Low-orbit Doppler positioning method and device based on dynamic empowerment of epoch by intelligent agent

Abstract

The invention discloses a low-orbit Doppler positioning method and device for dynamically weighting epochs based on an intelligent agent, and belongs to the technical field of satellite navigation positioning. According to the method, an initial position and an initial state covariance matrix of a receiver are determined based on an initial accumulation low-orbit non-cooperative signal of the receiver, an initial point location of a current epoch signal is determined when a new low-orbit non-cooperative signal is received so as to realize time frame synchronization, and a state characteristic of the current epoch signal is determined at least based on the initial point location. The state features at least comprise satellite features, epoch state features, signal state features and positioning state features, determining a weight of a current epoch signal based on the state features using a pre-training agent, and determining a real-time position of the receiver based on the current epoch signal and the weight of the current epoch signal using a recursive algorithm, according to the invention, adaptive empowerment based on the multi-dimensional features of the epoch signal is realized, the accuracy of the determined epoch signal weight is improved, and the positioning precision is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of satellite navigation and positioning, and particularly relates to a low-orbit Doppler positioning method and device based on epoch dynamic weighting by an intelligent agent. BACKGROUND

[0002] The Global Navigation Satellite System (GNSS) is the most widely used positioning, navigation and timing (PNT) solution at present, but its high orbital height (about 6300 kilometers) leads to significant signal path loss (-157 dB to -159 dB), and its performance decreases rapidly or even fails in shielded environments such as urban canyons, dense vegetation, tunnels and indoor environments. In recent years, low earth orbit (LEO) satellite signal of opportunity (SOP) has become an important supplement or alternative to GNSS due to its low orbital height (only 1 / 20 to 1 / 40 of GNSS), high signal power (more than 20 dB stronger than GNSS) and global coverage.

[0003] However, when using LEO SOP for real-time positioning, it is usually necessary to accumulate multiple epoch data. Traditional methods such as batch least squares have the problems of epoch redundancy and high computational complexity, and real-time positioning methods such as recursive least squares (RLS) do not fully consider the dynamic changes of epoch quality. The existing technology for processing the reliability of epochs is mostly fixed weight or simple heuristic rules, which only optimizes a single or a few features (such as azimuth, elevation or GDOP), and cannot comprehensively evaluate multi-dimensional epoch features (such as signal-to-noise ratio, ephemeris age, satellite ID, etc.), resulting in limited positioning accuracy. For example, the OECW algorithm only corrects the orbit error by azimuth and elevation, the LEO-NNPON architecture only optimizes the orbit extrapolation accuracy, and the satellite selection strategy based on GDOP only realizes 0-1 weighting, none of which realizes dynamic and flexible weighting of epochs.

[0004] Therefore, there is an urgent need for a positioning method that can dynamically evaluate multi-dimensional epoch features and adaptively weight them to improve the positioning accuracy and real-time performance of LEO SOP in complex environments. SUMMARY

[0005] To solve the above technical problems, the application provides a low-orbit Doppler positioning method and device based on epoch dynamic weighting of an intelligent agent, which processes the Iridium satellite signals received on the ground, extracts multi-dimensional feature parameters from the satellite epoch, uses a pre-trained neural network-based intelligent agent, considers the complex, multi-level and dynamically changing satellite epoch features by the intelligent agent, performs credibility assessment on the epoch, and converts the epoch credibility into adaptive weights for integration into RLS for iterative positioning.

[0006] To achieve the above object, the technical scheme adopted by the application is as follows:

[0007] In a first aspect, the application provides a low-orbit Doppler positioning method based on epoch dynamic weighting of an intelligent agent, which comprises:

[0008] Step 1: obtaining an initial accumulated low-orbit non-cooperative signal of a receiver, and determining an initial position and an initial state covariance matrix of the receiver based on the initial accumulated low-orbit non-cooperative signal;

[0009] Step 2: in response to receiving a low-orbit non-cooperative signal of a current epoch, determining a starting point of the current epoch signal to achieve time frame synchronization;

[0010] Step 3: determining a state feature of the current epoch signal; the state feature at least includes a satellite state feature, an epoch state feature, a signal state feature and a positioning state feature, the satellite state feature and the epoch state feature are determined based on at least the current epoch signal, and the signal state feature is determined based on at least the starting point of the current epoch signal; the positioning state feature is determined based on a state covariance matrix of the last epoch, and the state covariance matrix of each epoch is updated from the initial state covariance matrix at each positioning;

[0011] Step 4: using a pre-trained neural network intelligent agent to determine the weight of the current epoch signal based on the state feature of the current epoch signal;

[0012] Step 5: using a recursive algorithm to determine the real-time position of the receiver in the current epoch based on the current epoch signal and the weight of the current epoch signal, and updating the state covariance matrix of the current epoch.

[0013] In a second aspect, the application provides a low-orbit Doppler positioning device based on epoch dynamic weighting of an intelligent agent, which comprises:

[0014] An acquisition module configured to obtain an initial accumulated low-orbit non-cooperative signal of a receiver, and determine an initial position and an initial state covariance matrix of the receiver based on the initial accumulated low-orbit non-cooperative signal;

[0015] A determination module configured to, in response to receiving a low-orbit non-cooperative signal of a current epoch, determine a starting point of the current epoch signal to achieve time frame synchronization;

[0016] The determining module is further configured to determine a state feature of the current epoch signal; the state feature at least includes a satellite state feature, an epoch state feature, a signal state feature and a positioning state feature, the satellite state feature and the epoch state feature are determined based on at least the current epoch signal, and the signal state feature is determined based on at least a starting point of the current epoch signal; the positioning state feature is determined based on a state covariance matrix of a previous epoch, and the state covariance matrix of each epoch is updated at each positioning time from an initial state covariance matrix;

[0017] The weighting module is configured to determine a weight of the current epoch signal based on the state feature of the current epoch signal using a pre-trained neural network agent.

