Low-orbit satellite multi-source sensor data fusion processing method and system

Abstract

This invention discloses a method and system for fusion processing of multi-source sensor data from low-Earth orbit (LEO) satellites, belonging to the field of data processing technology. The method includes the following steps: Based on the remaining resources of the LEO satellite, dynamically assess bandwidth efficiency according to the sensor characteristic parameters and individual bandwidth occupancy rates of the multi-source sensors; perform adaptability screening on the raw data of the LEO satellite multi-source sensors based on mission requirements, and obtain valid data and their corresponding sensors to be fused based on the adaptability screening results; perform dynamic data compression processing on the valid data based on the dynamic bandwidth efficiency assessment results and link bandwidth occupancy rates; dynamically adjust the fusion weights based on the adaptability parameters of the sensors to be fused and the adaptability screening results; perform data fusion based on the adjusted sensor fusion weights and valid data; and self-optimize the fusion weights based on the fusion effect.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for fusion processing of multi-source sensor data from low-orbit satellites. Background Technology

[0002] Low Earth Orbit (LEO) satellite constellations, with their advantages of global coverage, low latency, and high revisit rate, have become a core platform for Earth observation and global information acquisition. These satellites typically carry multi-source sensors, including high-resolution optical cameras, synthetic aperture radar, infrared and spectrometers, enabling them to simultaneously acquire multi-dimensional and multi-scale information about the Earth's surface. However, single data sources have limitations, making data fusion technology crucial. This technology collaboratively processes heterogeneous data from different satellites and sensors, complementing and enhancing information at the information level to generate more complete, accurate, and reliable analytical results. This technological system is profoundly driving the transformation of decision-making models in fields such as Earth science, environmental monitoring, disaster prevention and mitigation, smart cities, and national defense towards real-time, global, and intelligent approaches.

[0003] Existing multi-source sensor data fusion processing technologies for low-Earth orbit satellites begin with data-level preprocessing, including radiometric and geometric correction of data from different sources such as optical, SAR, and infrared. High-precision ephemeris and attitude data are then used to unify the heterogeneous data to a single spatiotemporal reference through spatiotemporal registration, a prerequisite for effective fusion. This is followed by feature-level and decision-level fusion: through artificial intelligence and deep learning algorithms, common features (such as target contours and change information) from different data sources are automatically extracted and correlated. Alternatively, based on physical models and statistical methods, multi-source information is comprehensively interpreted to form more reliable and complete thematic information products than single data sources. Finally, on-orbit edge computing and satellite-ground collaborative processing architectures are becoming increasingly important. By performing initial fusion and filtering at the satellite end, only key information is transmitted, greatly improving the timeliness of data processing and reducing the pressure on ground stations, enabling rapid transformation from massive amounts of raw data into real-time usable knowledge.

[0004] For example, the Chinese invention patent with announcement number CN119538579B discloses a fusion positioning algorithm based on low-Earth orbit satellite Doppler measurement, DME, and VOR. This algorithm includes: acquiring the Doppler frequency shift of low-Earth orbit satellites, the slant range information of the DME, and the azimuth data of the VOR; constructing a combined observation equation and obtaining the corresponding Jacobian matrix; using a genetic algorithm based on the geometric accuracy factor to select stations and satellites, optimizing the selection of DME, VOR, and low-Earth orbit satellites to obtain the optimal combination of navigation sources; constructing motion models under different flight modes; and using interactive multi-model Kalman filtering to dynamically estimate the aircraft's state and update its position information in real time.

[0005] For example, the Chinese invention patent with announcement number CN119089377B discloses a data fusion method and system based on ERA5 reanalysis data and satellite observation data, which includes: first, converting the altitude coordinates of the reanalysis dataset from air pressure values ​​to corresponding altitude values ​​to facilitate subsequent matching and fusion; then, dividing the reanalysis dataset after altitude coordinate conversion into a gridded model according to a preset resolution, and calculating and fusing the parameters of the reanalysis data and observation data in units of spatiotemporal grids; for each spatiotemporal grid, first calculating the mean and standard deviation of the reanalysis data of the spatiotemporal grid, and then using the matched satellite observation data to calculate the mean and standard deviation of the observation data, and finally determining the gridded model of the fused data.

