Beidou terminal capable of intelligently evaluating multiple states

Abstract

The invention relates to the technical field of satellite navigation terminals, and discloses a Beidou terminal capable of intelligently evaluating multiple states. Comprising a Beidou positioning module used for acquiring a Beidou signal corresponding to the Beidou terminal; the control module is configured to execute the following steps: performing time sequence modeling on the propagation time delay, the geometric distribution factor, the ionosphere influence parameter and the troposphere influence parameter of the Beidou signal to generate a positioning reliability index so as to realize positioning state evaluation; performing conjoint analysis on the Beidou signal, and outputting a data credibility grade; processing the inertial navigation data, the terrain elevation data, the sky satellite geometric distribution factor and the multi-terminal altitude difference data corresponding to the Beidou terminal through a convolutional neural network to generate a three-dimensional environment interference coefficient; and fusing the channel fading parameter, the network topology data, the communication antenna signal-to-noise ratio, the transmission rate change parameter and the data integrity parameter corresponding to the Beidou terminal, and calculating a communication link stable value to realize communication state evaluation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite navigation terminal, and particularly relates to a Beidou terminal with intelligent evaluation of multiple states. BACKGROUND

[0002] In satellite navigation applications, the positioning reliability of a Beidou terminal is affected by multiple factors such as signal propagation environment, data quality, carrier motion state, and communication link stability. The traditional Beidou terminal evaluation method has the following defects: 1. Single evaluation dimension: Existing technologies mostly only evaluate single parameters such as positioning accuracy (e.g. geometric distribution factor PDOP) or signal quality (e.g. carrier-to-noise ratio), and do not form a multi-dimensional joint evaluation system for positioning state, data quality, use environment, and communication state, making it difficult to fully reflect the actual operation reliability of the terminal.

[0003] 2. Insufficient environmental adaptability: For complex terrain (such as mountainous areas, urban canyons) or dynamic carrier (such as vehicle-mounted, ship-mounted) scenarios, there is a lack of fusion analysis of three-dimensional space environmental interference (such as terrain obstruction, multi-terminal signal mutual interference) and inertial navigation data, resulting in insufficient environmental influence evaluation accuracy.

[0004] 3. Lack of weight adjustment mechanism: Traditional evaluation models mostly use fixed weight fusion parameters, which cannot dynamically optimize the importance of each evaluation index according to the real-time running state of the terminal (such as signal mutation, communication link fluctuation), making it difficult to adapt to evaluation needs in multiple scenarios.

[0005] Therefore, there is an urgent need for a Beidou terminal to solve at least one of the above problems. SUMMARY

[0006] The present application provides a Beidou terminal with intelligent evaluation of multiple states, aiming to solve the problem in the prior art that there is no scheme to systematically fuse positioning reliability index, data confidence level, three-dimensional environmental interference coefficient, and communication link stability value, and to realize weight self-adaptive adjustment by combining real-time data and historical data comparison analysis.

[0007] In a first aspect, the embodiments of the present application provide a Beidou terminal with intelligent evaluation of multiple states, comprising: a Beidou positioning module configured to obtain a Beidou signal corresponding to the Beidou terminal; a control module configured to perform: time series modeling on the propagation delay, geometric distribution factor, ionospheric influence parameter, and tropospheric influence parameter of the Beidou signal to generate a positioning reliability index, thereby realizing positioning state evaluation; The pseudorange residual, carrier-to-noise ratio, multipath effect parameters, number of satellite signals, data integrity parameters, and cycle slip parameters corresponding to the BeiDou signal are jointly analyzed to output the data credibility level, so as to achieve data quality assessment. By processing the inertial navigation data, terrain elevation data, sky satellite geometric distribution factor and multi-terminal elevation difference data corresponding to the Beidou terminal through a convolutional neural network, a three-dimensional environmental interference coefficient is generated to achieve environmental assessment. By integrating the channel fading parameters, network topology data, communication antenna signal-to-noise ratio, transmission rate variation parameters, and data integrity parameters corresponding to the BeiDou terminal, the stability value of the communication link is calculated to achieve communication status assessment. Based on the positioning reliability index, data credibility level, three-dimensional environmental interference coefficient, and communication link stability value, a real-time evaluation result of positioning credibility is generated. Based on the comparative analysis of the real-time operation data and historical operation data corresponding to the Beidou terminal, the weights of the positioning reliability index, data credibility level, three-dimensional environmental interference coefficient and communication link stability value are adaptively adjusted to update the real-time evaluation results.

[0008] In some embodiments, obtaining the BeiDou signal corresponding to the BeiDou terminal includes: receiving BeiDou B1, B2 and B3 frequency signals through a multi-band antenna, and preprocessing the signals through an anti-multipath filter and an adaptive gain control circuit to suppress reflected signal interference and optimize signal strength.

[0009] In some embodiments, the step of performing time-series modeling of the propagation delay, geometric distribution factor, ionospheric influence parameters, and tropospheric influence parameters of the BeiDou signal to generate a positioning reliability index includes: establishing a time-series model based on ARIMA or LSTM, inputting time-series data containing ionospheric puncture point delay, tropospheric zenith delay correction value, and geometric distribution factor, and obtaining a reliability index representing the error fluctuation trend through the training output corresponding to the time-series model.

[0010] In some embodiments, the joint analysis of the pseudorange residual, carrier-to-noise ratio, multipath effect parameters, number of satellite signals, data integrity parameters, and cycle slip parameters corresponding to the BeiDou signal, and the output of the data credibility level, includes: normalizing the parameters and using them as input features for the random forest algorithm, and outputting a credibility level containing multi-level data through a trained classifier, wherein the multipath effect parameters are calculated by the amplitude difference and phase difference between the direct path and the reflection path of the signal.

[0011] In some embodiments, the generating the three-dimensional environmental interference coefficient by processing the inertial navigation data, terrain elevation data, sky satellite geometric distribution factor and multi-terminal height difference data corresponding to the Beidou terminal through the convolutional neural network comprises: splicing the acceleration and angular velocity data of the inertial navigation, terrain elevation grid data, satellite azimuth or elevation angle distribution matrix and multi-terminal altitude difference tensor into a three-dimensional input tensor, extracting environmental occlusion, terrain undulation and multi-terminal signal mutual interference characteristics through a convolutional neural network containing a residual block, and outputting an interference coefficient in the interval of 0-1.

