Satellite space matching method, system and equipment based on global optimization and medium

Abstract

The invention discloses a global optimization-based satellite space matching method, system and device and a medium, and the method comprises the steps: obtaining satellite signal data, and obtaining a first satellite set in a first environment and a second satellite set in a second environment through screening and sorting; satellite in a first satellite set and satellites in a second satellite set serve as two node sets of a bipartite graph, connection is established between each pair of satellites from different sets, and a weighted bipartite graph model is constructed; on the basis of a weighted bipartite graph model, calculating the space geometric cost, the signal physical cost and the calculation geometric contribution cost between each pair of satellite nodes, and carrying out normalized weighted fusion to determine the weight for connecting each satellite node edge; and based on the weight of each satellite node edge, solving the minimum weight matching problem of the weighted bipartite graph model through a combinatorial optimization algorithm to obtain a global optimal satellite matching pair set. According to the method, the satellite matching accuracy and reliability can be remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite navigation, and in particular to a satellite space matching method, system, device and medium based on global optimization. BACKGROUND

[0002] In the application scenarios of satellite navigation positioning, especially in the operation in complex environments such as canyon geological disaster monitoring, the receiver has the possibility of obtaining satellite signals from multiple different data sources. For example, one of the data sources can be a reference station set in an open area, and the other can be a mobile station deployed in an area with relatively poor environmental conditions. In order to achieve the goal of high-precision differential positioning or realize the grafting operation of signals, the accurate space matching task between the satellites observed by the two data sources must be completed. The core of this task is to identify the observation data pairs that respectively come from two independent data sources but actually correspond to the same physical satellite.

[0003] The currently widely used satellite space matching method relies on a greedy strategy-based algorithm to perform the matching process. The algorithm first calculates the spatial angular distance between all possible pairs formed by the two satellite sets, and then selects the pair with the smallest angular distance as the first best matching result. Subsequently, the satellite pair that has been determined to be matched will be removed from the original set, and the same steps of recalculating the angular distance, finding the minimum value and completing the matching will be performed again in the remaining satellites. This process is repeated until all satellites that can participate in matching are processed. However, this algorithm itself has an unavoidable defect. It selects the current optimal matching scheme at each step, that is, the pair with the smallest angular distance, but this approach cannot ensure that the overall matching result formed is the most ideal from a global perspective. In particular, when the number of satellites that can be observed by the two data sources is not equal, or their distribution in the sky is significantly different, the pair with the absolute smallest angular distance is selected in the initial stage, which may cause the remaining satellites to have difficulty finding suitable matching objects, thereby making the overall matching effect very unsatisfactory. As can be seen, it is precisely because the local optimal choice is made that the global matching quality is reduced, which seriously affects the accuracy and reliability of subsequent data processing work. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a satellite space matching method, system, device and medium based on global optimization to solve the problem of non-global optimal satellite matching results obtained by the existing satellite space matching method.

[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a satellite space matching method based on global optimization, comprising: Obtaining satellite signal data, and through screening and sorting, obtaining a first satellite set under a first environment and a second satellite set under a second environment; Constructing a weighted bipartite graph model by taking the satellites in the first satellite set and the second satellite set as two node sets of a bipartite graph respectively, and establishing a connection between each pair of satellites from different sets; Based on the weighted bipartite graph model, the weight of the edge connecting each satellite node is determined by calculating the space geometric cost, signal physical cost and calculation geometric contribution cost between each pair of satellite nodes, and performing normalized weighted fusion; Based on the weight of each satellite node edge, the minimum weight matching problem of the weighted bipartite graph model is solved by a combinatorial optimization algorithm to obtain a globally optimal satellite matching pair set.

[0007] As a preferred scheme of the satellite space matching method based on global optimization, the present application comprises: Based on the first satellite set and the second satellite set, the structure framework of the bipartite graph is constructed by defining the two sets as mutually disjoint node sets; Based on the structure framework of the bipartite graph, all possible potential matching pairs are enumerated by establishing a connection relationship between each satellite in the first satellite set and each satellite in the second satellite set, forming a complete edge set; Based on the edge set and the node set, the weighted bipartite graph model is constructed by integrating the nodes, edges and their to-be-weighted attributes.

[0008] As a preferred scheme of the satellite space matching method based on global optimization, the present application comprises: The current candidate matching pair is taken as a satellite combination corresponding to a potential edge in the weighted bipartite graph model; Based on the candidate matching pair, the space geometric cost, signal physical cost and calculation geometric contribution cost between the corresponding satellites are calculated in parallel to obtain three independent matching quality evaluation components; Based on the three types of costs and environmental parameters, the space geometric cost, signal physical cost and calculation geometric contribution cost are normalized, and the weight coefficients of each cost are adjusted according to the current environment by using a fuzzy logic controller, and then weighted summation is performed; A dynamic cost function based on state prediction is introduced to integrate the current matching cost and the expected future matching cost. Combined with an adaptive adjustment mechanism for the cost function weights, the fusion weights of each cost component are dynamically adjusted according to real-time environmental parameters. The weights of the edges connecting each satellite node are determined by weighted fusion based on the output of the dynamic cost function and the adaptively adjusted weights.