[0006] However, in the process of implementing the inventive technical solution in the embodiments of this application, it was found that the above-mentioned technology has at least the following technical problems:

[0007] In existing technologies, during the on-orbit target identification process of low-Earth orbit satellites, when resources are abundant, the lack of dynamic control over bandwidth allocation may lead to redundant transmission of some sensor data, interfering with the extraction of effective features. When resources are scarce, bandwidth constraints may force excessive data compression, resulting in the loss of key target features and the inclusion of invalid data in the fusion process. At the same time, it also limits the computational depth of dynamic adjustment of fusion weights, making it difficult to optimize weight allocation based on the real-time performance of sensors, resulting in an imbalance in the contribution of each sensor's data and causing low data fusion accuracy. Summary of the Invention

[0008] To address the technical problem of decreased accuracy in multi-source data fusion during on-orbit target identification of low-Earth orbit (LEO) satellites due to differences and dynamic limitations in on-board resources, this invention provides a method and system for processing multi-source sensor data fusion for LEO satellites. The technical solution is as follows:

[0009] On the one hand, a method for fusion processing of multi-source sensor data from low-Earth orbit (LEO) satellites is provided. Based on the remaining resources of LEO satellites, bandwidth efficiency is dynamically evaluated according to the sensor characteristic parameters and individual bandwidth occupancy rates of the multi-source sensors. Based on mission requirements, the raw data from the LEO satellite multi-source sensors undergoes adaptability screening, and effective data and their corresponding sensors to be fused are obtained based on the adaptation screening results. Based on the dynamic bandwidth efficiency evaluation results and link bandwidth occupancy rates, the effective data undergoes dynamic data compression processing. Based on the adaptability parameters and adaptation screening results of the sensors to be fused, the fusion weights are dynamically adjusted, and data fusion is performed based on the adjusted sensor fusion weights and effective data. Finally, the fusion weights are self-optimized based on the fusion effect.

[0010] On the other hand, a low-Earth orbit (LEO) satellite multi-source sensor data fusion processing system is provided, including: a performance evaluation module, a data filtering module, a data compression module, and a data fusion module. The performance evaluation module dynamically evaluates bandwidth performance based on the remaining resources of the LEO satellite, according to the sensor characteristic parameters of the multi-source sensors and their individual bandwidth occupancy rates. The data filtering module performs adaptability filtering on the raw data from the LEO satellite multi-source sensors based on mission requirements, and obtains valid data and their corresponding sensors to be fused based on the adaptability filtering results. The data compression module dynamically compresses the valid data based on the dynamic bandwidth performance evaluation results and link bandwidth occupancy rates. The data fusion module dynamically adjusts the fusion weights based on the adaptability parameters of the sensors to be fused and the adaptability filtering results, performs data fusion based on the adjusted sensor fusion weights and valid data, and self-optimizes the fusion weights based on the fusion effect.

[0011] Beneficial effects

[0012] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0013] 1. This invention integrates dynamic assessment of onboard resources and communication status, multi-dimensional data screening oriented towards mission objectives, differentiated compression strategies based on link adaptation, and fusion weight adjustment and self-optimization mechanisms based on data quality into a collaborative closed-loop processing flow. This intelligently ensures the transmission quality and integrity of high-value data under resource-constrained conditions and ensures that the data on which the fusion algorithm is based has high mission relevance and credibility. In this way, it achieves simultaneous optimization and improvement of the accuracy, reliability, and overall resource utilization efficiency of multi-source sensor data fusion under the dynamic constraints of low-Earth orbit satellites.