[0012] In some embodiments, the fusing the channel fading parameter, network topology data, communication antenna signal-to-noise ratio, transmission rate change parameter and data integrity parameter corresponding to the Beidou terminal to calculate the communication link stability value comprises: constructing a Bayesian network model, defining the channel fading type, node connectivity, signal-to-noise ratio threshold and rate fluctuation coefficient as network nodes, quantifying the inter-node dependency relationship through a conditional probability table, and outputting a communication link stability value based on posterior probability.

[0013] In some embodiments, the generating the real-time evaluation result of the positioning credibility according to the positioning reliability index, data credibility level, three-dimensional environmental interference coefficient and communication link stability value comprises: using an analytic hierarchy process or a dynamic weighted fusion model to perform normalized weighting calculation on the four parameters, wherein the initial weight value is determined through historical data training, and a positioning credibility evaluation vector containing a confidence probability is output.

[0014] In some embodiments, the comparative analysis of the real-time running data and the historical running data corresponding to the Beidou terminal comprises: performing sliding window filtering and outlier detection on the real-time data and the historical data, calculating the deviation degree of the real-time value and the historical mean value of the characteristic parameters such as positioning error mean square error, data update rate and communication packet loss rate, and generating a parameter fluctuation warning signal.

[0015] In some embodiments, the self-adaptive adjustment of the weights of the positioning reliability index, data credibility level, three-dimensional environmental interference coefficient and communication link stability value comprises: through a gradient descent algorithm or a reinforcement learning model, taking the minimization of the deviation between the positioning result and the reference station data as an objective function, dynamically optimizing the weight coefficients of each parameter, and forming a self-adaptive weight adjustment strategy.

[0016] In some embodiments, further comprising: an inertial navigation fusion module integrated with a MEMS accelerometer and a MEMS gyroscope for outputting three-axis acceleration and angular velocity data of the carrier; the control module performs loose coupling or tight coupling fusion of the position and speed information output by the Beidou positioning module and the inertial measurement data output by the inertial navigation fusion module through an extended Kalman filter or an unscented Kalman filter algorithm to generate fusion positioning data; and during the generation of the three-dimensional environmental interference coefficient, an environmental occlusion feature extraction is further performed on the attitude angle of the fusion positioning data to enhance the use environment evaluation accuracy in a complex dynamic environment.

[0017] The application solves the one-sidedness of traditional single-index evaluation by modeling and analyzing parameters in four dimensions of positioning state, data quality, use environment and communication state, and builds a comprehensive evaluation system covering signal propagation, data features, environmental interference and communication stability. By introducing convolutional neural network processing of inertial navigation data, terrain elevation data and multi-terminal elevation difference data, the quantification evaluation of three-dimensional spatial environmental interference is realized, which significantly improves the environmental feature extraction accuracy in complex terrain or dynamic carrier scenarios. Through comparative analysis based on real-time data and historical data, the evaluation parameter weight is adaptively adjusted by the gradient descent algorithm, so that the terminal can dynamically optimize the evaluation model according to the current running state, effectively improving the accuracy and robustness of positioning reliability evaluation in different scenarios. By fusing multi-source data and intelligent algorithms (time series modeling, random forest, Bayesian network, etc.), full-link state monitoring from the signal layer to the application layer is realized, providing multi-level protection for the reliability of terminal positioning results, and being suitable for high-precision navigation, automatic driving and other scenes with high reliability requirements.

[0018] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0020] Figure 1 is a structure schematic diagram of a Beidou terminal for multi-state intelligent evaluation based on self-focusing fiber coupling scheme provided by an embodiment of the present application.

[0021] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work belong to the scope of protection of the present application.

[0023] The flowcharts shown in the drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor are they necessarily executed in the described order. For example, some operations / steps can be further decomposed, combined or partially merged, so that the actual execution order can be changed according to the actual situation.

[0024] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present application, in the embodiments of the present application, the terms "first", "second", etc. are used to distinguish the same items or similar items with basically the same functions and effects. Those skilled in the art can understand that the terms "first", "second", etc. do not limit the quantity and execution order, and the terms "first", "second", etc. also do not necessarily mean that they are different.

[0025] It should be understood that the terms used in the present application are only for the purpose of describing specific embodiments and do not intend to limit the present application. As used in the present application specification and the appended claims, unless otherwise clear from the context, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0026] It should also be understood that the term "and / or" used in the present application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0027] Some embodiments of the present application will be described in detail below in combination with the drawings. The following embodiments and features in the embodiments can be combined with each other without conflict.

[0028] In satellite navigation applications, the positioning reliability of Beidou terminals is affected by multiple factors such as signal propagation environment, data quality, carrier motion state and communication link stability. The traditional Beidou terminal evaluation method has the following defects: 1. Single evaluation dimension: existing technologies mostly only evaluate single parameters such as positioning accuracy (such as geometric distribution factor PDOP) or signal quality (such as carrier-to-noise ratio), and do not form a multi-dimensional joint evaluation system for positioning state, data quality, use environment and communication state, making it difficult to fully reflect the actual operation reliability of the terminal.

[0029] 2. Insufficient environmental adaptability: For complex terrain (such as mountainous areas, urban canyons) or dynamic carrier (such as vehicle-mounted, ship-mounted) scenarios, the fusion analysis of three-dimensional space environment interference (such as terrain shielding, multi-terminal signal mutual interference) and inertial navigation data is lacking, resulting in insufficient accuracy of environmental impact assessment.

[0030] 3. Lack of weight adjustment mechanism: Traditional evaluation models mostly use fixed weight fusion parameters, which cannot dynamically optimize the importance of each evaluation indicator according to the real-time running state of the terminal (such as signal mutation, communication link fluctuation), and are difficult to adapt to evaluation needs in multiple scenarios.