[0009] The beneficial effects of this preferred technical solution are that by integrating multi-dimensional matching costs, introducing an environment-adaptive dynamic weight adjustment mechanism, and a state-predictive forward-looking cost function, it achieves global optimality, robustness, and time stability of satellite matching in complex environments, and significantly improves the accuracy and reliability of positioning solutions in occlusion scenarios such as canyons.

[0010] As a preferred embodiment of the satellite spatial matching method based on global optimization described in this invention, the spatial geometric cost includes: calculating the spatial angular distance between the satellites in the first environment and the satellites in the second environment based on the azimuth and elevation angle information of the satellites in the first environment and the satellites in the second environment; and using the spatial angular distance as the spatial geometric cost. The signal physical cost includes: based on the signal carrier-to-noise ratio of satellites in the first environment and satellites in the second environment, calculating the sum of the reciprocals of the satellite signal carrier-to-noise ratio in the first environment and the satellite signal carrier-to-noise ratio in the second environment as the signal physical cost; The solution geometric contribution cost includes: constructing the benchmark Jacobian matrix and corresponding covariance matrix of the differential positioning system based on the high-confidence benchmark matching set; evaluating the degree of improvement of the geometric accuracy factor after introducing candidate matching pairs; determining the contribution of the candidate matching pairs to the positioning solution performance based on the degree of improvement; and using the quantification result of the contribution as the solution geometric contribution cost.

[0011] As a preferred embodiment of the satellite space matching method based on global optimization described in this invention, the dynamic cost function based on state prediction includes: The future states of satellites in the first and second environments are predicted based on an adaptive unscented Kalman filter, and the changes in satellite position, velocity and signal strength at multiple consecutive future moments are obtained based on the estimation results, which serve as the predicted future state sequence. Based on the predicted future state sequence, calculate the expected matching cost of the candidate matching pair at each future time. A dynamic cost function is constructed, which sums the matching cost at the current moment with the expected matching cost at each future moment in a weighted manner, and introduces a discount factor for the expected cost in the long term during the summation process. The calculation result of the dynamic cost function is used as the weight of the edge connecting the satellite nodes.

[0012] The beneficial effects of this preferred technical solution are that by using an adaptive unscented Kalman filter to predict the future state of the satellite and constructing a dynamic cost function that includes the current and future expected matching costs, the forward-looking optimization of the satellite matching edge weights is achieved, which significantly improves the stability and accuracy of satellite matching in complex environments and effectively reduces matching errors caused by environmental changes.

[0013] As a preferred embodiment of the satellite space matching method based on global optimization described in this invention, the adaptive unscented Kalman filter includes: The Josephus form is used for state covariance update; Based on the adaptive mechanism of the Sage-Houssa algorithm, the process noise covariance matrix and the observation noise covariance matrix are recursively estimated online by monitoring the filtered information sequence. The forgetting factor in the Saggi-Houssa algorithm is adjusted in real time based on the normalized squared innovation. The initial values ​​of the process noise covariance matrix and the observation noise covariance matrix include constructing an optimal experimental design protocol based on the Fisher information matrix, screening and collecting the observation data segment with the largest amount of information, and using the expectation-maximization algorithm combined with a fixed interval smoother to identify the parameters of the noise covariance until convergence.

[0014] As a preferred embodiment of the satellite spatial matching method based on global optimization described in this invention, the adaptive adjustment of the cost function weights includes: The average carrier-to-noise ratio of visible satellites, the total number of visible satellites, and the multipath effect index are used as environmental input parameters and input to the fuzzy logic controller. The fuzzy logic controller performs inference operations based on a preset fuzzy rule base, and dynamically outputs the weight coefficients corresponding to the spatial geometric cost, signal physical cost, and solution geometric contribution cost.

[0015] Secondly, the present invention provides a satellite space matching system based on global optimization, comprising: The acquisition module is used to acquire satellite signal data, and through filtering and sorting, obtain the first set of satellites under the first environment and the second set of satellites under the second environment; The bipartite graph modeling module is used to construct a weighted bipartite graph model by treating the satellites in the first satellite set and the second satellite set as two node sets of the bipartite graph, and establishing connections between each pair of satellites from different sets. The matching cost calculation and weighting module is used to determine the weight of the edges connecting each satellite node by calculating the spatial geometric cost, signal physical cost and solution geometric contribution cost between each pair of satellite nodes based on the weighted bipartite graph model, and performing normalized weighted fusion. The global optimal matching solution module is used to solve the minimum weight matching problem of the weighted bipartite graph model based on the weights of the edges of each satellite node, and obtain the globally optimal set of satellite matching pairs through a combinatorial optimization algorithm.

[0016] Thirdly, the present invention provides an electronic device, comprising: Memory, used to store programs; A processor for executing the computer-executable instructions, which, when executed by the processor, implement the steps of the globally optimized satellite space matching method.