[0014] 2. This invention combines the inherent characteristics of sensors with dynamic satellite remaining resources and real-time bandwidth occupancy rates, and uses a series of preset mapping tables for quantitative matching and coupling calculations. This generates personalized bandwidth occupancy rate thresholds for each sensor that dynamically change with the system state, and distinguishes them into two types of states: the first type and the second type. This enables refined and differentiated perception and identification of the bandwidth occupancy efficiency of multi-source sensors, providing a precise and objective basis for subsequent implementation of differentiated data compression or bandwidth control strategies for different types of sensors when resources are scarce.

[0015] 3. By establishing multi-dimensional quantitative matching standards for time, space, and observation angle, and calculating the actual matching degree parameters of each sensor data based on specific task objectives, observation data that simultaneously meets basic quality requirements and is highly relevant to the current task is selected at the source, and the sensors providing these data are identified as units to be fused. This achieves task adaptability purification and refinement of multi-source heterogeneous input data, effectively eliminating interference from invalid or low-relevance information, and establishing a high-quality, highly relevant input set for subsequent data compression and fusion processing, laying the data foundation for improving the accuracy and reliability of the final fusion results.

[0016] 4. This invention establishes a multi-level link bandwidth occupancy threshold linked to the system's operating mode, and adopts differentiated response strategies based on different occupancy ranges. This includes optimizing the transmission mode when resources are plentiful, limiting the bandwidth of non-critical sensors and calculating the overload rate when resources are strained, and then dynamically calculating and constraining the data compression ratio of each sensor by combining sensor priority and overload rate. This enables differentiated and refined compression control of data from sensors of different importance when link resources are scarce. Consequently, it maximizes the overall bandwidth utilization while strictly ensuring the accuracy of high-priority sensor data, ensuring reliable and efficient transmission of key observation information under limited channel conditions, and providing quality-controlled data input for backend fusion.

[0017] 5. This invention comprehensively considers the multi-dimensional fit between sensor data and the current task, its long-term reliability, and its predetermined importance. It quantifies this data into fusion weight adjustment values ​​and coefficients using a mapping mechanism, and dynamically synthesizes the baseline weights to assign appropriate initial weights for each fusion task. Furthermore, by establishing an adaptive feedback adjustment mechanism based on the consistency of fused data and using abnormal fluctuations in weights or data as diagnostic clues, the system can automatically identify and correct weight allocation inaccuracies caused by instantaneous sensor performance fluctuations or model mismatches. This closed-loop process ultimately achieves full-cycle dynamic and fine-grained management of fusion weights from initial allocation and real-time adjustment to continuous optimization, significantly improving the adaptability, robustness, and overall reliability of the output results of the multi-source data fusion system. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0019] Figure 1 A flowchart of a low-orbit satellite multi-source sensor data fusion processing method provided in this application embodiment;

[0020] Figure 2 This is a flowchart of the dynamic data compression processing method for the low-orbit satellite multi-source sensor data fusion processing method provided in the embodiments of this application;

[0021] Figure 3 This is a schematic diagram of the structure of a low-orbit satellite multi-source sensor data fusion processing system provided in an embodiment of this application. Detailed Implementation

[0022] The technical solution provided in this application will now be described with reference to the accompanying drawings.

[0023] To facilitate understanding of the embodiments of this application, the following description is provided first:

[0024] In this application, the use of prefixes such as "first" and "second" is solely for the purpose of distinguishing different things belonging to the same category, and does not constrain the order, size, or quantity of things. For example, "first message" and "second message" are simply different messages, and there is no chronological, size, or priority relationship between them.