[0031] In the prior art, there is no scheme to systematically fuse the positioning reliability index, data confidence level, three-dimensional environmental interference coefficient, and communication link stability value, and to realize weight self-adaptive adjustment combined with real-time data and historical data comparison analysis. Therefore, there is an urgent need for an intelligent evaluation method that can integrate multi-dimensional state parameters and has dynamic optimization capability to improve the application reliability of Beidou terminals in complex scenarios.

[0032] To solve the above problems, please refer to Figure 1 The Beidou terminal for multi-state intelligent evaluation provided in the embodiments of the present application includes a Beidou positioning module 10 for obtaining a Beidou signal corresponding to the Beidou terminal; a control module 20 configured to perform: time series modeling on the propagation delay, geometric distribution factor, ionospheric influence parameter and tropospheric influence parameter of the Beidou signal, generating a positioning reliability index to realize positioning state evaluation; joint analysis of the pseudorange residual, carrier-to-noise ratio, multipath effect parameter, number of satellite signals, data integrity parameter and cycle slip parameter corresponding to the Beidou signal, outputting a data confidence level to realize data quality evaluation; processing the inertial navigation data, terrain elevation data, sky satellite geometric distribution factor and multi-terminal height difference data corresponding to the Beidou terminal through a convolutional neural network, generating a three-dimensional environmental interference coefficient to realize usage environment evaluation; fusing the channel fading parameter, network topology data, communication antenna signal-to-noise ratio, transmission rate change parameter and data integrity parameter corresponding to the Beidou terminal, calculating a communication link stability value to realize communication state evaluation; generating a real-time evaluation result of positioning confidence according to the positioning reliability index, data confidence level, three-dimensional environmental interference coefficient and communication link stability value; according to the comparison analysis of real-time running data and historical running data corresponding to the Beidou terminal, self-adaptively adjusting the weights of the positioning reliability index, data confidence level, three-dimensional environmental interference coefficient and communication link stability value, and updating the real-time evaluation result.

[0033] Specifically, the Beidou terminal proposed in the application constructs an intelligent evaluation system covering positioning state, data quality, use environment and communication state through multi-dimensional state parameter fusion and dynamic weight adaptive mechanism, solving the defects of single dimension, poor environmental adaptability and fixed weight of traditional methods.

[0034] Positioning state evaluation: positioning reliability index generation includes: input parameters: Beidou signal propagation delay, geometric distribution factor (PDOP), ionospheric influence parameters (such as ionospheric delay correction value), tropospheric influence parameters (such as tropospheric delay model parameters). Processing method: time series analysis is performed on the above parameters by using time series modeling technology (such as ARIMA, LSTM neural network) to capture the dynamic change law of signal propagation. A positioning reliability model is constructed to calculate the positioning reliability index by comprehensively considering the time series characteristics (such as trend item, periodic fluctuation, abnormal mutation) of each parameter, reflecting the real-time state of the positioning signal affected by environmental interference (the higher the index, the stronger the reliability).

[0035] Data quality evaluation: data reliability level output includes: input parameters: pseudorange residual (reflecting measurement error), carrier-to-noise ratio (C / N0, signal strength index), multipath effect parameters (such as multipath delay error), number of satellite signals, data integrity parameters (such as parity check result), cycle slip parameters (carrier phase jump number). Processing method: a multi-parameter joint analysis model is established, and the data quality is comprehensively evaluated by statistical analysis (such as Bayesian inference, fuzzy logic) or machine learning algorithm (such as random forest).

[0036] Data quality is divided into multiple reliability levels (such as high, medium and low), quantifying the availability and accuracy of data. Use environment evaluation: three-dimensional environmental interference coefficient generation includes: input parameters: inertial navigation data (acceleration, angular velocity, reflecting the motion state of the carrier), terrain elevation data (such as DEM digital elevation model), satellite geometric distribution factor in the sky (such as satellite elevation angle, azimuth angle), multi-terminal height difference data (altitude difference between adjacent terminals, evaluating signal mutual interference risk). Processing method: convolutional neural network (CNN) is used to process three-dimensional spatial features, such as terrain occlusion (three-dimensional grid is constructed by elevation data), satellite visibility analysis (combined with elevation angle and terrain occlusion model), and multi-terminal signal mutual interference (based on height difference and signal radiation model). Output three-dimensional environmental interference coefficient to quantify the degree of environmental interference in complex terrain (mountainous area, urban canyon) or dynamic carrier (vehicle-mounted, ship-mounted) scenarios (the higher the coefficient, the stronger the interference).

[0037] Communication state evaluation: Communication link stability value calculation includes: input parameters: channel fading parameters (such as Rayleigh fading, Rice fading coefficient), network topology data (node connection relationship), communication antenna signal-to-noise ratio (reflecting the quality of received signal), transmission rate change parameter (bandwidth fluctuation), data integrity parameter (such as bit error rate). Processing method: fuse the above parameters to construct the communication link stability model, use analytic hierarchy process (AHP) or dynamic weighting algorithm to calculate the communication link stability value, reflect the reliability of data transmission (the higher the value, the more stable the link).

[0038] Dynamic weight adaptive adjustment mechanism includes: core logic: compare real-time running data (current time each dimension parameter) with historical data (long-term statistical normal / abnormal state characteristics), identify terminal running scene (such as static / dynamic, open / sheltered environment). Adjustment method: based on real-time state (such as signal mutation, link fluctuation), dynamically adjust the weight of each evaluation dimension through adaptive algorithm (such as particle swarm optimization, reinforcement learning), for example: increase the three-dimensional environmental interference coefficient weight in urban canyon scene, increase the link stability value weight when communication interruption. The weight adjustment period matches the terminal running state change frequency, ensuring that the evaluation model adapts to the scene difference in real time.

[0039] For example, data collection and preprocessing includes: Beidou positioning module: real-time receive Beidou satellite signal, extract propagation delay, pseudo-range, carrier phase, satellite ephemeris and other original data, synchronously collect motion data of inertial navigation module (such as MEMS inertial sensor). External data access: obtain terrain elevation data (loaded in real time through local storage or network), network state parameters of communication module (such as signal-to-noise ratio, transmission rate), integrate historical running data (stored in terminal or cloud database). Preprocessing: denoising (such as Kalman filtering) and format unification processing are performed on the original data to generate input feature vectors of each evaluation dimension.