[0017] Fourthly, the present invention provides a computer-readable storage medium, comprising: when the program is executed by a processor, the steps of implementing the satellite space matching method based on global optimization.

[0018] The beneficial effects of this invention are as follows: By constructing the satellite sets in the first and second environments as two node sets of a weighted bipartite graph and solving their minimum weight matching problem, this invention achieves the transformation of satellite matching from local optimum to global optimum, effectively avoiding the technical defects of traditional greedy algorithms that lead to overall performance degradation due to early mismatches, and improving the accuracy and reliability of matching results. By constructing a multi-factor weighted model that includes spatial geometric cost, signal physical cost, and solution geometric contribution cost, and introducing a fuzzy logic controller to dynamically adjust the fusion weight of each cost component according to real-time environmental parameters, this invention achieves adaptive response of the matching strategy to complex and variable environments (such as canyon occlusion and multipath interference), significantly enhancing the robustness of the matching process. By using an adaptive unscented Kalman filter to predict the future state of satellites, constructing a dynamic cost function that integrates the current matching cost and the expected future matching cost, and introducing a time decay factor to weight the long-term prediction, this invention achieves forward-looking optimization of matching decisions, effectively suppressing frequent jitter of matching results in the time series and improving matching stability. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a basic flowchart of a satellite space matching method based on global optimization, provided as an embodiment of the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0021] Example 1, referring to Figure 1 As an embodiment of the present invention, a satellite space matching method based on global optimization is provided, comprising: S100: Acquire satellite signal data, and through filtering and sorting, obtain the first satellite set under the first environment and the second satellite set under the second environment; S200: A weighted bipartite graph model is constructed by treating the satellites in the first and second satellite sets as two node sets of a bipartite graph, and establishing connections between each pair of satellites from different sets. S300: Based on a weighted bipartite graph model, the weights of the edges connecting each satellite node are determined by calculating the spatial geometric cost, signal physical cost, and solution geometric contribution cost between each pair of satellite nodes and performing normalized weighted fusion. S400: Based on the weights of the edges of each satellite node, a combinatorial optimization algorithm is used to solve the minimum weight matching problem of the weighted bipartite graph model, and obtain the globally optimal set of satellite matching pairs.

[0022] It should be noted that existing satellite spatial matching methods face a series of challenges during operation. In complex environments such as canyons and urban buildings, the large differences in observation conditions on both sides and the asymmetry in the number and spatial distribution of visible satellites make traditional greedy matching algorithms prone to getting trapped in local optima and unable to obtain globally optimal matching results. At the same time, factors such as signal obstruction, multipath effects, and ionospheric disturbances cause drastic fluctuations in satellite signal quality, and relying solely on single indicators such as spatial angular distance or carrier-to-noise ratio for matching can easily lead to mismatches. In addition, when the environment changes dynamically, the matching relationship switches frequently, resulting in severe jitter in the positioning results and poor stability. More importantly, existing methods are mostly based on the current instantaneous state for decision-making and lack the ability to predict future states, making it difficult to take into account the long-term quality of matching and meet the stringent requirements of application scenarios such as high-precision and high-reliability geological disaster monitoring.

[0023] Therefore, addressing the issue that existing satellite spatial matching methods do not yield globally optimal results, this paper proposes a fundamental breakthrough in satellite spatial matching from local optimum to global optimum through steps S100-S400. This breakthrough involves constructing a weighted bipartite graph model and solving the minimum weight matching problem, combined with a dynamic cost function based on multi-factor fusion and an environment-adaptive weight adjustment mechanism. This significantly improves the accuracy, robustness, and temporal continuity of satellite matching in demanding scenarios such as canyon geological disaster monitoring, providing a reliable data foundation for high-precision differential positioning and signal grafting.

[0024] Example 2, this is an embodiment of the present invention, which provides a satellite space matching method based on global optimization based on the previous embodiment, including: In this embodiment, observable satellite signal data are acquired in a first environment (e.g., an open environment) and a second environment (e.g., a complex environment). After filtering and processing, a first set of satellites in the first environment is obtained. ,in, The number of visible satellites in this environment; simultaneously, the second satellite set in the second environment is obtained. ,in, This represents the number of visible satellites in this environment.

[0025] In this embodiment of the application, step S200, which involves constructing a weighted bipartite graph model, includes: Based on the first satellite set Second satellite set By defining two sets as disjoint sets of nodes, a bipartite graph structural framework is constructed. ; Based on the bipartite graph framework, by establishing connections between each satellite in the first satellite set and each satellite in the second satellite set, enumeration... , Potential matching pairs between them form a complete edge set. ; Based on the edge set and node set, a weighted bipartite graph model is constructed by integrating the satellite state information of each node and the potential matching features corresponding to each edge. .

