A positioning system based on radio direction finding

Abstract

The present invention discloses a positioning system based on radio direction finding, comprising: a directional signal transmitting module, a signal receiving module, a signal processing module, and a positioning module; the signal transmitting module constructs a directional positioning network including at least three base stations, and the directional positioning network generates directional signals through beamforming technology; the signal receiving module receives directional signals from different base stations, extracts multi-dimensional parameters, and uploads the multi-dimensional parameters to the signal processing module; the signal processing module suppresses multipath interference on the extracted multi-dimensional parameters based on the generalized cyclic cross-correlation entropy function, uses the reverse ray tracing algorithm to inversely search for the propagation paths of the signals, excludes the non-line-of-sight propagation paths in the propagation paths, and obtains the line-of-sight propagation path parameters of each base station; the positioning module evaluates the signal reliability according to the multi-dimensional parameter quality of each base station, dynamically adjusts the weight coefficients of each base station, and performs weighted fusion on the line-of-sight path parameters and geographical location information of each base station to output the three-dimensional coordinates of the target device.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technologies, and particularly to a positioning system based on radio direction finding. Background Art

[0002] Offshore engineering involves multiple complex processes, covering the exploration, development, production, and abandonment of offshore oil and gas, as well as the installation, inspection, repair, and maintenance (IRM) of underwater structures. In these operations, accurately positioning underwater structures to their designed positions and enabling remotely operated vehicles (ROVs) to quickly and accurately locate underwater targets have become key issues to be urgently solved.

[0003] Currently, underwater positioning mainly relies on acoustic positioning technologies, including long baseline (LBL), short baseline (SBL), and ultra-short baseline (USBL) systems. The LBL has the highest accuracy, but laying out the array requires a large amount of time and cost; the SBL and USBL are relatively simple, but their working distances and accuracies are limited. Moreover, these high-precision positioning devices are usually bulky, expensive, and battery-powered, with complex installation and recovery processes, and are particularly risky in deep-water operations. To address the bottlenecks of traditional technologies, new positioning technologies, such as radio direction finding positioning, have been introduced in the field of offshore engineering. Radio direction finding positioning uses the propagation characteristics of radio waves for positioning and has advantages such as a wide coverage range and strong anti-interference ability. This method combines multiple technologies such as radio direction finding, satellite navigation, inertial navigation, and underwater acoustic positioning, and improves positioning accuracy and reliability by fusing data from different sensors.

[0004] However, radio direction finding positioning still has some limitations. In a complex marine environment, multipath interference will seriously affect the positioning accuracy. This interference will cause distortion of the positioning signal, making the received signal contain false path information, thereby affecting the accuracy of the positioning algorithm and reducing the positioning accuracy. Summary of the Invention

[0005] The present invention proposes a positioning system based on radio direction finding, which solves the problem of reduced accuracy of existing radio direction finding positioning methods caused by multipath interference.

[0006] To solve the above technical problems, the present invention provides a positioning system based on radio direction finding, including: a directional signal transmitting module, a signal receiving module, a signal processing module, and a positioning module;

[0007] The signal transmitting module: constructs a directional positioning network including at least three base stations, and the directional positioning network generates a directional signal through beamforming technology;

[0008] The signal receiving module: is disposed on the target device, receives the directional signals from different base stations, extracts multi-dimensional parameters, and uploads the multi-dimensional parameters to the signal processing module;

[0009] The signal processing module: based on the generalized cyclic cross-correlation entropy function, suppresses multipath interference for the extracted multi-dimensional parameters, uses the reverse ray tracing algorithm to inversely search for the signal propagation path, and excludes the non-line-of-sight propagation paths in the propagation path according to the multi-dimensional parameters after multipath interference suppression, obtaining the line-of-sight propagation path parameters of each base station;

[0010] The positioning module: evaluates the signal reliability according to the multi-dimensional parameter quality of each base station, dynamically adjusts the weight coefficients of each base station, and performs weighted fusion on the line-of-sight path parameters and geographical location information of each base station to output the three-dimensional coordinates of the target device.

[0011] Preferably, in the signal transmitting module, after generating the directional signal, suppresses the sidelobe interference of each beam through the least mean square error algorithm, and updates the weights of each beam. The expression of the least mean square error algorithm is:

[0012] ;

[0013] In the formula, and are the weight vectors of the th and th iterations respectively; is the step size parameter; is the conjugate of the error signal; is the input signal.

[0014] Preferably, after the signal receiving module extracts the signal strength, uses the sliding window average algorithm to denoise the signal strength, generates the smoothed signal strength by calculating the average value of the signal strength values within the window, and the size of the sliding window is adaptively adjusted according to the moving speed of the target device.

[0015] Preferably, an environmental attenuation compensation factor is introduced in the signal receiving module to correct the errors of the signal strength caused by height changes and device postures. The expression of the environmental attenuation compensation factor is:

[0016] ;

[0017] In the formula, is the environmental attenuation compensation factor; and are the weight coefficients; is the height attenuation term; is the height data measured by the barometer of the target device; is the attitude attenuation term; and are the pitch angle and roll angle output by the gyroscope of the target device respectively.