[0040] Multi-dimensional evaluation execution includes: positioning reliability index calculation: time series model is established for propagation delay, PDOP, ionosphere / troposphere parameters, short-term change trend is predicted, and reliability index is calculated combined with current value (such as through normalized weighted summation). Data confidence level analysis: construct multi-parameter correlation matrix, set threshold rule or train classification model, output data confidence level (such as distinguish high / low quality data through support vector machine SVM).

[0041] Three-dimensional environmental interference coefficient generation includes: input inertial navigation data (carrier attitude), terrain elevation (rasterization), satellite distribution (polar coordinate coding) into CNN, extract spatial features through convolution layer, output interference coefficient through full connection layer. Communication link stability value calculation: establish state space model of communication parameters, dynamically fuse channel fading, rate change and other parameters, use exponential smoothing method or state transition matrix to calculate stability value.

[0042] The comprehensive evaluation and result output includes: real-time evaluation fusion: based on the current weight, the positioning reliability index, the data reliability level (after numerical), the three-dimensional environment interference coefficient, and the communication link stability value are fused through a linear or nonlinear model (such as weighted average, neural network) to generate a real-time evaluation result of positioning reliability (such as 0-100 points, the higher the score, the stronger the reliability). Result application: output to the terminal display interface, or as a feedback signal to optimize the positioning algorithm (such as dynamically adjusting the filter parameters), while uploading to the cloud for long-term performance analysis. The weight self-adaptive adjustment process includes: state monitoring: real-time monitoring of the mutation of each dimension parameter (such as sudden drop of carrier-to-noise ratio, PDOP exceeding limit), to determine whether to trigger the weight adjustment mechanism. Historical data comparison: extract the current scene characteristics (such as carrier speed, terrain type), match the optimal weight combination of similar scenes in the historical database (based on the principle of minimizing historical evaluation error). If the deviation between real-time data and historical mean exceeds the threshold, start the online optimization algorithm (such as gradient descent) to adjust the weight until the evaluation result and the actual positioning effect are consistent. Weight update: dynamically update the weight matrix of each dimension to ensure that the latest weight is used for the next evaluation, forming a "evaluation-feedback-optimization" closed loop.

[0043] The application breaks through the limitation of traditional single parameter evaluation, systematically fuses positioning, data, environment, and communication four dimensions, and comprehensively reflects the terminal reliability. Through CNN processing three-dimensional space data and inertial navigation information, the interference influence in complex terrain and dynamic carrier scene is accurately quantified. Based on real-time state and historical data comparison, the weight is self-adaptively adjusted to improve the generalization ability of the model in multiple scenes. Through technology fusion and closed loop optimization, the evaluation accuracy and application reliability of Beidou terminal in complex environment are significantly improved, which provides intelligent support for high-precision positioning, navigation and communication.

[0044] In some embodiments, the Beidou signal corresponding to the Beidou terminal is acquired by receiving Beidou B1, B2 and B3 frequency point signals through a multi-band antenna, and pre-processing the signals through an anti-multipath filter and an adaptive gain control circuit to suppress reflected signal interference and optimize signal strength.

[0045] Through Beidou signal receiving and preprocessing technology, the signal quality is improved through multi-band antenna and anti-interference circuit.

[0046] Multi-band signal reception: Use a multi-band antenna that supports Beidou B1 (1561.098 MHz), B2 (1268.52 MHz), and B3 (1268.52 MHz) frequencies to simultaneously receive signals of different frequencies to enhance signal redundancy. Anti-multipath filtering: Deploy anti-multipath filters (such as narrowband filters, RAKE receivers) in the RF front end to separate effective signals by the time delay difference (Δτ) between direct and reflected paths, and suppress multipath interference (reduce bit error rate by more than 30%) caused by building / terrain reflections. Adaptive gain control: Through variable gain amplifiers (VGA) and automatic gain control (AGC) circuits, real-time monitoring of signal power (range: -130dBm ~ -100dBm), dynamic adjustment of amplification factor (gain range 0-60dB), to avoid signal saturation or excessive noise, optimize carrier-to-noise ratio (C / N0 improvement 5-10dB).

[0047] In some embodiments, the propagation time delay, geometric distribution factor, ionospheric influence parameter and tropospheric influence parameter of the Beidou signal are modeled in time sequence to generate a positioning reliability index, comprising: establishing an ARIMA or LSTM based time sequence model, inputting time sequence data containing ionospheric piercing point time delay, tropospheric zenith delay correction value and geometric distribution factor, and obtaining a reliability index representing error fluctuation trend through the training output of the corresponding time sequence model.

[0048] Through the positioning reliability index calculation based on the time sequence model, the error fluctuation trend is captured.

[0049] Input parameters: Ionospheric piercing point time delay (TIP, unit: ns): Calculate the ionospheric delay correction value through dual-frequency observation (B1 / B2); Tropospheric zenith delay correction value (ZTD, unit: mm): calculated using Saastamoinen model or NMF model; Geometric distribution factor (PDOP): calculated based on the three-dimensional coordinates of visible satellites.

[0050] Time sequence modeling includes: ARIMA model: identify the autocorrelation of data (ACF / PACF analysis), determine the difference order (d) and model order (p, q), fit the historical 10-minute data to predict the future 5-minute error trend; LSTM neural network: build a 3-layer LSTM layer (128 neurons per layer), input the previous 30 time steps (1 second / step) data, and output the current time reliability index (range 0-1, higher value indicates smaller error fluctuation). Output features: index fusion of short-term fluctuation (ARIMA residual) and long-term trend (LSTM hidden state), representing the stability of positioning error.