[0026] In this embodiment of the application, step S300, determining the weights of the edges connecting each satellite node, includes: The current candidate matching pair is treated as a satellite combination corresponding to a potential edge in the weighted bipartite graph model; Based on candidate matching pairs, three independent matching quality evaluation components are obtained by parallel computing of the spatial geometric cost, signal physical cost, and solution geometric contribution cost between corresponding satellites; Based on three types of costs and environmental parameters, the spatial geometric cost, signal physical cost, and solution geometric contribution cost are normalized, and the weight coefficients of each cost are dynamically adjusted according to the current environment using a fuzzy logic controller, and then a weighted sum is performed. A dynamic cost function based on state prediction is introduced to integrate the current matching cost and the expected future matching cost. Combined with an adaptive adjustment mechanism for the cost function weights, the fusion weights of each cost component are dynamically adjusted according to real-time environmental parameters. The weights of the edges connecting each satellite node are determined by weighted fusion based on the output of the dynamic cost function and the adaptively adjusted weights.

[0027] In this embodiment of the application, the multi-cost fusion strategy in step S300 calculates the spatial geometric cost, signal physical cost, and solution geometric contribution cost of the candidate matching pair respectively, normalizes the three types of costs, and uses a fuzzy logic controller to dynamically adjust the weight coefficients of each cost according to real-time environmental parameters. Finally, a weighted sum is performed to generate a comprehensive matching cost as the weight of the edge in the weighted bipartite graph.

[0028] In an optional implementation, the multi-cost fusion strategy in step S300 can also calculate the spatial geometric cost, signal physical cost, and solution geometric contribution cost of the candidate matching pair respectively. After normalizing the three types of costs, a weighted sum is performed using a pre-set fixed weight coefficient to generate a comprehensive matching cost as the weight of the edge in the weighted bipartite graph.

[0029] In an optional implementation, the multi-cost fusion strategy in step S300 can also extract the spatial geometric features, signal physical features, and environmental context parameters of the candidate matching pairs as inputs, and output the comprehensive matching cost through nonlinear mapping of a pre-trained lightweight neural network or support vector regression model, which is directly used as the weight of the edge in the weighted bipartite graph to achieve data-driven adaptive cost fusion.

[0030] In this embodiment of the application, for connecting satellites and satellite The weight of each edge Defined as the "quality-weighted matching cost" between them. This cost is a comprehensive indicator; the smaller the value, the higher the quality of the match. Preferably, this quality-weighted matching cost... It is specifically defined as a multi-factor weighted model, and its calculation method is as follows: in, These are the weights of the spatial geometric cost, the signal physical cost, and the solution geometric contribution cost, respectively, satisfying... . Normalization functions, such as max-min normalization, are used to scale the components to a comparable range (e.g., [0,1]).

[0031] In this embodiment of the application, the spatial geometric cost in step S300 This includes: azimuth angles of satellites in the first environment and satellites in the second environment. and elevation angle Information, calculate the spatial angular distance between satellites in the first environment and satellites in the second environment. And the spatial angular distance is used as the spatial geometric cost; In this embodiment of the application, the signal physical cost in step S300 This includes: calculating the satellite signal-to-noise ratio (CNR) in the first environment based on the CNR of satellites in the first environment and satellites in the second environment. Satellite signal carrier-to-noise ratio in the second environment The sum of the reciprocals serves as the physical cost of the signal; In this embodiment of the application, the geometric contribution cost is calculated in step S300. Including: matching sets based on high confidence benchmarks Constructing the baseline Jacobian matrix of the differential positioning system and the corresponding covariance matrix Evaluation of the introduction of candidate matching pairs The degree of improvement of the post-geometric accuracy factor is used to determine the contribution of candidate matching pairs to the positioning solution performance, and the quantification result of the contribution is used as the solution geometric contribution cost.

[0032] covariance matrix Efficient calculation using the Sherman-Morrison formula: in, This is the baseline covariance matrix. Finally, the geometric contribution cost is calculated. It can be defined as the reciprocal of the improvement in geometric precision factor (GDOP) after introducing the matching pair, for example... The larger this value, the smaller the contribution and the higher the matching cost.

[0033] In this embodiment of the application, the dynamic cost function based on state prediction in step S300 includes: The future states of satellites in the first and second environments are predicted based on an adaptive unscented Kalman filter, and the changes in satellite position, velocity and signal strength at multiple consecutive future moments are obtained based on the estimation results, which serve as the predicted future state sequence. Based on the predicted sequence of future states, calculate the expected matching cost of each candidate matching pair at each future time. A dynamic cost function is constructed, which sums the matching cost at the current moment with the expected matching cost at each future moment in a weighted manner, and introduces a discount factor for the expected cost in the long term during the summation process. The result of the dynamic cost function calculation is used as the weight of the edge connecting the satellite nodes.

[0034] In this embodiment of the application, the future cost processing in the dynamic cost function in step S300 calculates the expected matching cost of the candidate matching pair at each future time by using the satellite future state predicted by the adaptive unscented Kalman filter. When constructing the dynamic cost function, the matching cost at the current time and the expected cost in the next N steps are weighted and summed by an exponential discount factor, so that the long-term cost decays over time, thereby reducing the impact of prediction uncertainty on the current decision.