[0018] Preferably, the target device receives the directional signal from the base station through a regular tetrahedron antenna array, and the signal receiving module extracts the direction of arrival from the directional signal by using the quaternion MUSIC algorithm, including the following steps:

[0019] Step 1: Based on the three-dimensional coordinates of each antenna element in the regular tetrahedron antenna array and the incident direction vector of the signal, convert the position vector of each antenna element and the incident direction vector of the signal into quaternion form:

[0020] ;

[0021] In the formula, is the quaternion array manifold vector of the th antenna element; is the imaginary unit; is the spacing between antenna elements; is the wavelength of the signal; is the position vector of the th antenna element; is the incident direction vector of the signal;

[0022] Step 2: Calculate the covariance matrix of the quaternion signal to obtain the covariance matrix of the antenna array output;

[0023] Step 3: Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues and corresponding eigenvectors, and divide the eigenvectors into a signal subspace and a noise subspace according to the magnitude of the eigenvalues;

[0024] Step 4: Use the eigenvectors of the noise subspace to construct a spatial spectrum function, search the spatial spectrum function, find the direction that makes the spatial spectrum function obtain a minimum value, and obtain the direction of arrival of the directional signal.

[0025] Preferably, in the signal processing module, after receiving the multi-dimensional parameters uploaded by the target device, predict the direction of arrival angle according to the moving trajectory of the target device, and dynamically adjust the weight coefficients of each beam by using Kalman filtering:

[0026] ;

[0027] In the formula, is the Kalman gain matrix; is the prior estimation error covariance matrix; is the observation matrix; is the observation noise covariance matrix; is the posterior estimation error covariance matrix; is the identity matrix.

[0028] Preferably, the signal processing module suppresses multipath interference on the extracted multi-dimensional parameters based on the generalized cyclic cross-correlation entropy function, including the following steps:

[0029] S1: Perform cyclic spectral density analysis on the received multi-dimensional parameters, extract the cyclostationary characteristics of the received signal, and separate the interference components with different cyclic frequencies;

[0030] S2: Map the multi-dimensional parameters to the joint feature space, construct a three-dimensional matrix as the input of the generalized cyclic cross-correlation entropy function, and the generalized cyclic cross-correlation entropy function outputs the entropy matrix of the multi-dimensional parameters;

[0031] S3: Adjust the kernel width of the kernel function of the generalized cyclic cross-correlation entropy function using a reinforcement learning algorithm;

[0032] S4: Perform singular value decomposition on the entropy matrix, retain the first K principal components to construct a signal subspace, and suppress the components corresponding to the noise subspace;

[0033] S5: Analyze the eigenvalue distribution of the entropy matrix. When the ratio of the second largest eigenvalue to the largest eigenvalue is greater than the set threshold, it is determined as multipath interference, and the multipath interference in the multi-dimensional parameters is further reduced through an attenuation compensation mechanism.

[0034] Preferably, the expression of the generalized cyclic cross-correlation entropy function in S2 is:

[0035] ;

[0036] In the formula, represents the statistical correlation between signals and at time delay ; is the total number of signal samples; is the Gaussian kernel function; and are the directional signals from different base stations; is the time variable; is the time delay.

[0037] Preferably, S3 includes the following steps:

[0038] S31: Take the current environmental noise intensity, historical interference scenario data, and signal quality as the state space, take the kernel width adjustment rule as the action space, and calculate the reward value according to the signal processing performance and adaptability changes of the agent after performing the action. The expression for calculating the reward value is:

[0039] ;

[0040] In the formula, For signal processing performance reward; For adaptability enhancement reward; For signal-to-noise ratio improvement reward; For false detection rate reduction reward; For processing delay penalty term; For parameter adjustment speed reward; For parameter stability reward; For environmental mutation detection reward;

[0041] S32: Create a Q value table for storing the Q values of each state-action pair, where Q the value represents the expected long-term reward for performing a certain action in a certain state;

[0042] S33: The agent continuously tries different actions through interaction with the environment and updates the Q values in the Q value table based on the obtained rewards, gradually optimizing the decision-making strategy to obtain the optimal kernel width adjustment strategy.

[0043] Preferably, the signal processing module excludes the non-line-of-sight propagation paths in the propagation path according to the multi-dimensional parameters after multi-path interference suppression, including the following steps: calculating the signal propagation time difference through the time difference of arrival after multi-path interference suppression, combining the position information of each base station, inversely deriving the signal path length difference, and excluding the non-line-of-sight propagation paths; determining the signal arrival direction using the direction of arrival after multi-path interference suppression, optimizing the initial search direction of ray tracing, combining the geographic information system data of the search area to predict the main propagation path, and excluding the non-line-of-sight propagation paths; evaluating the path loss based on the signal strength after multi-path interference suppression, calculating the energy attenuation of different material surfaces in combination with the Fresnel equation and the material reflection coefficient database, verifying whether the path conforms to the line-of-sight propagation attenuation characteristics, and excluding the non-line-of-sight propagation paths.

