionization diagram O/X wave separation method based on polarization matched filtering is characterized by comprising the following steps of: S1, acquiring a north-south channel complex signal matrix and an east-west channel complex signal matrix of an orthogonal two-channel receiving system in ionosphere vertical detection, and performing incoherent accumulation to obtain an echo intensity matrix; Step S2, applying an ordered statistics constant false alarm rate detector to the echo intensity matrix, identifying potential effective signal points and generating a first mask matrix; Step S3, performing morphological processing on the first mask matrix, removing isolated noise points and micro-communication areas, and generating a second mask matrix; S4, calculating a phase difference matrix of the north-south channel and the east-west channel based on the second mask matrix, and performing double Gaussian mixture model fitting on a phase difference statistical histogram to obtain characteristic phases of the normal wave and the abnormal wave; S5, respectively estimating complex polarization ratios of the normal wave and the abnormal wave according to the characteristic phase, wherein the complex polarization ratios comprise amplitude ratio and phase difference information; S6, respectively constructing corresponding polarization matched filter weight vectors according to the estimated complex polarization ratio of the normal wave and the abnormal wave, multiplying the weight vectors by a dual-channel signal vector matrix, and separating to obtain complex signal matrixes of the normal wave and the abnormal wave; and S7, performing amplitude calculation and signal detection on the separated complex signal matrix of the normal wave and the abnormal wave to generate an ionization diagram for separating the normal wave and the abnormal wave.
2. The separation method according to claim 1, wherein in the step S2, for each detection unit, the ordered statistic constant false alarm rate detector is provided with a protection unit and a reference unit in front and behind the detection unit, echo intensity samples in the reference unit are arranged in ascending order, and a kth element is selected as an estimated value of the echo intensity of the background noise, wherein the value of k is between 50% and 75% of the total number of the reference units.
3. The method according to claim 1, wherein in the step S3, the morphological processing includes performing connected domain analysis on the binary mask image and filtering out the fragmented regions based on a preset area threshold.
4. The separation method according to claim 1, wherein in the step S4, the expression for performing nonlinear least squares fitting on the phase difference histogram by using the double Gaussian mixture model is: ; Wherein, the In order for the phase difference to be a phase difference, A function is fitted to the probability density of the phase difference, The mean value, standard deviation and amplitude of the corresponding O wave components are respectively, The mean value, standard deviation and amplitude of the corresponding X-wave components are respectively; wherein two peaks obtained by fitting And Characteristic phases corresponding to the normal wave and the abnormal wave, respectively.
5. The separation method according to claim 1, wherein in the step S5, the complex polarization ratio is determined The ratio of the north-south channel complex signal to the east-west channel complex signal is defined as: ; Wherein, the And The amplitudes of the north-south channel complex signal and the east-west channel complex signal are respectively, Is the phase difference of the two complex signals, In imaginary units.
6. The separation method according to claim 5, wherein in the step S6, the polarization ratio is determined based on the complex number The normalized jones vector of the construct is: ; polarization matched filter weight vector Taken as the normalized Jones vector Conjugate matching.
7. The separation method according to claim 6, wherein in the step S6, the obtained constant wave complex signal matrix is separated And an extraordinary wave complex signal matrix Is respectively: ; ; Wherein, the At the frequency of O wave after separation Deficiency of high qi The echo signal at the point is a signal, For X-wave after separation at frequency Deficiency of high qi The echo signal at the point is a signal, And Polarization matched filter weight vectors for normal and abnormal waves, respectively, the superscript H denotes the conjugate transpose, Vectors composed of corresponding elements for the north-south lanes and the east-west lanes.
8. The separation method according to claim 7, wherein in the step S6, since the polarization states of the normal wave and the non-normal wave are approximately orthogonal, the normal wave filter weight vector and the non-normal wave filter weight vector are constructed to be approximately orthogonal, so that the reception target wave signal is maximized and the orthogonal mode wave is suppressed.