[0051] In some embodiments, the joint analysis of the pseudo-range residual, carrier-to-noise ratio, multipath effect parameter, satellite signal number, data integrity parameter, and cycle slip parameter corresponding to the Beidou signal outputs a data credibility level, including: after normalizing the parameters, the normalized parameters are used as input features of a random forest algorithm, and a trained classifier outputs a credibility level containing multiple levels of data, wherein the multipath effect parameter is obtained by calculating the amplitude difference and phase difference between the direct path and reflected path of the signal.

[0052] Through random forest-based multi-parameter data credibility level classification, the data quality is quantified.

[0053] The parameter preprocessing includes: pseudo-range residual (Δρ, unit: m), carrier-to-noise ratio (C / N0, dB-Hz), cycle slip number (CycleSlip) and other parameters are normalized to the [0, 1] interval (formula: (x-μ) / σ, μ is the mean, σ is the standard deviation); the multipath effect parameter (MP) is calculated: quantified by the correlation peak amplitude difference (|A_direct - A_reflected|) and phase difference (|φ_direct - φ_reflected|), MP = 1 - (A_direct / (A_direct + A_reflected)) (the larger the value, the stronger the multipath interference).

[0054] The random forest model inputs a 6-dimensional feature vector (pseudo-range residual, C / N0, MP, satellite number, data integrity flag, cycle slip number), the training set contains 100,000 groups of normal / abnormal data (label: high / medium / low credibility); 50 decision trees are constructed, the Gini coefficient is used to select the split features, and the soft classification results are output (such as high credibility probability 0.85, medium 0.13, and low 0.02), and the highest probability level is taken as the final output.

[0055] In some embodiments, the three-dimensional environmental interference coefficient is generated by processing the inertial navigation data, terrain elevation data, sky satellite geometric distribution factor, and multi-terminal height difference data corresponding to the Beidou terminal through a convolutional neural network, including: the acceleration and angular velocity data of the inertial navigation, the terrain elevation raster data, the satellite azimuth or elevation angle distribution matrix, and the multi-terminal altitude difference tensor are spliced into a three-dimensional input tensor, the environmental occlusion, terrain undulation, and multi-terminal signal mutual interference features are extracted through a convolutional neural network containing a residual block, and the interference coefficient in the interval [0, 1] is output.

[0056] Through residual CNN-based three-dimensional environmental interference coefficient extraction, inertial and terrain data are fused.

[0057] The input tensor construction includes: inertial data: three-axis acceleration (m / s 2), time window length 10s, forming a 10x6 time sequence matrix; terrain elevation: DEM data within 500m range around the terminal is rasterized into a 32x32 matrix (resolution 15.625m / pixel); satellite distribution: azimuth (0-360°) and elevation (0-90°) of visible satellites are encoded into a 16x16 polar coordinate matrix; multi-terminal height difference: the height difference (Δh, unit: m) between adjacent terminals, forming an 8x8 tensor. Concatenated into a four-dimensional input tensor: (time step, height, width, channel number) = (10, 32, 32, 3).

[0058] The residual CNN structure includes: 4 residual blocks (each containing 2 3x3 convolution layers + jump connection), 2 global average pooling layers, and 1 fully connected layer; the activation function uses ReLU, and the output layer is normalized to [0, 1] by the Sigmoid function, with higher values indicating stronger environmental interference (e.g., urban canyon scene output 0.7, open scene 0.2).

[0059] In some embodiments, the fusion of the channel fading parameters corresponding to the Beidou terminal, network topology data, communication antenna signal-to-noise ratio, transmission rate variation parameter and data integrity parameter, and the calculation of the communication link stability value, include: constructing a Bayesian network model, defining the channel fading type, node connectivity, signal-to-noise ratio threshold, and rate fluctuation coefficient as network nodes, quantifying the dependency relationship between nodes through conditional probability table, and outputting the communication link stability value based on posterior probability.

[0060] Through the calculation of the communication link stability value based on the Bayesian network, the parameter dependency relationship is modeled.

[0061] The Bayesian network node definition includes: parent nodes: channel fading type (Rayleigh / Rice, discrete variable), node connectivity (0-10, continuous variable); child nodes: signal-to-noise ratio threshold (SNR_th, -10dB~20dB), rate fluctuation coefficient (ΔR / R0, 0-0.5), data integrity (bit error rate, BER<1e-6 is normal); root node: communication link stability state (stable / unstable, output variable).

[0062] The conditional probability table (CPT) includes: preset typical scene probability (e.g. the probability of SNR_th<0dB when in Rice fading is 0.6), the CPT parameters are trained and updated through historical communication logs (100,000 data); inference process: input real-time observation values (e.g. SNR=5dB, ΔR / R0=0.1), calculate the posterior probability P(stable|observation value) through joint tree algorithm, output the stability value (0-1, higher value means more stable).

[0063] In some embodiments, the real-time evaluation result of the positioning reliability is generated according to the positioning reliability index, the data reliability level, the three-dimensional environmental interference coefficient, and the communication link stability value, comprising: using the analytic hierarchy process or a dynamic weighted fusion model to normalize and weight the four parameters, wherein the initial weight is determined by historical data training, and a positioning reliability evaluation vector containing a confidence probability is output.

[0064] The positioning reliability is generated by multi-dimensional parameter weighted fusion, supporting dynamic weight.

[0065] Normalization processing: the positioning reliability index (0-1), the data reliability level (high=1, medium=0.6, low=0.3), the three-dimensional environmental interference coefficient (0-1), and the communication link stability value (0-1) are uniformly mapped to the interval [0, 100].

[0066] The weight initialization is based on historical data (covering 10 typical scenarios, such as static open and dynamic canyon), and the initial weight is calculated by the analytic hierarchy process (AHP) (such as positioning reliability 0.3, data reliability 0.25, environmental interference 0.25, and communication stability 0.2).

[0067] Dynamic weighted fusion: a linear weighted model is used: S=w1R+w2Q+w3E+w4C; wherein R, Q, E, and C are normalized parameters, wi is a dynamic weight, and the positioning reliability S (0-100 points) is output, with a confidence probability (such as S=85±5, indicating 80%-90% confidence).