[0035] In an optional implementation, the future cost processing in the dynamic cost function in step S300 can also calculate the expected matching cost of the candidate matching pair in the next N consecutive time steps based on the satellite state prediction results. When constructing the dynamic cost function, the arithmetic mean (or equal weight average) of the matching cost at the current time and the expected cost in the next N steps is weighted and fused to form a comprehensive cost as the weight of the edge of the bipartite graph, so as to simplify the calculation and improve short-term stability.

[0036] In an optional implementation, the future cost processing in the dynamic cost function in step S300 can also calculate the uncertainty measure of the satellite state at each future time based on the prediction covariance of the adaptive unscented Kalman filter output, thereby generating dynamic confidence weights, and weighting and summing the expected matching costs of candidate matching pairs at each future time, so that the long-term predictions with high confidence (low uncertainty) receive higher weights, and then constructing a comprehensive dynamic cost as the weight of the bipartite graph edge.

[0037] In this embodiment, the dynamic cost function based on state prediction includes employing a state prediction filter, preferably an adaptive unscented Kalman filter (AUKF), to predict the future state (such as position, velocity, signal strength change rate, etc.) of each satellite. Subsequently, the dynamic cost function is constructed. It considers not only the current moment Matching cost It also includes ideas about the future. The expected matching cost over a long period of time. The weighted summation is used. The expected matching cost is calculated based on AUKF's prediction of future states. The specific dynamic cost function is shown below: in, It is a weighting factor that balances current and future costs. It is a future cost discount factor, used to reduce the impact of uncertainty in long-term forecasts. The smaller the value, the more the system focuses on near-future stability. This is the predicted number of steps. This dynamic cost is used when solving bipartite graph matching problems. As the weight of the edge, it makes the decision more forward-looking.

[0038] In this embodiment of the application, the adaptive unscented Kalman filter in step S300 includes: The Josephus form is used for state covariance update; Based on the adaptive mechanism of the Sage-Houssa algorithm, the process noise covariance matrix and the observation noise covariance matrix are recursively estimated online by monitoring the filtered information sequence. The forgetting factor in the Saggi-Houssa algorithm is adjusted in real time based on the normalized squared innovation. The initial values ​​of the process noise covariance matrix and the observation noise covariance matrix are obtained by constructing an optimal experimental design protocol based on the Fisher information matrix, screening and collecting the observation data segment with the largest amount of information, and using the expectation-maximization algorithm combined with a fixed interval smoother to identify the parameters of the noise covariance until convergence.

[0039] In this embodiment, an adaptive unscented Kalman filter (AUKF) is used to predict the satellite state in both the first and second environments to address the strong nonlinear variations and dynamic uncertainties of satellite signals in complex environments. This filter significantly improves the accuracy and robustness of state estimation by introducing an adaptive mechanism and numerical stability optimization measures.

[0040] In this embodiment, the satellite state prediction filter in step S300 employs an adaptive unscented Kalman filter (AUKF). Based on real-time observation data of the satellite in the first and second environments, it uses unscented transformation to capture nonlinear dynamic characteristics, combines the Sage-Husa algorithm to estimate and update the process noise and observation noise covariance matrix online, and uses the Joseph form to ensure the numerical stability of the covariance update. This allows for the recursive prediction of the satellite position, velocity, and signal strength change rate at multiple future moments, providing a forward-looking state input for the dynamic cost function.

[0041] In an optional implementation, the satellite state prediction filter in step S300 can also be based on satellite observation data of the first environment and the second environment, and process the nonlinear state transition and observation model through first-order Taylor linearization, and recursively estimate the rate of change of satellite position, velocity and signal strength using a fixed or preset noise covariance matrix, and use the prediction results to calculate the expected matching cost at future times, thereby constructing a dynamic cost function as the weight of the bipartite graph edge.

[0042] In an optional implementation, the satellite state prediction filter in step S300 can also be based on the satellite observation data in the first environment and the second environment. By constructing a state-space model that includes the rate of change of position, velocity and signal strength, a particle set is generated by Monte Carlo sampling. The particle weights are evaluated according to the observation likelihood and resampled. The posterior probability distribution of the satellite state is recursively estimated, thereby predicting the state sequence at multiple future times. The predicted matching cost is used to construct a dynamic cost function as the weight of the bipartite graph edge.

[0043] In this embodiment, an adaptive mechanism based on the Sage-Husa algorithm is employed to recursively update the process noise covariance matrix by monitoring the filtering innovation sequence online. and observation noise covariance matrix This mechanism enables the filter to automatically adjust its noise statistics when the system's dynamic characteristics change abruptly, avoiding estimation divergence caused by prior noise model mismatch.