[0044] The advantages of the present invention at least include:

[0045] 1. In a complex environment, multi-path interference will seriously affect the positioning accuracy. The present invention suppresses multi-path interference on the extracted multi-dimensional parameters based on the generalized cyclic cross-correlation entropy function, removes the influence of noise on the positioning parameters, improves the quality of positioning data, and ensures the accuracy of positioning;

[0046] 2. The reverse ray tracing algorithm is adopted to inversely search the signal propagation path, generate a three-dimensional path map in combination with geographic information, and at the same time exclude the non-line-of-sight propagation paths according to the multi-dimensional parameters after multi-path interference suppression, enabling the system to more accurately analyze the signal propagation path, reduce the error caused by non-line-of-sight propagation, and provide reliable path information for precise positioning. Brief Description of the Drawings

[0047] Figure 1 Schematic diagram of the system framework according to an embodiment of the present invention;

[0048] Figure 2 Schematic diagram of the method flow according to an embodiment of the present invention. Detailed implementation manners

[0049] Next, in combination with the accompanying drawings in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] As Figure 1 shown, an embodiment of the present invention provides a positioning system based on radio direction finding, including: a directional signal transmitting module, a signal receiving module, a signal processing module, and a positioning module.

[0051] Signal transmitting module: Construct a directional positioning network including at least three base stations, and the directional positioning network generates directional signals through beamforming technology.

[0052] Deploy at least three base stations in an offshore platform or operation area to construct a directional positioning network. Each base station is configured with a multi-band antenna array. The multi-band antenna array should have high gain and low sidelobe level. The antenna spacing adopts an adjustable structure of 0.5-2λ to meet the requirements of different frequency bands and environments, and at the same time support signal coverage from low frequency (LF) to ultra-high frequency (UHF). The radiation performance and electromagnetic parameters of antennas in each frequency band are independent of each other and do not affect each other. There is sufficient isolation between different frequency bands to avoid signal interference. The low-frequency and high-frequency antenna units can be nested or interleaved to reduce the size of the antenna array and improve the frequency band isolation. The number of antenna units depends on the requirements of positioning accuracy and coverage. A high-precision positioning system needs to configure multiple antenna units to improve signal stability and positioning accuracy. In some applications, the antenna array may include dozens or even more antenna units. Integrate a Beidou / GPS dual-mode timing unit to ensure that the time synchronization accuracy between base stations reaches the nanosecond level, providing a basis for the measurement of time difference of arrival. Real-time calculate the beam weight coefficient through a field programmable gate array (FPGA) chip, and combine the least mean square (LMS) algorithm to suppress sidelobe interference and achieve dynamic beamforming. Specifically, in a low-speed scenario of 0-10 km / h, a narrow beam is used to enhance the signal strength; in a high-speed scenario of 30-100 km / h, it switches to a wide beam to expand the coverage range.

[0053] Meanwhile, connect the base station to the environmental database, link the marine geological database with the Geographic Information System (GIS), and obtain data such as seabed terrain elevation, seawater refractive index, and reflection coefficient of obstacle materials to provide environmental parameter support for the improved reverse ray tracing algorithm. Establish a database of reflection coefficients of urban building materials, including Fresnel reflection attenuation models for materials such as glass, concrete, and metal.

[0054] Specifically, generate a directional signal through beamforming technology. The beamforming technology can adopt digital beamforming (DBF) or analog beamforming technology to achieve flexible beam control. In low-speed scenarios, use narrow-beam directional scanning to improve signal strength and positioning accuracy; in high-speed scenarios, switch to wide-beam fast scanning to expand the coverage range and track moving targets. Calculate the beam weight coefficients in real time based on the FPGA chip, combine the least mean square error algorithm to suppress sidelobe interference, support multi-user spatial division multiplexing, and reduce co-channel interference through orthogonal beam allocation.

[0055] The expression of the least mean square error algorithm is:

[0056] ;

[0057] In the formula, 、 are the weight vectors of the -th and -th iterations respectively; is the step size parameter; is the conjugate of the error signal; is the input signal.

[0058] Signal receiving module: Set on the target device, receive directional signals from different base stations, extract multi-dimensional parameters including signal strength, time difference of arrival, and direction of arrival, and upload the multi-dimensional parameters to the signal processing module.

[0059] Specifically, install a multi-band receiving module on the underwater device or remotely operated underwater vehicle (ROV), support the demodulation of Beidou / GPS signals, and be equipped with a high-dynamic range receiver with a dynamic range ≥ 80 dB to adapt to base station signals of different intensities. The device is built-in with a regular tetrahedron antenna array, constructs a three-dimensional spatial spectrum based on the quaternion MUSIC algorithm, breaks through the estimation limit of the traditional two-dimensional direction of arrival, and realizes sub-degree resolution of azimuth angles from 0° to 360° and elevation angles from -90° to 90°. Integrate a barometer and a gyroscope to obtain the device height and attitude data in real time, where the attitude data includes the pitch angle and the roll angle , which are used for signal strength environmental attenuation compensation.

[0060] Introduce an environmental attenuation compensation factor, combined with the barometer height data of the target device With the gyroscope attitude data , correct the signal strength deviation. The expression of the compensation model is:

[0061] ;

[0062] In the formula, is the environmental attenuation compensation factor; , are the weight coefficients; is the height attenuation term; is the height data measured by the barometer of the target device; is the attitude attenuation term; , are the pitch angle and roll angle respectively output by the gyroscope of the target device.