[0018] The positioning module is configured to determine a real-time position of the receiver at the current epoch based on the current epoch signal and the weight of the current epoch signal using a recursive algorithm, and update the state covariance matrix of the current epoch.

[0019] In a third aspect, the present application provides an electronic device, comprising a memory and a processor, the memory stores a computer program, and the processor implements the low-orbit Doppler positioning method based on dynamic weighting of epochs by agents as described above when executing the computer program.

[0020] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the low-orbit Doppler positioning method based on dynamic weighting of epochs by agents as described above.

[0021] The present application has the following beneficial effects:

[0022] The technical scheme provided by the embodiments of the present application determines the initial position and the initial state covariance matrix of the receiver based on the initial accumulated low-orbit non-cooperative signal of the receiver, determines the starting point of the current epoch signal to realize time frame synchronization when receiving a new low-orbit non-cooperative signal, determines the state feature of the current epoch signal based on at least the starting point, uses a pre-trained agent to determine the weight of the current epoch signal based on the state feature, and uses a recursive algorithm to determine the real-time position of the receiver based on the current epoch signal and the weight of the current epoch signal, thereby realizing adaptive weighting based on the multi-dimensional features of the epoch signal, improving the accuracy of the determined weight of the epoch signal, and improving the positioning accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a flowchart of the low-orbit Doppler positioning method based on dynamic weighting of epochs by agents provided by the embodiments of the present application.

[0024] Figure 2 is a flowchart of a method for determining the weight of the current epoch signal based on the state features of the current epoch signal using a pre-trained neural network agent according to an embodiment of the present application.

[0025] Figure 3 is a flowchart of a method for determining the real-time position of the receiver at the current epoch and updating the state covariance matrix of the current epoch according to an embodiment of the present application.

[0026] Figure 4 is a flowchart of another low earth orbit Doppler positioning method based on dynamic weighting of epochs by an agent according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] The present application will be further described below in conjunction with the accompanying drawings and embodiments.

[0028] Figure 1 is a flowchart of a low earth orbit Doppler positioning method based on dynamic weighting of epochs by an agent according to an embodiment of the present application. As shown in the figure, the method comprises the following steps: Figure 1

[0029] In step 1, the initial accumulated low earth orbit non-cooperative signal of the receiver is obtained, and the initial position and the initial state covariance matrix of the receiver are determined based on the initial accumulated low earth orbit non-cooperative signal.

[0030] In step 2, in response to receiving the low earth orbit non-cooperative signal of the current epoch, the starting point of the current epoch signal is determined to achieve time frame synchronization.

[0031] In step 3, the state features of the current epoch signal are determined.

[0032] The state features include at least satellite state features, epoch state features, signal state features, and positioning state features. The satellite state features and the epoch state features are determined based on at least the current epoch signal, and the signal state features are determined based on at least the starting point of the current epoch signal. The positioning state features are determined based on the state covariance matrix of the previous epoch, and the state covariance matrix of each epoch is updated at each positioning starting from the initial state covariance matrix.

[0033] In step 4, a pre-trained neural network agent is used to determine the weight of the current epoch signal based on the state features of the current epoch signal.

[0034] In step 5, a recursive algorithm is used to determine the real-time position of the receiver at the current epoch based on the current epoch signal and the weight of the current epoch signal, and to update the state covariance matrix of the current epoch.

[0035] ​In some embodiments of the present application, the method can be performed by a server or a terminal device with certain processing capability. In an example, the terminal device can be a receiver.

[0036] In some embodiments of the present application, an initial accumulated low earth orbit non-cooperative signal of the receiver can be obtained first, and an initial position and an initial state covariance matrix of the receiver can be determined based on the initial accumulated low earth orbit non-cooperative signal.

[0037] Further, when the low earth orbit non-cooperative signal of the current epoch is received, a starting point of the current epoch signal can be determined to achieve time frame synchronization. That is, the starting point of the pilot signal can be determined by determining the starting point of the current epoch signal, and then the current epoch signal frame is demodulated, thereby providing a basis for subsequent extraction of the state features of the current epoch signal.

[0038] In some embodiments of the present application, the state features of the current epoch can be determined, which at least include satellite state features, epoch state features, signal state features and positioning state features.

[0039] Among them, the satellite state features and the epoch state features are determined based on at least the current epoch signal, the signal state features are determined based on at least the starting point of the current epoch signal, and the positioning state features are determined based on the state covariance matrix of the previous epoch, and the state covariance matrix of each epoch is updated at each positioning starting from the initial state covariance matrix.

[0040] In some embodiments of the present application, a pre-trained neural network agent can be used to determine the weight of the current epoch signal based on the determined state features of the current epoch. Then, a recursive algorithm can be used to determine the real-time position of the receiver in the current epoch based on the current epoch signal and the weight of the current epoch signal, and to update the state covariance matrix of the current epoch.

[0041] By using the technical solutions provided in the embodiments of the present application, the initial position and the initial state covariance matrix of the receiver are determined based on the initial accumulated low earth orbit non-cooperative signal of the receiver, the starting point of the current epoch signal is determined to achieve time frame synchronization when a new low earth orbit non-cooperative signal is received, the state features of the current epoch signal are determined based on at least the starting point, the state features at least include satellite features, epoch state features, signal state features and positioning state features, a pre-trained agent is used to determine the weight of the current epoch signal based on the state features, and a recursive algorithm is used to determine the real-time position of the receiver based on the current epoch signal and the weight of the current epoch signal, thereby realizing adaptive weighting based on the multi-dimensional features of the epoch signal, improving the accuracy of the determined weight of the epoch signal, and improving the positioning accuracy.