[0025] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0026] like Figure 1 The diagram shows a flowchart of a low-Earth orbit (LEO) satellite multi-source sensor data fusion processing method provided in this application embodiment. The method includes the following steps: Based on the remaining resources of the LEO satellite, dynamically assess bandwidth efficiency according to the sensor characteristic parameters and individual bandwidth occupancy rates of the multi-source sensors. This dynamic bandwidth efficiency assessment represents a dynamic evaluation of the bandwidth occupancy of each sensor under satellite resource conditions, providing a performance basis for subsequent data compression strategies. Based on mission requirements, perform adaptability screening on the raw data from the LEO satellite multi-source sensors, and obtain valid data and their corresponding sensors to be fused based on the adaptability screening results. This adaptability screening represents the selection of valid data that meets the mission objectives and the corresponding sensors to be fused. The system integrates sensors to establish a high-quality input foundation for subsequent data compression strategies. Based on dynamic bandwidth efficiency assessment results and link bandwidth occupancy, it performs dynamic data compression processing on the effective data. This dynamic data compression processing represents a real-time processing mechanism that dynamically adjusts the data compression strategy to match the current low-Earth orbit satellite communication channel conditions and mission requirements. Based on the adaptability parameters and adaptability screening results of the sensors to be fused, the system dynamically adjusts the fusion weights. Data fusion is then performed based on the adjusted sensor fusion weights and the effective data, and the fusion weights are self-optimized based on the fusion effect. This dynamic adjustment of fusion weights represents dynamically adjusting the fusion weights of the sensors to be fused to optimize the overall quality and reliability of the fused data.

[0027] In this embodiment, when performing data fusion based on the adjusted sensor fusion weights and valid data, the valid data of each sensor to be fused is first preprocessed. Through time synchronization calibration, spatial registration, and data format standardization, fusion barriers caused by data heterogeneity are eliminated, laying a consistent foundation for subsequent fusion. Using the dynamically adjusted and optimized fusion weights as the core basis, an appropriate fusion algorithm is selected based on the type and characteristics of the valid data: For numerical monitoring data (such as temperature and humidity), a weighted fusion method is used, allocating data contribution according to the weight ratio of each sensor; the higher the weight of the sensor data, the larger the weight coefficient in the fusion calculation. For feature-based data (such as target contours and spectral features), a feature-layer fusion algorithm is used, extracting core features from high-weight sensor data as the fusion leader through weight guidance, while low-weight sensor data features serve as supplementary verification. For decision-based data… (For example, target recognition results) Algorithms such as DS evidence theory are used to transform weights into evidence credibility, achieving weighted fusion of multi-source decisions. During the fusion process, the comprehensive index of observation matching degree of each sensor data and historical accuracy are introduced in real time as dynamic correction factors. If the real-time fluctuation of a certain sensor data exceeds the threshold, its weight ratio can be temporarily adjusted to reduce interference. Finally, the fusion result is double-verified. On the one hand, the deviation value between the fusion result and the effective data of each sensor is calculated to ensure that the deviation is within the preset reasonable range. On the other hand, the practicality of the fusion result is verified in combination with the mission objective requirements (such as whether the target positioning accuracy and monitoring data error meet the requirements). If the standard is not met, the weight adjustment logic and data preprocessing stage are traced back until the fusion data with accuracy, reliability and mission adaptability is output, providing the final data support for core tasks such as low-orbit satellite on-orbit target recognition.This invention provides a dynamic bandwidth efficiency assessment based on satellite remaining resources, sensor characteristic parameters, and individual bandwidth occupancy rates. This breaks the limitations of traditional fixed bandwidth allocation models, enabling real-time quantification of the bandwidth occupancy rationality of each sensor under different resource conditions. This provides precise performance data for subsequent compression strategies, effectively avoiding bandwidth waste when resources are abundant and insufficient bandwidth for critical sensors when resources are scarce, thus improving the refined scheduling level of onboard resources. Based on a mission-requirement-based adaptive screening mechanism, it extracts effective data and corresponding sensors to be fused by focusing on mission objectives, eliminating redundant and interfering data irrelevant to the mission from the data source. This establishes a high-quality data input foundation for subsequent processing, significantly reducing the negative impact of invalid data on the fusion results and reducing the computational overhead of subsequent compression and fusion processing. Combining the bandwidth efficiency assessment results with dynamic compression processing based on link bandwidth occupancy rates, the compression strategy can be flexibly adjusted according to the real-time status of the satellite communication channel and mission requirements, even when link resources are abundant. The system employs a low compression ratio to ensure data integrity, and balances data volume with key feature retention at a reasonable compression ratio when link load is high or resources are scarce. This avoids the problems of over-compression leading to feature loss or under-compression consuming bandwidth caused by fixed compression strategies, achieving dynamic adaptation of data transmission to channel conditions. Based on the adaptability parameters of the sensors to be fused and the screening results, the fusion weights are dynamically adjusted. Combined with a weight self-optimization mechanism driven by the fusion effect, the weight allocation can be adjusted in real time according to the fit between each sensor data and the task and the data reliability. This allows sensor data with high adaptability and high accuracy to contribute more in the fusion process. At the same time, the weight parameters are continuously optimized through fusion effect feedback, effectively solving the drawback of treating good and bad data equally in traditional fixed-weight fusion. This significantly improves the overall quality, reliability, and matching degree of fused data with mission objectives, ultimately achieving efficient processing and accurate fusion of multi-source sensor data from low-Earth orbit satellites under complex resource conditions, providing stable and reliable data support for core tasks such as on-orbit target identification.