[0063] Perform weighted fusion on the signal strengths of different frequency bands, and assign weights to the signal strengths of each frequency band according to the signal quality and anti-interference ability:

[0064] Signal quality: Monitor the signal quality of each frequency band in real time, such as signal-to-noise ratio (SNR), signal strength, etc. When the signal quality of a certain frequency band decreases, reduce its weight, and at the same time increase the weights of other frequency bands to ensure that the frequency bands with good signal quality have higher weights in the fusion and make greater contributions to the fusion result.

[0065] Anti-interference ability: The low-frequency band (such as 433 MHz) has strong penetration but low resolution, and the high-frequency band (such as 5.8 GHz) has high resolution but is easily blocked. Therefore, after evaluating the electromagnetic environment of each frequency band, assign higher weights to the frequency bands with less interference to reduce the impact of interference on the fused signal.

[0066] Utilize artificial intelligence and machine learning algorithms, such as neural networks, fuzzy logic, etc., to automatically learn and adjust the weights of each frequency band according to historical data and real-time information to adapt to the complex and changing signal environment.

[0067] The target device receives directional signals from different base stations and extracts the following multi-dimensional parameters:

[0068] Signal strength (RSSI): Use the sliding window average algorithm to denoise, and the window size is adaptively adjusted according to the target moving speed.

[0069] First, according to the initial moving speed of the target device, set an initial window size. For example, if the target device is in a stationary or low-speed moving state, a larger window size can be set. In the embodiment of the present invention, it is set to 100-200 sampling points to obtain a better smoothing effect; if the target device is in a high-speed moving state, a smaller window size is set. In the embodiment of the present invention, it is set to 20-50 sampling points to quickly respond to signal changes.

[0070] The moving speed of the target is monitored in real time through the sensors built in the target device, and the window size is dynamically adjusted according to the moving speed of the target device. When the moving speed of the target increases, the window size is reduced to quickly respond to signal changes; when the moving speed of the target decreases, the window size is increased to better smooth the noise.

[0071] Time Difference of Arrival (TDOA): The measurement accuracy needs to reach the nanosecond level. The generalized cyclic cross-correlation entropy algorithm is used to suppress multipath interference and eliminate the time delay error caused by frequency offset.

[0072] Direction of Arrival (DOA): Based on the regular tetrahedron antenna array to receive signals, subspace projection technology is adopted to separate co-frequency interference and improve the direction resolution in a low signal-to-noise ratio environment. The signal receiving module uses the quaternion MUSIC algorithm to extract the direction of arrival from the directional signals, including the following steps:

[0073] Step 1: Based on the three-dimensional coordinates of each antenna element in the regular tetrahedron antenna array and the incident direction vector of the signal, convert the position vector and the incident direction vector of each antenna element into quaternion form:

[0074] ;

[0075] In the formula, is the quaternion array manifold vector of the th antenna element; is the imaginary unit; is the spacing between antenna elements; is the wavelength of the signal; is the position vector of the th antenna element; is the incident direction vector of the signal. Both the position vector and the incident direction vector are represented by quaternions to more comprehensively describe the signal propagation characteristics in three-dimensional space.

[0076] In three-dimensional space, the azimuth angle usually ranges from 0° to 360°, while the elevation angle generally ranges from -90° to 90°.

[0077] According to the requirements of the angle resolution for actual applications, the ranges of the azimuth angle and the elevation angle are respectively divided into multiple discrete angle points. For example, it can be selected to divide both the azimuth angle and the elevation angle at 1° intervals. In this way, there will be 360 points in the azimuth angle direction and 180 points in the elevation angle direction.

[0078] By combining the divided azimuth angle and elevation angle points, a three-dimensional grid is generated. Specifically, for each azimuth angle and each elevation angle , calculate the corresponding direction of the array manifold matrix , and use these matrices to construct a gridded search space in three-dimensional space.

[0079] Step 2: Calculate the covariance matrix of the quaternion signal to obtain the covariance matrix of the antenna array output.

[0080] Step 3: Perform eigenvalue decomposition on the covariance matrix to obtain the eigenvalues and the corresponding eigenvectors. Divide the eigenvectors into the signal subspace and the noise subspace according to the magnitudes of the eigenvalues.

[0081] Step 4: Use the eigenvectors of the noise subspace to construct a spatial spectrum function, search the spatial spectrum function, find the direction that makes the spatial spectrum function obtain a minimum value, and obtain the direction of arrival of the directional signal.

[0082] Signal processing module: Based on the generalized cyclic cross-correlation entropy function, perform multipath interference suppression on the extracted multi-dimensional parameters, use the reverse ray tracing algorithm to reverse-search the propagation path of the signal, generate a three-dimensional path map, and exclude the non-line-of-sight propagation paths in the three-dimensional path map according to the multi-dimensional parameters after multipath interference suppression to obtain the line-of-sight propagation path parameters of each base station.

[0083] Specifically, use the generalized cyclic cross-correlation entropy function to construct a noise suppression model, and combine Monte Carlo path screening to exclude non-line-of-sight signals, reducing the positioning deviation caused by the multipath effect.

[0084] The expression of the noise suppression model is:

[0085] ;

[0086] In the formula, represents the statistical correlation between signals and at time delay ; is the total number of signal samples; is the Gaussian kernel function; , are the directional signals from different base stations; is the time variable; is the time delay.