[0068] In some embodiments, the comparison and analysis of the real-time running data and the historical running data corresponding to the Beidou terminal comprises: performing sliding window filtering and outlier detection on the real-time data and the historical data, calculating the deviation of the real-time value and the historical mean value of the feature parameters such as positioning error mean square deviation, data update rate, and communication packet loss rate, and generating a parameter fluctuation warning signal.

[0069] Parameter abnormal fluctuation is detected by comparing and analyzing real-time data and historical data.

[0070] Data preprocessing includes: sliding window filtering: using a 5-second sliding window to calculate the mean and standard deviation of real-time data (1 second / point), and filtering out high-frequency noise; outlier detection: based on the 3σ principle, when the parameter deviates from the historical mean value by more than 3 times the standard deviation, it is marked as abnormal (such as PDOP>6 for 10 seconds to trigger a warning).

[0071] The feature parameter calculation is performed by the positioning error mean square deviation (RMSE, unit: m): compared with the reference station data; the data update rate (UR, %): the actual received data frame number / theoretical frame number; and the communication packet loss rate (PLR, %): the number of lost packets / the total number of packets.

[0072] Deviation is calculated by the relative deviation between the real-time value and the historical average. When the deviation exceeds, for example, 20%, an early warning signal is generated (such as LED flashing or log recording).

[0073] In some embodiments, the adaptive adjustment of the weights of the positioning reliability index, data credibility level, three-dimensional environmental interference coefficient, and communication link stability value includes: using a gradient descent algorithm or reinforcement learning model, with the objective function of minimizing the deviation between the positioning result and the base station data, dynamically optimizing the weight coefficients of each parameter to form an adaptive weight adjustment strategy.

[0074] Localization bias is minimized by adaptively adjusting weights based on gradient descent / reinforcement learning.

[0075] The objective function is defined to minimize the Euclidean distance between the positioning result and the base station: Where (x0, y0, z0) are the reference coordinates, and (x, y, z) are the terminal positioning results.

[0076] Weight optimization algorithm: Gradient descent: Calculate the gradient of L with respect to each weight wi, update rule: (Learning rate α=0.01, iteration step size 100ms); Reinforcement learning: Construct a state space (current weights, parameters of each dimension), an action space (weight adjustment step size ±0.05), and a reward function (-L). Train using the Q-learning algorithm, updating the optimal weight policy every 5 minutes. Constraints: Ensure the sum of weights is 1 (∑wi=1), and prevent overfitting (add L2 regularization term).

[0077] In some embodiments, the system further includes: an inertial navigation fusion module, which integrates a MEMS accelerometer and a MEMS gyroscope to output triaxial acceleration and angular velocity data of the carrier; the control module uses an extended Kalman filter or an unscented Kalman filter algorithm to loosely or tightly couple the position and velocity information output by the BeiDou positioning module with the inertial measurement data output by the inertial navigation fusion module to generate fused positioning data; the generation of the three-dimensional environmental interference coefficient also includes extracting environmental occlusion features from the attitude angles of the fused positioning data to enhance the accuracy of environmental assessment in complex dynamic environments.

[0078] Improve the accuracy of dynamic scenes by combining inertial navigation fusion with attitude angle-assisted environmental assessment.

[0079] Inertial navigation module integrates MEMS accelerometer (accuracy ±0.1 m / s²) and gyroscope (accuracy ±5° / h) to output three-axis acceleration (a_x, a_y, a_z) and angular velocity (ω_x, ω_y, ω_z) at a frequency of 100 Hz. Fusion positioning algorithm: loosely coupled fusion: through extended Kalman filter (EKF), the position (x, y, z) and velocity (v_x, v_y, v_z) of Beidou positioning are taken as observations, and the inertial navigation output is taken as state prediction to update the state vector (position, velocity, attitude); tightly coupled fusion: directly fuse the Beidou pseudo-range / carrier phase observations and inertial measurement data to improve the continuity of positioning in dynamic scenarios (such as vehicle turning).

[0080] Environment evaluation enhancement extracts the carrier attitude angle (pitch angle θ, roll angle φ, heading angle ψ) from the fusion positioning data, constructs a three-dimensional shielding model (such as when θ>30° and there are high-rise buildings within 50 meters, the environmental interference coefficient weight is enhanced) combined with terrain elevation data, judges the signal shielding direction through the attitude angle, and optimizes the spatial dimension of CNN feature extraction.

[0081] In some embodiments, the evaluation model migration for different terrains (mountainous area / city / ocean) uses source domain (such as mountainous area) training data to quickly adapt to target domain (such as urban canyon), solving the problem of insufficient model generalization across scenarios.

[0082] Transfer learning architecture: feature extraction layer sharing: build a shared backbone network containing convolutional layers (process terrain grid) and LSTM layers (process time-series signals), pre-train to extract general environmental features (such as shielding patterns, signal multipath features) in the source domain (mountainous area). Domain adaptation layer: add a domain classifier (Domain Classifier) to align the feature distribution of the source domain (mountainous area) and the target domain (urban canyon) through a gradient reversal layer (GradientReversal Layer), and minimize the difference between domains.

[0083] Adaptation process: pre-train the base model with multi-scenario data such as mountainous area / ocean in the cloud to extract general features of signal propagation and terrain influence; when deployed in an urban canyon scenario, the terminal only needs a small amount of local data (500 groups) to fine-tune the domain classifier and output layer, quickly generating an evaluation model adapted to the current environment (shortening training time); when outputting three-dimensional environmental interference coefficients, automatically weighting the migrated terrain shielding features (such as increasing the weight of the city high-rise reflection coefficient by 30%). Advantage: solves the high cost of cross-scenario retraining of traditional models, supports terminal rapid self-adaptation in new environments.

[0084] In some embodiments, multi-Beidou terminal data collaborative modeling is achieved through federated learning (Federated Learning), which shares environmental interference evaluation experience under the premise of protecting user privacy (such as not uploading terminal location data).

[0085] The federated learning architecture includes: a terminal layer: each terminal (vehicle-mounted / ship-mounted / handheld) locally collects data (inertial data, signal parameters, terrain fragments), extracts features (such as edge features of building obstructions, frequency domain features of multipath signals) through a lightweight CNN, and only uploads model gradient update values (not raw data).