[0042] In some embodiments of the present application, obtaining the initial accumulated low earth orbit non-cooperative signal of the receiver and determining the initial position and initial state covariance matrix of the receiver based on the initial accumulated low earth orbit non-cooperative signal can be, first, obtaining the initial accumulated low earth orbit non-cooperative signal of the receiver, the initial accumulated low earth orbit non-cooperative signal including the low earth orbit non-cooperative signals accumulated and received by the receiver within a preset time since the receiver is started or the positioning function is turned on. Then, constructing an initial Doppler linear observation equation based on the Doppler observation of the low earth orbit non-cooperative signal. Finally, using a least squares algorithm to solve the initial Doppler linear observation equation to obtain the initial position of the receiver, and determining the initial state covariance matrix based on the coefficient matrix of the initial Doppler linear observation equation.

[0043] That is, the low earth orbit Doppler positioning method based on epoch dynamic weighting of the agent provided by the embodiments of the present application processes the low earth orbit non-cooperative signal received on the ground, such as the Iridium signal, extracts multi-dimensional feature parameters from the satellite epoch, uses a pre-trained neural network-based agent, and considers the complex, multi-level and dynamically changing satellite epoch features by the agent to evaluate the credibility of the epoch and convert the epoch credibility into adaptive weights, and integrates the adaptive weights into a recursive algorithm, such as a recursive least squares (RLS) algorithm, for iterative positioning, thereby solving the epoch redundancy problem and improving the positioning accuracy.

[0044] Since the result of each step of the RLS iterative positioning is recursively obtained based on the result of the previous step and the newly added observation, a reliable initial positioning solution is constructed in the initial observation stage as the initial position of the receiver, and the initial state covariance matrix is determined as the basis for the first step of iteration.

[0045] Suppose the time when the receiver starts to receive the signal is , the complete Iridium signal received in can be cached locally for initial positioning. Then, the Iridium signals received in are divided into signal blocks with a length of 4.32s, and each signal block is bandpass filtered. Each signal block is further divided into equal-length signal segments, and each signal segment is subjected to Fast Fourier Transform (FFT) and then peak detection.

[0046] Since the pilot signal is an unmodulated single carrier with concentrated energy in the frequency domain, the approximate position of the pilot signal in the signal block can be determined by comparing the peak values of the FFT of each signal segment. The estimation result of the FFT is the coarse estimation value of the Doppler frequency corresponding to the epoch , wherein, is a coarse estimate of Doppler frequency for the current epoch, is the serial number of the largest spectral line in the current FFT, is the sampling frequency, is the number of operation points of the FFT operation.

[0047] This step is only a rough search and search for the signal, the number of points of the FFT is small, resulting in limited estimation accuracy, but the estimate is sufficient for the accuracy requirement of the initial positioning solution. Assuming , the total number of epochs of the i-th satellite is , the total number of epochs of the i-th satellite is , the initial observation equation can be constructed as:

[0048] ;

[0049] where is the approximate value of the Taylor expansion of the rate of change of pseudo-range, , and are the partial derivatives of the rate of change of pseudo-range in three directions, is the covariance matrix of the joint Gaussian distribution white noise. For the sake of simplicity, the above initial observation equation can be written as

[0050] , where is the initial observation vector, is the initial coefficient matrix, is the initial state vector. According to the solution of the least squares method, the minimum variance unbiased estimate of is , is the initial position solution, i.e., the initial position of the receiver.

[0051] On the other hand, the initial state covariance matrix can be expressed as , where is the initial state covariance matrix.

[0052] In some embodiments of the present application, after the initial position of the receiver and the initial state covariance matrix are determined, the starting point of the current epoch signal received this time can be determined to achieve time frame synchronization every time a new low-orbit non-cooperative signal is received.

[0053] In an example, the starting point of the current epoch signal can be determined in the following manner: first, the received current epoch signal is divided into blocks according to a preset period, and each block signal is preprocessed to obtain a target time frame signal, which at least includes the pilot and unique word of the current epoch signal; wherein the preset period matches the beam polling period of the current epoch signal.

[0054] Then, the target time frame signal is segmented and processed, and each segmented signal of the target time frame signal is subjected to fast Fourier transform (FFT) to obtain a frequency domain peak value of each segmented signal, the segmented signal satisfying a preset peak value condition is determined as a target segmented signal, and a Doppler frequency coarse estimation value of the pilot is determined based on the frequency domain peak value in the target segmented signal.

[0055] Next, a local unique word signal is generated using prior information, and a peak value of a cross-correlation function of the local unique word signal and a pilot rough position signal is determined as a pilot starting position, the pilot rough position signal being a signal with a distance to the pilot rough position less than a preset distance threshold; wherein the prior information at least includes prior pilot information and prior unique word information.

[0056] Finally, the starting position of the time frame is determined based on the pilot starting position.

[0057] In RLS iterative positioning, each new captured epoch is subjected to one iteration. However, each input epoch is not equal in status, but needs to be weighted. The greater the weight of the epoch, the greater the influence on the positioning result; otherwise, the smaller the influence. Therefore, the weight of each epoch signal needs to be determined. The technical solution provided by the embodiments of the present application can use an agent to determine the weight of each epoch signal. When determining the weight, the state features of each epoch signal need to be extracted, and then the state features are input into the agent to obtain the weight of each epoch signal.

[0058] When extracting the state features, the starting point of the pilot signal needs to be determined first, which is also the starting point of the time frame signal. Based on the starting point, the time frame signal can be demodulated, and then at least part of the state features can be extracted.

[0059] In some embodiments of the present application, the above Doppler frequency coarse estimation value can be used to generate a local signal in combination with the prior pilot and unique word modulation information of the Iridium frame:

[0060] ;

[0061] wherein, and are the amplitudes of the pilot and unique word, is a sampling rate, is the number of sampling points of each Iridium symbol, for the prior iridium star unique character symbol information, for the upward rounding symbol.