[0087] Perform singular value decomposition (SVD) on the entropy matrix, retain the first K principal components to construct an effective signal subspace, and suppress the noise subspace components.

[0088] Through the analysis of the eigenvalue distribution of the entropy matrix, if the ratio of the second largest eigenvalue to the largest eigenvalue exceeds a threshold (such as > 0.3), it is determined as multipath interference, and the attenuation compensation mechanism is triggered.

[0089] The scale (σ) of the kernel function in the generalized cyclic cross - correlation entropy function is dynamically adjusted using a reinforcement learning algorithm. The state space includes environmental noise intensity, historical interference scenario data, and signal quality; the action space is the kernel width adjustment strategy; the reward function synthesizes the improvement in signal processing performance and the improvement in adaptability.

[0090] Among them, the kernel width adjustment strategy is used to adjust the width parameter of the Gaussian kernel function to adapt to different environmental noise intensities. The width parameter ( ) of the Gaussian kernel function needs to be dynamically adjusted according to the environmental noise intensity to achieve a balance between noise suppression and signal retention. Specifically, when the noise statistical characteristics are known, it can be directly derived through the noise variance :

[0091] ;

[0092] In the formula, is the noise variance, reflecting the noise intensity; k is the empirical coefficient, usually taking values from 1 to 3, which needs to be calibrated through measured data.

[0093] If the statistical characteristics of the noise are unknown at this time, the optimal kernel width adjustment strategy is selected according to the current state. According to the dynamic characteristics of the noise and system requirements, common strategies can be divided into the following four categories:

[0094] Empirical threshold method: Preset the mapping relationship between the noise intensity interval and the kernel width .

[0095] Model adaptive strategy: Establish a compensation model by combining sensor data of barometers and gyroscopes;

[0096] Reinforcement learning strategy: Use the improvement in signal - to - noise ratio or edge retention degree as the reward function to train the agent to select the optimal ;

[0097] Multi - scale fusion strategy: Apply multiple - valued Gaussian kernels in parallel and fuse the results with weights.

[0098] The reward function synthesizes the improvement in signal processing performance and the improvement in adaptability. Specifically, when the signal processing performance improves, the reward is measured by increasing the signal - to - noise ratio or reducing the false detection rate: when the adaptability enhances, the agent is rewarded according to its ability to quickly adapt to environmental changes. To achieve a balance between performance improvement and adaptability, the expression of the comprehensive reward function is:

[0099] ;

[0100] In the formula, is the signal processing performance reward; is the adaptability enhancement reward; It is the reward for SNR improvement; It is the reward for false detection rate reduction; It is the processing delay penalty term; It is the reward for parameter adjustment speed; It is the reward for parameter stability; It is the reward for environmental mutation detection.

[0101] The environmental adaptation ability of the agent needs to be comprehensively evaluated through five-dimensional indicators of dynamic response, stability, detection sensitivity, generalization, and resource efficiency, and relies on high-fidelity simulation and real-scenario iterative verification; when actually deployed, 3-4 core indicators are preferably selected to construct the reward function to avoid excessive complexity.

[0102] Create a Q value table for storing the Q values of each state-action pair, where the Q value represents the expected long-term reward for performing a certain action in a certain state. The agent continuously tries different actions through interaction with the environment and updates the Q values in the Q value table, gradually optimizing the decision-making strategy to obtain the optimal kernel width adjustment strategy.

[0103] After calculating the real-time position and motion state of the target device based on the TDOA and DOA reported by the target device, the base station predicts the future position of the target device through Kalman filtering and adjusts the beam weight coefficient of the antenna array in real time to make the main lobe of the beam accurately point to the predicted position.

[0104] The state equation of the Kalman filter model is used to predict the target position at the next moment, and the observation equation corrects the predicted value by combining the target signal received by the base station. Output the optimal estimated target position and motion trend, and update the weight coefficient of the beam through the FPGA chip. The expression of the Kalman filter model is:

[0105] ;

[0106] In the formula, is the Kalman gain matrix; is the prior estimation error covariance matrix; is the observation matrix; is the observation noise covariance matrix; is the posterior estimation error covariance matrix; is the identity matrix.

[0107] When the target is moving at high speed, adjusting the beam direction in advance through prediction can avoid signal loss caused by beam switching delay.

[0108] The improved reverse ray tracing algorithm traces the possible paths of the signal reversely from the receiving end, optimizes the judgment of the ray propagation direction by using the cross-scanning algorithm, and reduces the amount of invalid calculations. Combining GIS data with the Fresnel equation, it calculates the reflection attenuation on the surface of the obstacle and establishes a diffraction wave field strength model. It performs time delay alignment and phase compensation on all effective paths at the receiving point, and constructs a composite channel impulse response model including path loss and Doppler frequency shift.

[0109] The Monte Carlo method is used to randomly generate path samples, and the line-of-sight (LOS) paths are screened through the signal strength threshold and time delay consistency, excluding the non-line-of-sight (NLOS) interference.