[0086] A cloud aggregation layer: the central server receives the gradient parameters of each terminal, aggregates to generate a global evaluation model (such as a three-dimensional environmental interference coefficient prediction model) using the FedAvg algorithm, and periodically issues updates (update period 5 minutes).

[0087] Privacy protection mechanism: add differential privacy noise (Laplace mechanism, ε=0.5) to the uploaded gradient to prevent feature reverse deduction of terminal location; the terminal retains data ownership locally and only participates in model training without revealing raw observation values. Collaborative application scenario: multiple vehicle-mounted terminals in the city collaboratively learn high-rise obstruction rules to generate a regional three-dimensional interference map, improving the environmental evaluation accuracy (obstruction judgment accuracy improved by 25%) of all terminals in the region.

[0088] In some embodiments, a digital twin is constructed for a Beidou terminal to pre-evaluate positioning reliability in different environments through virtual simulation and early warning of potential failures.

[0089] The digital twin architecture includes: a physical terminal layer: real-time collection of signal parameters (carrier-to-noise ratio, pseudorange residual), inertial data (acceleration, angular velocity), and environmental parameters (altitude, temperature); a virtual twin layer: based on Unity / UE engine to build a three-dimensional scene (restore the terminal surrounding 500 meters terrain), integrate signal propagation model (such as ray tracing method to simulate multipath effect), positioning algorithm (consistent with physical terminal); data mapping: through space-time calibration algorithm (such as ICP matching) to synchronize the position and attitude of physical terminal and twin, error controlled within 0.1 meters.

[0090] Pre-evaluation application: When the terminal enters a new area (such as a tunnel entrance), the twin body simulates the signal shielding situation in the next 10 seconds in advance, predicts the three-dimensional environmental interference coefficient mutation threshold (such as triggering the inertial navigation dominant strategy when the interference coefficient > 0.8), and trains the LSTM fault prediction model based on historical fault data (such as sudden drop of carrier-to-noise ratio leading to positioning failure). The twin body outputs the fault probability every 2 seconds (such as "75% probability of communication interruption in the next 30 seconds"), and the terminal caches key data in advance. By realizing forward-looking evaluation, the positioning interruption time caused by sudden environmental changes is shortened.

[0091] In some embodiments, by designing a lightweight evaluation model based on model distillation and neural architecture search (NAS) for low-power handheld terminals, a balance between performance and power consumption is achieved.

[0092] Model lightweight technology includes: knowledge distillation: taking a complex CNN model (such as the residual network in embodiment 4) in the cloud as a teacher model, training a student model (such as MobileNetV3), reducing the parameter amount by 70% (from 12M to 3.6M) while maintaining 85% accuracy; neural architecture search: through reinforcement learning to search for the optimal network structure, optimizing the convolution kernel size (3x3→2x2) and the number of channels (256→128) according to the terminal CPU / GPU characteristics (such as ARM Cortex-A53), and improving the inference speed by 40% (single-frame processing time <5ms).

[0093] Low-power deployment: dynamic wake-up mechanism: when it is detected that the carrier is stationary (acceleration <0.1m / s² for 30 seconds), switch to a lightweight model (power consumption from 150mW to 50mW); data compression: lossy compression of inertial data (100Hz→10Hz downsampling), terrain data (grid resolution from 1m→5m) to reduce computational load. Adapt to handheld terminals, wearable devices, extend the battery life by more than 20%, while maintaining the environmental interference evaluation accuracy greater than the preset threshold (such as 90%).

[0094] The application solves the one-sidedness problem of traditional single index evaluation by modeling and analyzing parameters in four dimensions of positioning state, data quality, use environment and communication state, and builds a comprehensive evaluation system covering signal propagation, data characteristics, environmental interference and communication stability. Through the convolutional neural network processing of inertial navigation data, terrain elevation data and multi-terminal elevation difference data, the quantitative evaluation of three-dimensional spatial environmental interference is realized, and the environmental feature extraction accuracy in complex terrain or dynamic carrier scene is significantly improved. Through the comparison and analysis based on real-time data and historical data, the evaluation parameter weight is adaptively adjusted by the gradient descent algorithm, so that the terminal can dynamically optimize the evaluation model according to the current running state, and the accuracy and robustness of positioning credibility evaluation in different scenes are effectively improved. Through the fusion of multi-source data and intelligent algorithms (time series modeling, random forest, Bayesian network, etc.), the whole link state monitoring from the signal layer to the application layer is realized, which provides multi-level guarantee for the credibility of terminal positioning results, and is suitable for high-precision navigation, automatic driving and other scenes with high reliability requirements.

[0095] It should be understood that the terms used herein in the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. It should be understood that when an element or layer is referred to as "on", "adjacent to", "connected to", or "coupled to" another element or layer, it can be directly on, adjacent to, connected or coupled to the other element or layer, or there can be an intervening element or layer. Conversely, when an element is referred to as "directly on", "directly adjacent to", "directly connected to" or "directly coupled to" another element or layer, there is no intervening element or layer. It should be understood that although the terms first, second, third, etc. are used to describe various elements, components, regions, layers and / or parts, these elements, components, regions, layers and / or parts should not be limited by these terms. These terms are only used to distinguish one element, component, region, layer or part from another element, component, region, layer or part. Therefore, the first element, component, region, layer or part discussed below can be represented as the second element, component, region, layer or part without departing from the teachings of the present application.

[0096] Spatially relative terms, such as "beneath", "below", "lower", "under", "above", "upper" and the like, can be used herein for ease of description to describe one element or feature's relationship to another element(s) or feature(s) as illustrated in the figures. It will be understood that the spatially relative terms are intended to encompass different orientations of the device in use and / or operation in addition to the orientations depicted in the figures. For example, if a device in the figures is inverted, then a dependent- element described as "below" or "beneath" another element or feature would then be oriented "above" and "over" the other element or feature. Thus, the exemplary term "below" can encompass both an orientation of above and below. The device can be otherwise oriented (rotated 90 degrees or at other orientations) and the spatially relative descriptors used herein interpreted accordingly.