9. The separation method according to claim 7, wherein in the step S6, the weight vector of the polarization matched filter is determined And After being multiplied by the corresponding elements of the two-channel signal vector matrix respectively, the signal energy of the orthogonal polarization mode in the output signal is restrained, and the energy of the restrained residual signal is lower than one ten thousandth of the energy of the original signal.
10. The method of separation according to claim 1, wherein in step S7, the ionospheric echographic traces of the normal wave and the abnormal wave after separation are marked in the same frequency-virtual high coordinate system by different colors to generate a final normal wave and abnormal wave separation ionization map.

Description

Vertical measurement ionization diagram O/X wave separation method based on polarization matched filtering Technical Field The invention belongs to the technical field of vertical detection of a short wave ionized layer, and particularly relates to a vertical detection ionization diagram O/X wave separation method based on polarization matched filtering. Background Ionospheric vertical probing has so far remained one of the most basic and reliable techniques in ionospheric scientific research and in radio wave propagation and application research. The working principle is that high-frequency pulse signals are vertically and upwardly transmitted, reflected echoes from each layer of the ionized layer are received, the virtual height of the ionized layer is obtained by measuring echo time delay, then a frequency-virtual height relation curve, namely frequency Gao Tu (or vertical measurement ionization diagram) is obtained, and key parameters such as critical frequency of each layer and the like can be extracted based on the frequency height diagram, and the vertical distribution profile of the ionized layer electron concentration can be inverted. Ionosphere vertical detection has irreplaceable practical value in the fields of scientific research, national defense, communication, navigation and the like. Due to the presence of the geomagnetic field, the normally incident radio wave undergoes magnetic ion splitting as it propagates in the ionosphere, forming two elliptically polarized components, the ordinary wave (O-wave) and the extraordinary wave (X-wave). These two waves have different refractive properties and propagation paths, resulting in them appearing as two separate echographic traces on the frequency hypergraph. The correct separation of the O-wave and X-wave is crucial for interpreting the vertical ionization diagrams, directly affecting the accuracy of ionospheric parameter measurements. The realization of high-precision separation of the two characteristic wave traces is a necessary condition for accurately extracting the characteristic parameters of the ionosphere and realizing inversion of the vertical distribution profile of high-precision electron density, and is also an important link for improving the application value of ionosphere detection data. However, in actual detection, the detection device is affected by a plurality of factors such as environmental radio noise, short wave radio station interference, ionosphere time-varying, chromatic dispersion, dissipation, multimode multipath propagation, technical limitations of the detection device (such as antenna orthogonality, different channel amplitude-phase consistency, antenna placement mode) and the like, and the accurate separation of O/X waves is very difficult. The existing O/X wave separation method mainly extends around three directions of digital image processing, deep learning and polarization information processing. The method based on digital image processing realizes separation from the trace form of the ionization image by means of morphology, graph theory and the like, but the performance of the method is seriously dependent on the integrity of the image, and the robustness is insufficient under the scenes of complex trace overlapping and the like caused by noise, interference or ionosphere disturbance. The deep learning-based method realizes end-to-end separation by training the neural network, and has good performance on normative data, however, under the complex physical conditions of ionosphere time-varying, chromatic dispersion and the like, the generalization capability and stability of the method still have limitations, and the method cannot physically express the polarization characteristics of the OX wave. The O/X wave separation method based on polarization information processing directly utilizes the relation between the amplitude and the phase of two paths of orthogonal polarization receiving signals, and has definite physical significance of ionosphere wave propagation. The conventional method generally assumes that the O-wave and the X-wave exhibit ideal left-hand and right-hand circular polarizations, respectively, and the separation is achieved by applying a theoretical ±90° fixed phase shift between the two signals. However, ionosphere wave propagation cannot be ideal circular polarization due to the influence of various factors, in addition, the orthogonality, amplitude consistency and antenna arrangement mode (strict arrangement mode is in the directions of geomagnetism, north-south and east-west) of an antenna system, and the fact that the polarization state of an actual received signal is deviated from theoretical circular polarization due to the fact that environment and equipment generate phase noise and the like often cause the fact that the polarization state of the actual received signal deviates from theoretical circular polarization, elliptical polarization characteristics are shown, and the phase