[0028] Furthermore, the sensor characteristic parameters include data dimension, data transmission volume, and sensor priority level. Based on the remaining resources of the low-Earth orbit satellite, the steps for dynamically evaluating bandwidth efficiency according to the sensor characteristic parameters and individual bandwidth occupancy rates of multiple sensors include: matching the sensor priority level of each sensor with a preset standard occupancy rate mapping table to obtain the corresponding baseline bandwidth occupancy rate threshold; the standard occupancy rate mapping table defines the baseline bandwidth occupancy rate thresholds corresponding to different sensor priority levels; matching the remaining resources of the low-Earth orbit satellite with a preset operation mode mapping table to obtain the corresponding system operation mode; the operation mode mapping table defines the system resource scheduling strategy mode corresponding to different remaining resource ranges, and obtaining the corresponding preset bandwidth adjustment ratio based on the system operation mode; and matching the remaining resources of each sensor with a preset standard bandwidth occupancy rate mapping table to obtain the corresponding system operation mode. The data dimensions and data transmission volume are matched with preset first and second bandwidth adjustment mapping tables to obtain corresponding first and second bandwidth utilization adjustment values. Each second bandwidth adjustment mapping table defines the impact of different data dimensions and data transmission volume levels on bandwidth utilization. The coupling processing results of the first and second bandwidth utilization adjustment values ​​are weighted using the bandwidth adjustment ratio to obtain the bandwidth utilization adjustment value for each sensor. The bandwidth utilization adjustment values ​​of each sensor are then summed with the corresponding baseline bandwidth utilization thresholds to obtain the individual utilization thresholds for each sensor. Based on the bandwidth utilization adjustment values ​​of each sensor and the individual utilization thresholds, bandwidth performance is dynamically evaluated.