[0044] In the embodiments of this application, the strategy dynamically adjusts the forgetting factor based on the real-time value of the normalized innovation square (NIS): when the NIS is large, indicating a significant mismatch in the system model, the forgetting factor is automatically reduced to accelerate the tracking response to new dynamics; when the NIS is small, indicating a good model match, the forgetting factor is increased to improve the smoothness and stability of the estimation results.

[0045] In this embodiment, the Josephus form is used for calculation. This form, by explicitly constructing a complete expression for the covariance matrix, effectively enhances the numerical stability of the algorithm and ensures the mathematical consistency and convergence of the filtering process throughout the entire running cycle.

[0046] In this embodiment, to improve the initial convergence speed of AUKF and ensure the estimation quality during the "cold start" phase, an offline initialization process is also disclosed to determine the process noise covariance matrix. and observation noise covariance matrix High-quality initial values. The process is as follows: 1. Constructing an optimal experimental design protocol. To ensure that the dataset used for parameter identification contains sufficient information, an optimal experimental design protocol is first constructed based on the Fisher Information Matrix (FIM). By maximizing the determinant of the FIM (D-optimality criterion) or other optimality criteria, an excitation input sequence that maximizes the accuracy of parameter identification can be actively designed, or the most informative segments can be selected from massive historical data.

[0047] 2. Collect or screen sample data. According to the designed protocol, collect an observation sequence of length N, or screen data segments that meet the protocol requirements from existing data.

[0048] 3. Parameter identification is performed using the Expectation-Maximization (EM) algorithm. To avoid the computational burden and susceptibility to local optima that may arise from directly optimizing complex likelihood functions, this embodiment employs the EM algorithm to identify the initial values ​​of the process noise covariance matrix Q and the observation noise covariance matrix R. The EM algorithm treats the system state sequence as a latent variable and converges by alternately executing the following two steps: 1) E-Step (Expectation Step): Given the current parameter estimates, run a fixed interval smoother (such as an RTS smoother) on the collected sample data once to obtain the optimal estimate of the state sequence.

[0049] 2) M-Step (Maximization Step): Using the smooth state sequence obtained by E-step, the estimated values ​​of new Q and R that maximize the expectation are calculated through analytical or approximate analytical update formulas.

[0050] After the iteration converges, the optimal estimates of Q and R will be used as the initial values ​​for AUKF. and .

[0051] In this embodiment of the application, the adaptive adjustment of the cost function weights in step S300 includes: The average carrier-to-noise ratio of visible satellites, the total number of visible satellites, and the multipath effect index are used as environmental input parameters and input to the fuzzy logic controller. The fuzzy logic controller performs inference operations based on a preset fuzzy rule base, and dynamically outputs the weight coefficients corresponding to the spatial geometric cost, signal physical cost, and solution geometric contribution cost.

[0052] In this embodiment, to enable the cost model to adapt to different external environments, the adaptive adjustment of the cost function weights further includes a dynamic weight adjustment module, preferably a fuzzy logic controller. This controller acquires environmental parameters in real time, such as the average carrier-to-noise ratio of all visible satellites, the total number of visible satellites, and multipath effect indices obtained through signal analysis. Based on these inputs, the controller dynamically adjusts the weights of the three cost components according to a preset fuzzy rule base (e.g., "if the signal environment is poor, increase the weight of the signal physical cost"). This ensures that matching decisions focus on the most important factors at any given time, regardless of the circumstances.

[0053] In this embodiment of the application, solving for the minimum weight matching in step S400 includes: Compare the number of nodes in the first satellite set with the number of nodes in the second satellite set; If the number of nodes in the first satellite set is less than the number of nodes in the second satellite set, then virtual nodes are added to the first satellite set until its number of nodes is equal to that of the second satellite set. If the number of nodes in the second satellite set is less than the number of nodes in the first satellite set, then virtual nodes are added to the second satellite set until its number of nodes is equal to that of the first satellite set. In a bipartite graph where the number of nodes on both sides is equal, set initial weights for all edges; For the edge between real satellite pairs, if the initial weight of the edge exceeds a preset threshold, the weight of the edge is updated to infinity to obtain the updated edge weight. Based on the updated edge weights, the Kuhn-Munkres algorithm is used to find the minimum weight perfect matching of the weighted bipartite graph, thus obtaining the globally optimal set of satellite matching pairs.

[0054] Set initial weights for all edges, including the edge weight between real satellite pairs which is the matching cost corresponding to the satellite pair, and the edge weight between virtual nodes and real nodes which is set to a preset penalty value. In this embodiment of the application, the minimum weight matching algorithm in step S300 constructs a bipartite graph by adding virtual nodes to the side with fewer nodes and setting penalty weights to achieve matching of node numbers. A weighted bipartite graph model is constructed based on the matching cost of real satellite pairs and virtual nodes. The Kuhn-Munkres algorithm is used to solve the minimum weight perfect matching of the graph to obtain the globally optimal set of satellite matching pairs.