[0097] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising", when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0098] It will be further understood that the terms "comprises" and / or "comprising", when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0099] The above description is only specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A BeiDou terminal for intelligent assessment of multiple states, characterized in that, include: The BeiDou positioning module is used to acquire the BeiDou signal corresponding to the BeiDou terminal; The control module is configured to execute: The propagation delay, geometric distribution factor, ionospheric influence parameters, and tropospheric influence parameters of the BeiDou signal are time-series modeled to generate a positioning reliability index, thereby achieving positioning status assessment. The pseudorange residual, carrier-to-noise ratio, multipath effect parameters, number of satellite signals, data integrity parameters, and cycle slip parameters corresponding to the BeiDou signal are jointly analyzed to output the data credibility level, so as to achieve data quality assessment. By processing the inertial navigation data, terrain elevation data, sky satellite geometric distribution factor and multi-terminal elevation difference data corresponding to the Beidou terminal through a convolutional neural network, a three-dimensional environmental interference coefficient is generated to achieve environmental assessment. By integrating the channel fading parameters, network topology data, communication antenna signal-to-noise ratio, transmission rate variation parameters, and data integrity parameters corresponding to the BeiDou terminal, the stability value of the communication link is calculated to achieve communication status assessment. Based on the positioning reliability index, data credibility level, three-dimensional environmental interference coefficient, and communication link stability value, a real-time evaluation result of positioning credibility is generated. Based on the comparative analysis of the real-time operation data and historical operation data corresponding to the Beidou terminal, the weights of the positioning reliability index, data credibility level, three-dimensional environmental interference coefficient and communication link stability value are adaptively adjusted to update the real-time evaluation results.

2. The Beidou terminal according to claim 1, characterized in that, The step of obtaining the BeiDou signal corresponding to the BeiDou terminal includes: The system receives BeiDou B1, B2 and B3 frequency signals through a multi-band antenna, and preprocesses the signals using an anti-multipath filter and adaptive gain control circuit to suppress reflected signal interference and optimize signal strength.

3. The Beidou terminal according to claim 1, characterized in that, The process of performing time-series modeling on the propagation delay, geometric distribution factor, ionospheric influence parameters, and tropospheric influence parameters of the BeiDou signal to generate a positioning reliability index includes: Establish a time series model based on ARIMA or LSTM, inputting time series data including ionospheric puncture point delay, tropospheric zenith delay correction, and geometric distribution factor, and obtain a reliability index representing the error fluctuation trend through the training output corresponding to the time series model.

4. The Beidou terminal according to claim 1, characterized in that, The joint analysis of the pseudorange residual, carrier-to-noise ratio, multipath effect parameters, number of satellite signals, data integrity parameters, and cycle slip parameters corresponding to the BeiDou signal outputs a data reliability level, including: The parameters are normalized and used as input features for the random forest algorithm. The trained classifier outputs a confidence level containing multi-level data. The multi-path effect parameters are calculated by the amplitude difference and phase difference between the direct path and the reflection path of the signal.

5. The Beidou terminal according to claim 1, characterized in that, The process of processing the inertial navigation data, terrain elevation data, sky satellite geometric distribution factor, and multi-terminal elevation difference data corresponding to the BeiDou terminal through a convolutional neural network to generate a three-dimensional environmental interference coefficient includes: The acceleration and angular velocity data of inertial navigation are concatenated with terrain elevation grid data, satellite azimuth or elevation distribution matrix and multi-terminal altitude difference tensor into a three-dimensional input tensor. Environmental occlusion, terrain undulation and multi-terminal signal interference features are extracted by a convolutional neural network containing residual blocks, and the interference coefficient in the 0-1 range is output.

6. The Beidou terminal according to claim 1, characterized in that, The calculation of the communication link stability value by integrating the channel fading parameters, network topology data, communication antenna signal-to-noise ratio, transmission rate variation parameters, and data integrity parameters corresponding to the BeiDou terminal includes: A Bayesian network model is constructed, defining channel fading type, node connectivity, signal-to-noise ratio threshold, and rate fluctuation coefficient as network nodes. The dependencies between nodes are quantified using a conditional probability table, and the stable value of the communication link based on posterior probability is output.

7. The Beidou terminal according to claim 1, characterized in that, The step of generating a real-time evaluation result of positioning reliability based on the positioning reliability index, data credibility level, three-dimensional environmental interference coefficient, and communication link stability value includes: The four parameters are normalized and weighted using the analytic hierarchy process or a dynamic weighted fusion model. The initial weight values ​​are determined through training with historical data, and the output is a location reliability assessment vector containing confidence probabilities.

8. The Beidou terminal according to claim 1, characterized in that, The comparative analysis based on the real-time operation data and historical operation data corresponding to the BeiDou terminal includes: Sliding window filtering and outlier detection are performed on real-time and historical data. The deviation between the real-time values ​​and historical averages of characteristic parameters such as the root mean square error of positioning error, data update rate, and communication packet loss rate is calculated, and parameter fluctuation warning signals are generated.

9. The Beidou terminal according to claim 1, characterized in that, The adaptive adjustment of the weights of the positioning reliability index, data credibility level, three-dimensional environmental interference coefficient, and communication link stability value includes: By using gradient descent algorithm or reinforcement learning model, with the objective function of minimizing the deviation between the positioning result and the base station data, the weight coefficients of each parameter are dynamically optimized to form an adaptive weight adjustment strategy.

10. The Beidou terminal according to claim 1, characterized in that, Also includes: An inertial navigation fusion module, which integrates a MEMS accelerometer and a MEMS gyroscope, is used to output the three-axis acceleration and angular velocity data of the carrier; The control module uses extended Kalman filtering or unscented Kalman filtering algorithms to loosely or tightly couple the position and velocity information output by the BeiDou positioning module with the inertial measurement data output by the inertial navigation fusion module to generate fused positioning data. The generation process of the three-dimensional environmental interference coefficient also includes extracting environmental occlusion features from the attitude angles of the fused positioning data to enhance the accuracy of environmental assessment in complex dynamic environments.

Images