[0062] Then the cross-correlation function of the generated signal and the signal near the rough search position is calculated, and the peak value is the starting position of the iridium star frame pilot signal , wherein, is the generated local signal , the number of sampling points, is the signal near the position of the segmented FFT rough estimation. Once the accurate starting position of each pilot signal is determined, the subsequent algorithm operation can accurately segment each part of the iridium star frame, so as to estimate all characteristic parameters of the epoch.

[0063] In some embodiments of the present application, the satellite state characteristics can include the age of ephemeris corresponding to the current epoch, the satellite azimuth angle of the current epoch, the satellite elevation angle of the current epoch, and the satellite-geodetic distance of the current epoch.

[0064] In some embodiments of the present application, the age of ephemeris corresponding to the current epoch can be determined in the following way: first, the update time of the two-line orbital data TLE ephemeris used in the current epoch and the signal receiving time of the current epoch are obtained. Then, the difference between the signal receiving time of the current epoch and the update time of the TLE ephemeris used in the current epoch is determined as the age of ephemeris corresponding to the current epoch.

[0065] That is, if the update time of the TLE ephemeris used in the current epoch is , the receiving time of the current epoch is , then the age of ephemeris is .

[0066] In some embodiments of the present application, the satellite azimuth angle of the current epoch, the satellite elevation angle of the current epoch, and the satellite-geodetic distance of the current epoch can be determined in the following way:

[0067] First, the first coordinate point of the satellite sending the signal of the current epoch in the Earth-Centered Earth-Fixed coordinate system is determined according to the TLE ephemeris used in the current epoch, and the second coordinate point of the real-time position of the receiver in the previous epoch in the Earth-Centered Earth-Fixed coordinate system is obtained.

[0068] Then, the three-dimensional coordinate components of the second coordinate point are converted into latitude, longitude and height respectively, and a rotation matrix is constructed, wherein, is the latitude, is the longitude. The formula is used to convert the three-dimensional coordinate components of the first coordinate point into the east, north and zenith components in the station-centered coordinate system based on the rotation matrix, wherein is the east component, is the north component, the zenith component, 、 and are three-dimensional coordinate components of the first coordinate point in the Earth-Centered Earth-Fixed coordinate system.

[0069] Next, the satellite azimuth angle at the current epoch is calculated using the formula and the satellite elevation angle at the current epoch is calculated using the formula .

[0070] Finally, the Euclidean distance between the first coordinate point and the second coordinate point is determined as the satellite-geodetic distance at the current epoch.

[0071] In some embodiments of the present application, the epoch state feature can also include a satellite unique identifier at the current epoch and an epoch number at the current epoch.

[0072] In certain embodiments of the present application, if the current epoch signal is an Iridium non-cooperative signal received at the current epoch, the satellite unique identifier at the current epoch can be determined in the following manner:

[0073] Based on the starting point position at the current epoch, a simplex channel signal of the current epoch signal is determined, which is a signal composed of 163 characters taken backward from the starting point position; wherein the signal composed of the first 64 characters is a pilot signal, and the signal composed of the last 99 characters is an independent word and data signal.

[0074] Doppler frequency estimation is performed on the current epoch signal to obtain a Doppler estimation frequency of the current epoch signal.

[0075] First, the formula is used to down-convert the simplex channel signal based on the Doppler estimation frequency to obtain a down-converted simplex channel signal; wherein is the down-converted simplex channel signal, is the simplex channel signal, is an imaginary symbol, is the Doppler estimation frequency, is a sampling rate when performing segmented Fast Fourier Transform (FFT) on the pilot signal in determining the Doppler estimation frequency, and n is a time domain sampling point index of the pilot, each n value representing a sampling point.

[0076] Then, the formula is used to determine a phase reference of the pilot signal; wherein is the phase reference, is the number of sampling points contained in the pilot signal, is a conjugate symbol. Further, the phase of the i-th character of the simplex channel signal is determined as ; wherein is the phase of the i-th character, ms represents the time slot unit in milliseconds, and i is a positive integer greater than or equal to 1 and less than or equal to 163.

[0077] Then, each character is demodulated by quadrature phase shift keying (QPSK), and the demodulation result of the i-th character is obtained as ;in is the demodulation result of the i-th character. And, the characters of the independent word and data signal are converted according to the formula , obtaining the bit data of the independent word and data signal, the bit data includes 198 bits of data.

[0078] Then the adjacent bits in the bit data and exchange, is an integer greater than or equal to 0 and less than 99. Data bits 31 to 94 of the swapped bit data are obtained and standard deinterleaving is performed on the obtained data to obtain 64-bit data. Standard linear error correction code (BCH) decoding is performed on the first 31 bits of the 64-bit data to obtain 21 valid data bits and 10 check bits.

[0079] Finally, the first 7 bits of the 21 valid data bits are converted into a decimal number to obtain the unique satellite identifier of the current epoch.

[0080] In certain embodiments of the present invention, the epoch number of the current epoch can be determined as follows: first, epoch signals received by the receiver during local positioning iterations are sorted according to satellite unique identifiers. Then, within each satellite unique identifier category, the epoch signals are sorted in order of their reception time. Finally, the sorted number corresponding to the current epoch signal within the corresponding satellite unique identifier category is determined as the epoch number of the current epoch.

[0081] In other words, the sequence number of an epoch can be defined as the number of epochs received from the satellite from which it originated in this positioning iteration. That is, each epoch is classified by satellite ID, and then numbered in the order of reception time under each satellite ID.

[0082] In some embodiments of the present invention, the signal state characteristics may further include the Doppler estimation frequency of the current epoch signal and the signal-to-noise ratio of the current epoch signal.

[0083] In some embodiments of the present invention, the Doppler estimated frequency of the current epoch signal may be determined in the following manner:

[0084] First, perform a segmented fast Fourier transform (FFT) on the current epoch signal to obtain a rough estimate of the Doppler frequency of the current epoch signal. ; The sampling rate of the segmented FFT is .