[0110] The pre-trained CNN network classifies path features such as path length and angle consistency, outputs the path credibility weights, and combines the Kalman filter to dynamically correct the weights to improve the screening accuracy in complex environments. According to the multi-dimensional parameter quality of each base station, it evaluates the signal reliability, dynamically adjusts the weight coefficients of each base station, and performs weighted fusion of the LOS path parameters and geographical location information of each base station to output the three-dimensional coordinates of the target device.

[0111] Positioning module: According to the multi-dimensional parameter quality of each base station, it evaluates the signal reliability, dynamically adjusts the weight coefficients of each base station, and performs weighted fusion of the LOS path parameters and geographical location information of each base station to output the three-dimensional coordinates of the target device.

[0112] Specifically, the local coordinate systems of each base station are converted to the global coordinate system through the rotation and translation matrix. Weights are assigned to the positioning data of each base station, and the weight coefficients are dynamically adjusted based on the signal quality (SNR, multipath error, clock synchronization deviation) evaluated by the Kalman filter.

[0113] The improved particle swarm optimization algorithm PSO is used to optimize the weight coefficients. The improved PSO algorithm can improve the global search ability and convergence speed by dynamically adjusting the inertia weight. For example, the linear decreasing weight strategy (LDW) or the adaptive inertia weight strategy is adopted, so that the particles can find the global optimal solution faster.

[0114] In the improved PSO algorithm, the weight update formula of the particle can be expressed as:

[0115] ;

[0116] In the formula, 、 are the weights of the i th particle at the t +1 and the t th iteration respectively; 、 are the learning factors, which are taken as 2 in the embodiments of the present invention; 、 is a random number between [0, 1]; is the historical best position of the i th particle; is the global best position; is the i th particle's position in the t th iteration.

[0117] As Figure 2 shown, an embodiment of the present invention further provides a positioning method based on radio direction finding, which is implemented based on the above-mentioned radio direction finding-based positioning system, and includes the following steps:

[0118] S1. Construct a directional positioning network composed of at least three base stations. Each base station is configured with a multi-band antenna array, and a directive signal is generated through beamforming technology.

[0119] S2. The target device receives the directional signals from different base stations and extracts multi-dimensional parameters such as signal strength, time difference of arrival, and direction of arrival.

[0120] S3. Based on the generalized cyclic cross-correlation entropy function, construct a multipath interference suppression model to perform noise suppression processing on the multi-dimensional parameters.

[0121] S4. Adopt an improved reverse ray tracing algorithm, combine with geographic information system data, construct a three-dimensional signal propagation path map, and exclude non-line-of-sight propagation paths.

[0122] S5. Through the dynamic weighted particle swarm optimization algorithm, fuse the positioning data of multiple base stations and output the target three-dimensional coordinates in real time, where the weight coefficient is dynamically adjusted according to the signal quality of the base stations.

[0123] Suppose in the development project of deep-sea oil and gas fields, it is necessary to accurately install and regularly maintain the key equipment of the subsea production system. These equipment include subsea wellheads, pipelines, valves, and sensors, etc., which are distributed in a vast seabed area. At the same time, it is necessary to conduct equipment inspection and repair operations through a remotely operated underwater vehicle (ROV).

[0124] System Deployment

[0125] 1. Base Station Deployment

[0126] Deploy at least three base stations on the ocean platform. Each base station is configured with a multi-band antenna array, and the antenna spacing is adjustable, with a range of 0.5 - 2 . The base station integrates a Beidou / GPS dual-mode timing unit to ensure high-precision time synchronization. The base station generates a directive signal through beamforming technology to cover the entire operation area.

[0127] 2. Target Device

[0128] Install the target device on the underwater equipment and ROV, which has multi-band receiving capabilities, can receive directional signals from the base station, and extract multi-dimensional parameters such as signal strength (RSSI), time difference of arrival (TDOA), and direction of arrival (DOA).

[0129] 3. Signal Processing and Optimization Module

[0130] Deploy an embedded signal processing module on the ocean platform, equipped with an FPGA chip, to achieve real-time calculation of generalized cyclic cross-correlation entropy for suppressing multipath interference. At the same time, combine the dynamic weighted particle swarm optimization algorithm (DPSO) and Kalman filtering to output the three-dimensional coordinates of the target device in real time.

[0131] 4. Three-Dimensional Path Modeling

[0132] The system is connected to the marine geological database, which contains information such as seabed topography, seawater refractive index, and ocean platform structure. Through an improved reverse ray tracing algorithm, combined with GIS data, construct a three-dimensional signal propagation path map to exclude non-line-of-sight propagation paths.

[0133] Implementation Steps

[0134] 1. Base Station Initialization and Calibration

[0135] After deployment, initialize and calibrate the base station to ensure the beamforming performance of the antenna array and the synchronization accuracy of the timing unit. Verify the communication link and signal coverage between base stations through test signals.

[0136] 2. Target Device Positioning

[0137] When the ROV or underwater equipment starts operating, the target device receives directional signals from multiple base stations and extracts RSSI, TDOA, and DOA parameters. The signal processing module suppresses multipath interference through the generalized cyclic cross-correlation entropy function, and fuses multi-base station data through the dynamic weighted particle swarm optimization algorithm to output the three-dimensional coordinates of the target in real time.