[0055] In an optional implementation, the minimum weight matching algorithm in step S300 can also take the first satellite set and the second satellite set as two node sets of a bipartite graph, construct a network flow model and introduce a source and a sink, set the matching cost between satellites as the cost of the edge and set the capacity to 1, and obtain the optimal matching scheme by solving the minimum cost maximum flow from the source to the sink, thereby directly realizing the global optimal satellite matching under the condition of unequal number of nodes.

[0056] In an optional implementation, the minimum weight matching algorithm in step S300 can also encode the satellite matching relationship as a state in the solution space, take the total matching cost as the objective function, generate new solutions through random perturbation and accept or reject them according to the Metropolis criterion, and gradually reduce the temperature parameter to converge to the globally approximately optimal set of satellite matching pairs.

[0057] In this embodiment of the application, solving the minimum weight matching problem of the weighted bipartite graph model in step S400 includes formalizing the satellite matching problem into the classic assignment problem, i.e., finding a matching set. , making The problem requires that no two edges in the given graph share the same node, and that the sum of the weights of all matched edges be minimized. This problem can be solved in polynomial time using a mature combinatorial optimization algorithm, namely the Kuhn-Munkres algorithm.

[0058] In this embodiment of the application, in order to prevent matching with satellite pairs of extremely poor quality, a maximum allowable cost threshold is set. .any The edge weights are set to infinity, indicating that matching is not possible. If the two satellite sets have unequal numbers of satellites, a virtual node can be added to the set with fewer satellites. The edge weights between the virtual node and all real nodes are set to... Or a larger penalty value to achieve an imperfect match.

[0059] In this embodiment, the output of the algorithm in step S400 is the globally optimal set of satellite matching pairs. If a real node matches a virtual node, it means that the satellite has not found a suitable pair in this round of matching and should be discarded.

[0060] Example 3 is an embodiment of the present invention. This embodiment differs from the first embodiment in that it provides a satellite space matching system based on global optimization.

[0061] It should be noted that the technical solution of the satellite space matching system based on global optimization is based on the same concept as the technical solution of the satellite space matching method based on global optimization described above. For details not described in detail in the technical solution of the satellite space matching system based on global optimization in this embodiment, please refer to the description of the technical solution of the satellite space matching method based on global optimization described above.

[0062] This embodiment provides a satellite space matching system based on global optimization, comprising: The acquisition module is used to acquire satellite signal data, and through filtering and sorting, obtain the first set of satellites under the first environment and the second set of satellites under the second environment; The bipartite graph modeling module is used to construct a weighted bipartite graph model by treating the satellites in the first satellite set and the second satellite set as two node sets of the bipartite graph, and establishing connections between each pair of satellites from different sets. The matching cost calculation and weighting module is used to determine the weight of the edges connecting each satellite node by calculating the spatial geometric cost, signal physical cost and solution geometric contribution cost between each pair of satellite nodes based on the weighted bipartite graph model, and performing normalized weighted fusion. The global optimal matching solution module is used to solve the minimum weight matching problem of the weighted bipartite graph model based on the weights of the edges of each satellite node, and obtain the globally optimal set of satellite matching pairs through a combinatorial optimization algorithm.

[0063] This embodiment also provides an electronic device applicable to a satellite space matching method based on global optimization, comprising: The system includes a memory and a processor. The memory stores computer-executable instructions, and the processor executes these instructions to implement a satellite space matching method based on global optimization, as proposed in the above embodiments.

[0064] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a satellite space matching method based on global optimization as proposed in the above embodiments.

[0065] The storage medium proposed in this embodiment and the satellite space matching method based on global optimization proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0066] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0067] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A satellite space matching method based on global optimization, characterized in that, The method comprises the following steps: Satellite signal data is acquired, and through screening and sorting, a first satellite set in a first environment and a second satellite set in a second environment are obtained; A weighted bipartite graph model is constructed by taking the satellites in the first satellite set and the second satellite set as two node sets of a bipartite graph respectively, and establishing a connection between each pair of satellites from different sets; Based on the weighted bipartite graph model, the weight of the edge connecting each satellite node is determined by calculating the spatial geometric cost, signal physical cost and solving geometric contribution cost between each pair of satellite nodes, and then performing normalized weighted fusion; Based on the weight of each satellite node edge, the minimum weight matching problem of the weighted bipartite graph model is solved by a combinatorial optimization algorithm to obtain a globally optimal satellite matching pair set.

2. The global optimization based satellite space matching method of claim 1, wherein: The construction of the weighted bipartite graph model comprises: Based on the first satellite set and the second satellite set, the structural framework of the bipartite graph is constructed by defining the two sets as mutually exclusive node sets; Based on the structural framework of the bipartite graph, all possible potential matching pairs are enumerated by establishing a connection relationship between each satellite in the first satellite set and each satellite in the second satellite set, and a complete edge set is formed; Based on the edge set and the node set, the satellite state information of each node and the corresponding potential matching features of each edge are integrated to construct the weighted bipartite graph model.