[0085] Then the pilot signal is determined based on the starting point of the current epoch The pilot signal is subjected to a fast Fourier transform (FFT) again to obtain a pilot spectrum ; wherein n is a time-domain sampling point index of the pilot, each n value represents a sampling point, and k is a spectrum line sequence number of the pilot spectrum.

[0086] Then a Doppler frequency estimation compensation value of the current epoch signal is determined using a formula ; wherein is the Doppler frequency estimation compensation value of the current epoch signal, is a number of sampling points contained in the pilot signal, is a spectrum line sequence number corresponding to a spectrum line with the maximum length in the pilot spectrum, and a is a compensation direction. .

[0087] Finally, a sum of a Doppler frequency coarse estimation value and the Doppler frequency estimation compensation value is the Doppler estimation frequency of the current epoch signal.

[0088] In some embodiments of the present application, a signal-to-noise ratio of the current epoch signal can be determined in the following manner:

[0089] The signal-to-noise ratio of the current epoch signal is calculated using a formula ; wherein is the signal-to-noise ratio of the current epoch signal, and N pilot is equal to N1.

[0090] In some embodiments of the present application, the positioning state feature includes a geometric dilution of precision (GDOP) of a previous epoch.

[0091] In some embodiments, the GDOP of the previous epoch can be determined in the following manner: first, a state covariance matrix of the previous epoch is obtained , and then a GDOP is calculated using a formula ; wherein tr() is a trace function.

[0092] That is, the GDOP is not a feature of a single epoch, but it reflects the recursive state when the epoch is added, and thus it also has important significance. Since the feature parameter is obtained, the new epoch has not participated in the RLS iteration, and thus the covariance matrix obtained in the previous step is used to obtain the GDOP.

[0093] By the above method, the accurate Doppler frequency, signal-to-noise ratio, satellite ID, elevation angle, azimuth angle, satellite-ground distance, ephemeris age, epoch number, and generalized GDOP of each epoch can be extracted, totaling 9-dimensional feature parameters. Next, the state features of the signals of each epoch composed of the 9-dimensional feature parameters can be input into the pre-trained neural network agent, and the agent can determine the weight of the signals of each epoch based on the feature parameters.

[0094] Figure 2 is a flowchart of a method for determining the weight of a current epoch signal based on the state features of the current epoch signal using a pre-trained neural network agent according to an embodiment of the present application. As shown in Figure 2 , the method comprises the following steps:

[0095] In step 41, a pre-trained neural network agent is obtained.

[0096] The pre-trained neural network agent is trained based on historical low-orbit non-cooperative signals obtained by a receiver.

[0097] In step 42, the state features of the current epoch signal are input into the pre-trained neural network agent to obtain the weight of the current epoch signal.

[0098] In some embodiments of the present application, a pre-trained neural network agent trained based on historical low-orbit non-cooperative signals obtained by a receiver can be obtained, and then the state features of the current epoch signal are input into the pre-trained neural network agent to obtain the weight of the current epoch signal.

[0099] That is, the 9-dimensional feature parameters determined above are denoted as , n represents the n-th positioning iteration, and the parameters from left to right are Doppler frequency, signal-to-noise ratio, satellite-ground distance, azimuth angle, elevation angle, ephemeris age, epoch number, and generalized GDOP. The pre-trained neural network-based agent can determine the weight of the new epoch as .

[0100] Figure 3 is a flowchart of a method for determining the real-time position of a receiver at a current epoch and updating the state covariance matrix of the current epoch according to an embodiment of the present application. As shown in Figure 3 , the method comprises the following steps:

[0101] In step 51, the Doppler observation in the current epoch signal is obtained, and a Doppler linear observation equation of the current epoch is constructed based at least on the Doppler observation.

[0102] The Doppler linear observation equation of the current epoch includes a coefficient matrix of the current epoch.

[0103] In step 52, a state covariance matrix of the current epoch is determined based on at least the state covariance matrix of the previous epoch, the coefficient matrix of the current epoch and the weight of the current epoch signal.

[0104] In one example, the formula may be used to determine the state covariance matrix of the current epoch, wherein is the state covariance matrix of the current epoch, is the state covariance matrix of the previous epoch, is the coefficient matrix of the current epoch, is the weight of the current epoch signal, and is an identity matrix.

[0105] In step 53, a gain matrix of the current epoch is determined based on at least the state covariance matrix of the current epoch and the coefficient matrix of the current epoch.

[0106] In one example, the formula may be used to determine the gain matrix of the current epoch, wherein is the gain matrix of the current epoch.

[0107] In step 54, a real-time position of the receiver at the current epoch is determined based on at least the real-time position of the receiver at the previous epoch, the Doppler observation in the current epoch signal and the coefficient matrix of the current epoch.

[0108] In one example, the formula may be used to determine the real-time position of the receiver at the current epoch, wherein is the real-time position of the receiver at the current epoch, is the real-time position of the receiver at the previous epoch, is the Doppler observation in the current epoch signal.

[0109] That is, the receiver performs one RLS iteration every time a new epoch is captured. The input quantities participating in the iteration are the results of the previous iteration and , the weight of the newly added epoch , the observation of the newly added epoch and . The output quantities are the iteration results of this step and , which will participate in the next RLS iteration.

[0110] Suppose in the nth iteration, the observation of the newly added epoch is , and the coefficient matrix is , their relationship with the rate of change of pseudo-range is:

[0111] ;

[0112] The steps of the single recursive least squares iteration are as follows:

[0113] First step: the covariance matrix is recursively updated. When a new observation is added at the nth step, the new covariance matrix is updated based on the covariance matrix at the (n-1)th step and the weight obtained by the agent, expressed as:

[0114] .