[0138] 3. Path Optimization and Real-Time Adjustment

[0139] The system dynamically optimizes signal processing parameters through an adaptive kernel width adjustment unit according to the real-time monitored environmental noise intensity. The Kalman filtering module dynamically adjusts the weight coefficient according to the signal quality to ensure positioning accuracy. At the same time, through GIS data and the reverse ray tracing algorithm, the three-dimensional signal propagation path map is updated in real time to exclude the interference of non-line-of-sight paths.

[0140] 4. Operation Monitoring and Feedback

[0141] In the control center of the offshore platform, the three-dimensional coordinates and motion trajectories of the target devices are monitored in real time. The operator adjusts the operation path of the ROV through the visualization interface to ensure its accurate arrival at the target position. Meanwhile, the system records the signal data and positioning accuracy during the operation process for subsequent performance evaluation and optimization.

[0142] A positioning system and method based on radio direction finding provided by an embodiment of the present invention can achieve centimeter-level positioning accuracy in a complex marine environment through multi-parameter fusion and dynamic optimization algorithms, providing reliable technical support for precise operations in ocean engineering. By adjusting the beamforming parameters and the scale of the signal processing kernel function in real time to adapt to the dynamic changes in the marine environment. For example, in response to complex situations such as signal refraction and electromagnetic interference caused by seawater flow, the system can quickly respond, optimize the signal transmission path and processing method, and ensure the stability and accuracy of the positioning signal. By enhancing the signal transmission quality and reliability, the system enables the remotely operated underwater vehicle (ROV) and underwater devices to reach the target position quickly and accurately. This not only reduces the operation time and cost but also improves the operation safety and reliability, providing a strong guarantee for the efficient execution of ocean engineering. The generalized cyclic cross-correlation entropy function and the multipath suppression model effectively reduce the influence of complex reflection and refraction signals in the marine environment on the positioning accuracy. The system can identify and suppress multipath interference, remove the influence of noise on the positioning parameters, and thus improve the quality of the positioning data. The multi-band antenna array can work simultaneously in multiple frequency bands to adapt to different signal environments and application requirements. The beamforming technology focuses the signal energy in a specific direction, enhancing the signal strength in the target area and expanding the signal coverage range.

[0143] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity in description, not all possible combinations of the various technical features in the above embodiments are described. Only the preferred embodiments of the present invention are expressed. The description is relatively specific and detailed, but it should not be construed as a limitation to the scope of the present invention. As long as the combinations of these technical features do not conflict, they should be considered as falling within the scope described in this specification.

[0144] It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several deformations and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the appended claims.

Claims

1. A positioning system based on radio direction finding, characterized in that, Including: A directional signal transmission module, a signal reception module, a signal processing module, and a positioning module; The signal transmission module: constructs a directional positioning network including at least three base stations, and the directional positioning network generates directional signals through beamforming technology; The signal reception module: is arranged on the target device, receives directional signals from different base stations, extracts multi-dimensional parameters, and uploads the multi-dimensional parameters to the signal processing module; The signal processing module: suppresses multipath interference for the extracted multi-dimensional parameters based on the generalized cyclic cross-correlation entropy function, uses the reverse ray tracing algorithm to inversely search for the signal propagation path, and excludes non-line-of-sight propagation paths in the propagation path according to the multi-dimensional parameters after multipath interference suppression to obtain the line-of-sight propagation path parameters of each base station; The suppressing of multipath interference for the extracted multi-dimensional parameters based on the generalized cyclic cross-correlation entropy function includes the following steps: S1: Conducts cyclic spectral density analysis on the received multi-dimensional parameters, extracts the cyclostationary characteristics of the received signal, and separates interference components with different cyclic frequencies; S2: Maps the multi-dimensional parameters to the joint feature space, constructs a three-dimensional matrix as the input of the generalized cyclic cross-correlation entropy function, and the generalized cyclic cross-correlation entropy function outputs the entropy matrix of the multi-dimensional parameters; S3: Uses a reinforcement learning algorithm to adjust the kernel width of the kernel function of the generalized cyclic cross-correlation entropy function; S4: Performs singular value decomposition on the entropy matrix, retains the first K principal components to construct a signal subspace, and suppresses the components corresponding to the noise subspace; S5: Analyzes the eigenvalue distribution of the entropy matrix. When the ratio of the second largest eigenvalue to the largest eigenvalue is greater than the set threshold, it is determined as multipath interference, and the multipath interference in the multi-dimensional parameters is further reduced through an attenuation compensation mechanism; The positioning module: evaluates the signal reliability according to the multi-dimensional parameter quality of each base station, dynamically adjusts the weight coefficients of each base station, and performs weighted fusion on the line-of-sight path parameters and geographical location information of each base station to output the three-dimensional coordinates of the target device.

2. The positioning system based on radio direction finding according to claim 1, characterized in that: In the signal transmission module, after generating the directional signal, the side lobe interference of each beam is suppressed by the least mean square error algorithm, and the weights of each beam are updated. The expression of the least mean square error algorithm is: W(k + 1) = w(k) + μe * (k)X(k); where \(w(k)\) and \(W(k + 1)\) are the weight vectors of the \(k\)-th and \((k + 1)\)-th iterations respectively; \(\mu\) is the step size parameter; \(e * (k)\) is the conjugate of the error signal; \(X(k)\) is the input signal.