3. The global optimization based satellite space matching method of claim 1 or 2, wherein: The determination of the weight of the edge connecting each satellite node comprises: The current candidate matching pair is taken as a satellite combination corresponding to a potential edge in the weighted bipartite graph model; Based on the candidate matching pair, the spatial geometric cost, signal physical cost and solving geometric contribution cost between the corresponding satellites are calculated in parallel to obtain three independent matching quality evaluation components; Based on the three types of costs and environmental parameters, the spatial geometric cost, signal physical cost and solving geometric contribution cost are normalized, and the weight coefficients of each cost are dynamically adjusted according to the current environment by using a fuzzy logic controller, and then weighted summation is performed; A dynamic cost function based on state prediction is introduced to integrate the current matching cost and the expected future matching cost, and the fusion weight of each cost component is dynamically adjusted according to the real-time environmental parameters in combination with the adaptive adjustment mechanism of the cost function weight; Based on the output of the dynamic cost function and the adaptively adjusted weight, weighted fusion is performed to finally determine the weight of the edge connecting each satellite node.

4. The global optimization based satellite space matching method of claim 3, wherein: The spatial geometric cost comprises: based on the azimuth and elevation angle information of the satellite in the first environment and the satellite in the second environment, the spatial angular distance between the satellite in the first environment and the satellite in the second environment is calculated; and the spatial angular distance is taken as the spatial geometric cost; The signal physical cost comprises: based on the signal carrier-to-noise ratio of the satellite in the first environment and the satellite in the second environment, the sum of the inverses of the signal carrier-to-noise ratio of the satellite in the first environment and the signal carrier-to-noise ratio of the satellite in the second environment is taken as the signal physical cost. The solving geometry contribution cost comprises: constructing a reference Jacobian matrix and a corresponding covariance matrix of a differential positioning system based on a high-confidence reference matching set, evaluating the degree of improvement of a geometric precision factor after introducing a candidate matching pair, determining the contribution of the candidate matching pair to positioning solution performance according to the degree of improvement, and taking the quantization result of the contribution as the solving geometry contribution cost.

5. The global optimization based satellite space matching method of claim 4, wherein: The state prediction-based dynamic cost function comprises: predicting the future state of the satellite in the first environment and the second environment based on an adaptive unscented Kalman filter, and obtaining the satellite position, velocity and signal strength change rate at multiple consecutive future time points based on the estimation result as a predicted future state sequence; according to the predicted future state sequence, calculating the expected matching cost of the candidate matching pair at each future time point; constructing a dynamic cost function, weighting and summing the matching cost at the current time and the expected matching cost at each future time, and introducing a discount factor to the long-term expected cost during the summation process; taking the calculation result of the dynamic cost function as the weight of the edge connecting the satellite nodes.

6. The global optimization based satellite space matching method of claim 5, wherein: The adaptive unscented Kalman filter comprises: state covariance update in Joseph form; an adaptive mechanism based on the Sage-Husa algorithm, which recursively estimates the process noise covariance matrix and the observation noise covariance matrix online by monitoring the filter innovation sequence; real-time adjustment of the forgetting factor in the Sage-Husa algorithm according to the normalized innovation square; the initial values of the process noise covariance matrix and the observation noise covariance matrix comprise constructing an optimal experimental design protocol based on the Fisher information matrix, screening and collecting the observation data segment with the largest information amount, and using the expectation maximization algorithm combined with a fixed interval smoother to identify the parameters of the noise covariance until convergence.

7. The global optimization based satellite space matching method of claim 6, wherein: The adaptive adjustment of the cost function weight comprises: inputting the average carrier-to-noise ratio of the visible satellite, the total number of visible satellites and the multipath effect index as environmental input parameters into the fuzzy logic controller; performing inference operation based on the preset fuzzy rule base through the fuzzy logic controller to dynamically output the weight coefficients corresponding to the spatial geometry cost, the signal physical cost and the solving geometry contribution cost.

8. A satellite space matching system based on global optimization, applying the method according to any one of claims 1 to 7, characterized in that, comprises: an acquisition module for acquiring satellite signal data, and obtaining a first satellite set in a first environment and a second satellite set in a second environment through screening and sorting; a bipartite graph modeling module for constructing a weighted bipartite graph model by taking the satellites in the first satellite set and the second satellite set as two node sets of the bipartite graph respectively, and establishing a connection between each pair of satellites from different sets; a matching cost calculation and weighting module for determining the weight of the edge connecting the satellite nodes based on the weighted bipartite graph model by calculating the spatial geometry cost, the signal physical cost and the solving geometry contribution cost between each pair of satellite nodes, and performing normalized weighted fusion; a global optimal matching solving module for solving the minimum weight matching problem of the weighted bipartite graph model based on the weight of each satellite node edge through a combinatorial optimization algorithm to obtain a globally optimal satellite matching pair set.

9. An electronic device, comprising: comprises: a memory for storing programs; a processor for loading said program to perform the steps of the method according to any of claims 1-7.

10. A computer-readable storage medium storing a program, characterized in that, said program, when executed by a processor, implements the steps of the method according to any of claims 1-7.

Images