[0115] Second step: determine the gain matrix at the nth step for balancing the influence of the new observation on the state update, which can be expressed as .

[0116] Third step: obtain the weighted recursive update of the positioning solution at the nth step, expressed as: .

[0117] By continuously adding new epochs, weighting, and cyclically iterating the three recursive steps, the positioning solution after the addition of each epoch can be recursively obtained, and the final position estimate obtained at the last epoch.

[0118] Figure 4 is another flowchart of a low-orbit Doppler positioning method based on the dynamic weighting of epochs by the agent provided by the embodiments of the present application. As shown in Figure 4 , the initial position can be first solved using the batch least squares method, and then the observation characteristics of the new epoch and the latest GDOP are obtained. Next, the agent performs credibility evaluation on the epoch, maps the evaluated credibility to the epoch weight, and performs recursive least squares iteration combined with the determined epoch weight to iteratively solve the receiver position in real time. Then, it is determined whether to continue to obtain the epoch signal. If yes, the steps are sequentially executed from the observation characteristics of the new epoch and the latest GDOP, and the new receiver real-time position is obtained. Otherwise, if the acquisition of the epoch signal is stopped, the positioning is ended.

[0119] The technical solution provided by the embodiments of the present application proposes a recursive least squares low-orbit satellite opportunity signal positioning framework with real-time weighting by the agent, which can evaluate the credibility of each epoch according to the multi-dimensional characteristic parameters of the epoch in real time, and convert it into a weight participating in the iteration. This framework can enhance the contribution of high-quality epochs to the positioning accuracy and weaken the adverse effects of poor epochs on the positioning accuracy, thus significantly improving the positioning accuracy in the iteration process.

[0120] Meanwhile, a fine processing flow for low-orbit non-cooperative signals such as Iridium satellites is designed, which can extract the multi-dimensional characteristic parameters of each epoch from the Iridium satellite signals, such as Doppler frequency, signal-to-noise ratio, satellite ID, pitch angle, azimuth angle, satellite-ground distance, ephemeris age, epoch number, etc.

[0121] The technical scheme provided by the embodiment of the application has the following advantages:

[0122] Improving positioning accuracy: By dynamically evaluating the multi-dimensional feature parameters of the epoch (including Doppler frequency, signal-to-noise ratio, satellite ID, elevation angle, azimuth angle, satellite-ground distance, ephemeris age, epoch number and generalized GDOP) of the agent, and adaptively assigning weights, high-quality epochs can be more accurately screened, and the interference of low-quality epochs can be suppressed. Compared with the traditional fixed weight or single feature optimization method (such as the OECW algorithm which only relies on azimuth and elevation angles), the present application comprehensively considers multi-level error sources, significantly improves the stability and accuracy of the positioning result.

[0123] Enhancing real-time performance: Using the recursive least squares (RLS) framework combined with real-time weighting of the agent, the high computational complexity problem of batch least squares method is avoided, and efficient iterative solution is realized. At the same time, the dynamic decision of the agent optimizes the epoch utilization rate, reduces redundant calculation, makes the positioning process converge faster, and is suitable for real-time navigation requirements.

[0124] Adapting to complex environments: The present application can dynamically cope with complex scenarios such as signal obstruction, multipath effect, ephemeris error, etc. through multi-dimensional feature fusion. For example, epochs with high signal-to-noise ratio and new ephemeris will be assigned higher weights, while epochs with interference or old ephemeris will be de-weighted, thereby improving the robustness of the system in harsh environments.

[0125] Flexibility and scalability: The agent is trained based on neural networks, and the weight strategy can be continuously optimized through data learning, adapting to different satellite constellations (such as Iridium, Starlink) or signal characteristics. In addition, the dimension of feature parameters can be further expanded (such as adding carrier phase, Doppler rate, etc.), leaving room for future technology upgrades.

[0126] On the other hand, a low-orbit Doppler positioning device based on the dynamic weighting of epochs by the agent, the device comprises:

[0127] The acquisition module is configured to acquire the initial accumulated low-orbit non-cooperative signal of the receiver, and determine the initial position and initial state covariance matrix of the receiver based on the initial accumulated low-orbit non-cooperative signal;

[0128] The determination module is configured to determine the starting point of the current epoch signal to realize time frame synchronization in response to receiving the low-orbit non-cooperative signal of the current epoch;

[0129] The determining module is further configured to determine a state feature of the current epoch signal; the state feature comprises at least a satellite state feature, an epoch state feature, a signal state feature and a positioning state feature, the satellite state feature and the epoch state feature are determined based on at least the current epoch signal, and the signal state feature is determined based on at least a starting point of the current epoch signal; and the positioning state feature is determined based on a state covariance matrix of a previous epoch, and the state covariance matrix of each epoch is updated at each positioning time from an initial state covariance matrix.

[0130] The weighting module is configured to determine a weight of the current epoch signal based on the state feature of the current epoch signal using the pre-trained neural network agent.

[0131] The positioning module is configured to determine a real-time position of the receiver at the current epoch based on the current epoch signal and the weight of the current epoch signal using a recursive algorithm, and update the state covariance matrix of the current epoch.

[0132] In a third aspect, the present application provides an electronic device, comprising a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned low-orbit Doppler positioning method based on dynamic weighting of epochs by agents when executing the computer program.

[0133] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the above-mentioned low-orbit Doppler positioning method based on dynamic weighting of epochs by agents.