3. A positioning system based on radio direction finding according to claim 1, characterized in that: After the signal reception module extracts the signal strength, it uses a sliding window averaging algorithm to denoise the signal strength, generates a smoothed signal strength by calculating the average value of the signal strength values within the window, and the size of the sliding window is adaptively adjusted according to the moving speed of the target device.

4. A positioning system based on radio direction finding according to claim 1, characterized in that: An environmental attenuation compensation factor is introduced in the signal reception module to correct the errors in the signal strength caused by height changes and device postures. The expression of the environmental attenuation compensation factor is: C env = α·f(h) + β·g(θ, φ); where C env is the environmental attenuation compensation factor; α and β are weight coefficients; f(h) is the height attenuation term; h is the height data measured by the barometer of the target device; g(θ, φ) is the attitude attenuation term; θ and φ are the pitch angle and roll angle output by the gyroscope of the target device, respectively.

5. A positioning system based on radio direction finding according to claim 1, characterized in that: The target device receives directional signals from the base station through a regular tetrahedron antenna array. The signal reception module uses the quaternion MUSIC algorithm to extract the direction of arrival from the directional signals, including the following steps: Step 1: Based on the three-dimensional coordinates of each antenna element in the regular tetrahedron antenna array and the incident direction vector of the signal, convert the position vector of each antenna element and the incident direction vector of the signal into quaternion form: where a m (φ,θ) is the quaternion array manifold vector of the m-th antenna element; j is the imaginary unit; d is the spacing between antenna elements; λ is the wavelength of the signal; P m is the position vector of the m-th antenna element; u(φ,θ) is the incident direction vector of the signal; Step 2: Calculates the covariance matrix of the quaternion signal to obtain the covariance matrix output by the antenna array; Step 3: Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues and corresponding eigenvectors, and divide the eigenvectors into a signal subspace and a noise subspace according to the magnitudes of the eigenvalues; Step 4: Use the eigenvectors of the noise subspace to construct a spatial spectrum function, search the spatial spectrum function, find the direction that makes the spatial spectrum function obtain a minimum value, and obtain the direction of arrival of the directional signal.

6. The positioning system based on radio direction finding according to claim 1, characterized in that: In the signal processing module, after receiving the multi-dimensional parameters uploaded by the target device, the direction-of-arrival angle is predicted according to the moving trajectory of the target device, and the weight coefficients of each beam are dynamically adjusted by using Kalman filtering: where Q(q) is the Kalman gain matrix; P(q|q - 1) is the prior estimation error covariance matrix; H is the observation matrix; R is the observation noise covariance matrix; P(q|q) is the posterior estimation error covariance matrix; I is the identity matrix.

7. A positioning system based on radio direction finding according to claim 1, characterized in that: The expression of the generalized cyclic cross-correlation entropy function described in S2 is: where C XY (τ) represents the statistical correlation between the signals X(t n ) and Y(t n ) at the time delay τ; N is the total number of signal samplings; K is the Gaussian kernel function; X(t n ) and Y(t n ) are the directional signals from different base stations at time t n ; t n is the time variable; τ is the time delay.

8. A positioning system based on radio direction finding according to claim 1, characterized in that: S3 includes the following steps: S31: Take the current ambient noise intensity, historical interference scenario data, and signal quality as the state space, and the kernel width adjustment rule as the action space, and calculate the reward value according to the signal processing performance and adaptability change of the agent after performing the action. The expression for calculating the reward value is: R total = R performance + R adaptability =(R SNR + R F-error + R latency ) + (R speed + R stable + R detect ); where R performance is the signal processing performance reward; R adaptability is the adaptability enhancement reward; R SNR is the signal-to-noise ratio improvement reward; R F-error is the missed detection rate reduction reward; R latency is the processing delay penalty term; R speed is the parameter adjustment speed reward; R stable is the parameter stability reward; R detect is the environmental mutation detection reward; S32: Create a Q-value table for storing the Q-values of each state-action pair, where the Q-value represents the expected long-term reward for performing a certain action in a certain state; S33: Through the interaction between the agent and the environment, continuously try different actions, update the Q-values in the Q-value table according to the obtained rewards, gradually optimize the decision-making strategy, and obtain the optimal kernel width adjustment strategy.

9. The positioning system based on radio direction finding according to claim 1, characterized in that: The signal processing module excludes the non-line-of-sight propagation paths in the propagation path according to the multi-dimensional parameters after multi-path interference suppression, including the following steps: calculate the signal propagation time difference through the time difference of arrival after multi-path interference suppression, combine the position information of each base station, and reverse-derive the signal path length difference to exclude the non-line-of-sight propagation paths; use the direction of arrival after multi-path interference suppression to determine the signal arrival direction, optimize the initial search direction of ray tracing, combine the geographical information system data of the search area to predict the main propagation path, and exclude the non-line-of-sight propagation paths; evaluate the path loss based on the signal strength after multi-path interference suppression, combine the Fresnel equation and the material reflection coefficient database to calculate the energy attenuation of different material surfaces, and verify whether the path conforms to the line-of-sight propagation attenuation characteristics to exclude the non-line-of-sight propagation paths.

Images