[0134] The above specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above are only specific embodiments of the present application and are not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A low-orbit Doppler positioning method based on dynamic weighting of epochs by an intelligent agent, characterized in that: The method comprises: Step 1: Obtain an initial accumulated low-orbit non-cooperative signal of the receiver, and determine an initial position and an initial state covariance matrix of the receiver based on the initial accumulated low-orbit non-cooperative signal; Step 2: In response to receiving the low-orbit non-cooperative signal of the current epoch, determining the starting point of the current epoch signal to achieve time frame synchronization; Step 3: Determine the state characteristics of the current epoch signal; the state characteristics include at least satellite state characteristics, epoch state characteristics, signal state characteristics and positioning state characteristics, the satellite state characteristics and the epoch state characteristics are determined based on at least the current epoch signal, and the signal state characteristics are determined based on at least the starting point of the current epoch signal; the positioning state characteristics are determined based on the state covariance matrix of the previous epoch, and the state covariance matrix of each epoch starts from the initial state covariance matrix and is updated at each positioning; Step 4: Using a pre-trained neural network agent, determine the weight of the current epoch signal based on the state characteristics of the current epoch signal; Step 5: Using a recursive algorithm, determine the real-time position of the receiver at the current epoch based on the current epoch signal and the weight of the current epoch signal, and update the state covariance matrix of the current epoch.

2. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 1, characterized in that: The step 1 comprises: Obtaining an initial accumulated low-orbit non-cooperative signal of the receiver, where the initial accumulated low-orbit non-cooperative signal includes a cumulative low-orbit non-cooperative signal received within a preset time after the receiver is powered on or the positioning function is enabled; constructing an initial Doppler linear observation equation based on the Doppler observation quantity of the low-orbit non-cooperative signal; The initial Doppler linear observation equation is solved using a least squares algorithm to obtain the initial position of the receiver, and the initial state covariance matrix is ​​determined based on the coefficient matrix of the initial Doppler linear observation equation.

3. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 2, characterized in that: The step 2 includes: Dividing the received current epoch signal into blocks according to a preset period, and preprocessing each block signal to obtain a target time frame signal, wherein the target time frame signal includes at least a pilot and a unique word of the current epoch signal; wherein the preset period matches a beam polling period of the current epoch signal; Segment processing is performed on the target time frame signal, and a fast Fourier transform (FFT) is performed on each segment signal of the target time frame signal to obtain a frequency domain peak value of each segment signal, determining a segment signal whose frequency domain peak value meets a preset peak condition as a target segment signal, and determining a rough Doppler frequency estimate value of the pilot signal based on the frequency domain peak value in the target segment signal; Generate a local independent word signal using prior information, determine the peak of the cross-correlation function between the local independent word signal and the panned pilot coarse position signal as the pilot start position, and the panned pilot coarse position signal is a signal whose distance from the pilot coarse position is less than a preset distance threshold; wherein the prior information includes at least prior pilot information and prior unique word information; The starting position of the time frame is determined based on the pilot starting position.

4. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 1, characterized in that: The satellite status characteristics include the ephemeris age corresponding to the current epoch, the satellite azimuth angle of the current epoch, the satellite pitch angle of the current epoch and the satellite-to-ground distance of the current epoch.

5. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 4, characterized in that: The ephemeris age corresponding to the current epoch is determined in the following manner: Get the update time of the two-line orbit data TLE ephemeris used in the current epoch and the signal reception time of the current epoch; The difference between the signal reception time of the current epoch and the update time of the TLE ephemeris used in the current epoch is determined as the ephemeris age corresponding to the current epoch.

6. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 4, characterized in that: The satellite azimuth of the current epoch, the satellite pitch angle of the current epoch, and the satellite-to-ground distance of the current epoch are determined in the following manner: Determine, based on the TLE ephemeris used for the current epoch, a first coordinate point of the satellite that sends the current epoch signal in the Earth-centered Earth-fixed coordinate system; Obtain the second coordinate point of the real-time position of the receiver in the previous epoch in the Earth-centered Earth-fixed coordinate system; Convert the three-dimensional coordinate components of the second coordinate point into latitude, longitude and altitude respectively, and construct a rotation matrix ,in, is the latitude, is the longitude; Using the formula The three-dimensional coordinate components of the first coordinate point are converted into the east, north and zenith components in the station-centered coordinate system based on the rotation matrix, where is the east component, is the north component, is the zenith component, 、 and is the three-dimensional coordinate component of the first coordinate point in the Earth-centered Earth-fixed coordinate system; Using the formula Calculating the satellite azimuth of the current epoch; Using the formula Calculating the satellite pitch angle of the current epoch; The Euclidean distance between the first coordinate point and the second coordinate point is determined as the satellite-to-earth distance at the current epoch.

7. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 1, characterized in that: The epoch state characteristics include a satellite unique identifier of the current epoch and an epoch sequence number of the current epoch.

8. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 7, characterized in that: In response to determining that the current epoch signal is an Iridium non-cooperative signal received at the current epoch, the satellite unique identifier of the current epoch is determined in the following manner: Determining a simplex channel signal of the current epoch signal based on the starting point of the current epoch, wherein the simplex channel signal is a signal consisting of 163 characters intercepted from the starting point backward; wherein the signal consisting of the first 64 characters is a pilot signal, and the signal consisting of the last 99 characters is an independent word and data signal; Performing Doppler frequency estimation on the current epoch signal to obtain the Doppler estimated frequency of the current epoch signal; Using the formula Down-converting the simplex channel signal based on the Doppler estimation frequency to obtain a down-converted simplex channel signal; wherein is the simplex channel signal after down-conversion, is a simplex channel signal, is the imaginary number symbol, is the Doppler estimated frequency, is the sampling rate when performing segmented fast Fourier transform (FFT) on the pilot signal when determining the Doppler estimation frequency, n is the time domain sampling point index of the pilot, and each value of n represents a sampling point; Using the formula Determine the phase reference of the pilot signal; wherein is the phase reference, is the number of sampling points contained in the pilot signal, is the conjugate symbol; Determine the phase of the i-th symbol of the simplex channel signal as ;in is the phase of the i-th character, ms represents the time slot unit in milliseconds, and i is a positive integer greater than or equal to 1 and less than or equal to 163; Perform quadrature phase shift keying (QPSK) demodulation on each character, and the demodulation result of the i-th character is ;in is the demodulation result of the i-th character; The independent words and characters of the data signal are converted according to the formula , obtaining bit data of the independent word and the data signal, wherein the bit data includes 198 bits of data; The adjacent bits in the bit data and exchange, is an integer greater than or equal to 0 and less than 99; Obtaining data from bits 31 to 94 of the swapped bit data, and performing standard deinterleaving processing on the obtained data to obtain 64-bit data; Performing standard linear error correction code (BCH) decoding on the first 31 bits of the 64-bit data to obtain 21 valid data bits and 10 check bits; The first 7 bits of the 21 valid data bits are converted into a decimal number to obtain a unique satellite identifier of the current epoch.

9. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 7, characterized in that: The epoch number of the current epoch is determined in the following manner: Classify the epoch signals received by the receiver in the local positioning iteration according to the satellite unique identifier; Under each satellite unique identifier category, sort the epoch signal numbers in the order of reception time; Determine the sorting sequence number corresponding to the current epoch signal under the corresponding satellite unique identifier classification as the epoch sequence number of the current epoch.

10. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 1, characterized in that: The signal state characteristics include the Doppler estimated frequency of the current epoch signal and the signal-to-noise ratio of the current epoch signal.

11. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 10, characterized in that: The Doppler estimated frequency of the current epoch signal is determined in the following manner: Perform a segmented fast Fourier transform (FFT) on the current epoch signal to obtain a rough estimate of the Doppler frequency of the current epoch signal. ; The sampling rate of the segmented FFT is ; Determine the pilot signal based on the starting point of the current epoch , perform fast Fourier transform FFT on the pilot signal again to obtain the pilot spectrum ; Where n is the time domain sampling point index of the pilot, each n value represents a sampling point, and k is the spectral line number of the pilot spectrum; Using the formula Determine the Doppler frequency estimation compensation value of the current epoch signal; wherein is the Doppler frequency estimation compensation value of the current epoch signal, is the number of sampling points contained in the pilot signal, is the spectral line number corresponding to the spectral line with the largest modulus length in the pilot spectrum, a is the compensation direction, ; Determine the rough estimate of the Doppler frequency and the Doppler frequency estimation compensation value The sum is the Doppler estimated frequency of the current epoch signal.

12. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 11, characterized in that: The signal-to-noise ratio of the current epoch signal is determined as follows: Using the formula Calculate the signal-to-noise ratio of the current epoch signal; is the signal-to-noise ratio of the current epoch signal, N pilot Equal to N1.

13. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 1, characterized in that: The positioning state characteristics include the generalized geometric dilution of precision GDOP of the previous epoch; The GDOP is determined as follows: Get the state covariance matrix of the previous epoch ; Using the formula The GDOP is obtained by calculation; wherein tr() is a trace function.

14. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 1, characterized in that: The step 4 comprises: Obtaining a pre-trained neural network agent, wherein the pre-trained neural network agent is trained based on historical low-orbit non-cooperative signals acquired by the receiver; The state characteristics of the current epoch signal are input into the pre-trained neural network agent to obtain the weight of the current epoch signal.

15. The low-orbit Doppler positioning method based on agent-based dynamic weighting of epochs according to claim 1, characterized in that: The step 5 comprises: Obtaining a Doppler observation quantity in the current epoch signal, and constructing a Doppler linear observation equation for the current epoch based at least on the Doppler observation quantity, wherein the Doppler linear observation equation for the current epoch includes a current epoch coefficient matrix; Using the formula The state covariance matrix of the current epoch is determined based on at least the state covariance matrix of the previous epoch, the current epoch coefficient matrix and the weight of the current epoch signal; wherein is the state covariance matrix of the current epoch, is the state covariance matrix of the previous epoch, is the coefficient matrix of the current epoch, is the weight of the current epoch signal, is the identity matrix; Using the formula Determine the gain matrix for the current epoch; where, is the gain matrix of the current epoch; Using the formula Determine the real-time position of the receiver at the current epoch; wherein, is the real-time position of the receiver at the current epoch, is the real-time position of the receiver in the previous epoch, is the Doppler observation in the current epoch signal.

16. A low-orbit Doppler positioning device based on dynamic weighting of epochs by an intelligent agent, characterized in that: The device comprises: an acquisition module configured to acquire an initial accumulated low-orbit non-cooperative signal of the receiver, and determine an initial position and an initial state covariance matrix of the receiver based on the initial accumulated low-orbit non-cooperative signal; a determination module configured to, in response to receiving a low-orbit non-cooperative signal of a current epoch, determine a starting point of the current epoch signal to achieve time frame synchronization; The determination module is further configured to determine state characteristics of the current epoch signal; the state characteristics include at least satellite state characteristics, epoch state characteristics, signal state characteristics and positioning state characteristics, the satellite state characteristics and the epoch state characteristics are determined based on at least the current epoch signal, and the signal state characteristics are determined based on at least the starting point position of the current epoch signal; the positioning state characteristics are determined based on the state covariance matrix of the previous epoch, and the state covariance matrix of each epoch starts from the initial state covariance matrix and is updated at each positioning; a weighting module configured to determine a weight of the current epoch signal based on state characteristics of the current epoch signal using a pre-trained neural network agent; The positioning module is configured to use a recursive algorithm to determine the real-time position of the receiver at the current epoch based on the current epoch signal and the weight of the current epoch signal, and update the state covariance matrix of the current epoch.

17. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein the device is characterized in that when the processor executes the computer program, it implements the low-orbit Doppler positioning method based on dynamic weighting of epochs by an intelligent agent as described in any one of claims 1 to 15.

18. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the low-orbit Doppler positioning method based on dynamic weighting of epochs by an intelligent agent is implemented as described in any one of claims 1 to 15.

Images