[0029] In this embodiment, the present invention establishes a priority-oriented bandwidth allocation framework by matching sensor priority levels with a standard occupancy mapping table to determine a baseline bandwidth occupancy threshold. This ensures that high-priority sensors, such as those undertaking core target identification tasks, receive reasonable bandwidth resource guarantees, preventing critical tasks from being affected by bandwidth allocation imbalances. This approach aligns resource allocation with core task requirements from the outset. Furthermore, by matching the remaining satellite resources with an operational mode mapping table to determine the system's operational mode and corresponding bandwidth adjustment ratio, bandwidth assessment is deeply integrated with the actual satellite resource status. This overcomes the drawbacks of traditional fixed bandwidth assessments being disconnected from resources. When satellite resources are abundant, bandwidth restrictions can be appropriately relaxed to ensure data transmission integrity; when resources are scarce, the assessment standard is tightened through adjustment ratios, achieving dynamic adaptation between bandwidth assessment and resource status, and improving the flexibility and targeting of onboard resource utilization. Finally, by matching sensor data dimensions and data transmission volume with a dedicated adjustment mapping table to obtain quantitative adjustment values, the present invention achieves precise consideration of the sensor's own data characteristics, solving the problem of prioritizing priority over data characteristics in traditional assessments. For example, it allows for targeted adjustments to sensors with high data dimensions and large transmission volumes. This approach avoids the failure to identify sensors with abnormal bandwidth usage due to their own data attributes, making bandwidth assessment more aligned with the actual operational needs of each sensor. By weighting the coupled processing results of the dual-adjustment values ​​through bandwidth adjustment ratios, and then summing them with a baseline threshold to obtain an individual occupancy rate threshold, a differentiated assessment standard is formed between the baseline threshold and personalized adjustments. This provides each sensor's bandwidth occupancy assessment with its own quantitative basis, replacing the traditional, coarse-grained approach of uniform threshold assessment. It can accurately identify the first category of sensors with individual bandwidth occupancy rates exceeding the threshold and the compliant second category of sensors, providing a clear and quantitative basis for subsequent bandwidth optimization and data compression strategies. The entire assessment process, through the combination of multiple mapping tables and quantitative calculations, transforms abstract factors such as satellite resource status, sensor priority, and data characteristics into calculable and comparable specific parameters. This upgrades bandwidth performance assessment from experience-based judgment to data-driven assessment, significantly improving the objectivity, accuracy, and reliability of the assessment results. This provides solid performance support for the formulation of subsequent data compression strategies, effectively reducing over- or under-compression problems caused by inaccurate bandwidth assessments, and laying an important foundation for the efficient operation of the entire multi-source sensor data processing workflow.

[0030] Furthermore, the step of dynamically evaluating bandwidth performance based on the bandwidth occupancy adjustment value of each sensor and the individual occupancy threshold includes: if the individual bandwidth occupancy of any sensor exceeds the individual occupancy threshold, it is marked as a first-class sensor; if the individual bandwidth occupancy of any sensor does not exceed the individual occupancy threshold, it is marked as a second-class sensor.

[0031] In this embodiment, the present invention transforms the abstract quantitative indicators previously calculated through multi-dimensional parameters into concrete and directly applicable sensor classification results. This allows the originally complex bandwidth performance evaluation to be grounded in the application level from the data level, solving the problem of the disconnect between quantitative evaluation results and actual processing actions. Bandwidth performance evaluation is no longer a simple numerical calculation, but a practical judgment that can directly guide subsequent operations. By using a clear binary judgment standard, the vague qualitative descriptions of reasonable and unreasonable in traditional evaluations are replaced, eliminating subjective biases from human judgment and ensuring that the classification of all sensors is based on unified and objective quantitative criteria. This significantly improves the consistency and reliability of the evaluation results and avoids subsequent processing chaos caused by unclear classification standards. The classification method of marking sensors exceeding the threshold as Category 1 and compliant sensors as Category 2 provides a clear target orientation for subsequent bandwidth optimization and data processing strategies, enabling subsequent operations to accurately focus on Category 1 sensors with abnormal bandwidth usage, avoiding... This invention avoids the waste of resources and inefficiency caused by indiscriminate processing of all sensors. For example, bandwidth compression or resource scheduling optimization can be prioritized for the first type of sensors, while the existing state of the second type of sensors can be maintained to reduce unnecessary processing overhead. The classification results of this invention build an efficient bridge between problem location and precise policy implementation in the entire data processing flow. It can quickly screen out sensors with bandwidth occupancy risks, so that subsequent bandwidth adjustment, data compression and other strategy formulation do not need to repeat the analysis of complex evaluation parameters. Targeted processing can be quickly initiated based on the classification results, which greatly improves the process response efficiency. This classification step is essentially a purification and simplification of the bandwidth performance evaluation results. While retaining the core evaluation value, it reduces the information processing cost of subsequent links, enabling technicians or automated processing systems to grasp the bandwidth status of each sensor with very low cognitive load. This provides simple and reliable decision support for the efficient and orderly operation of the entire low-orbit satellite multi-source sensor data fusion processing flow.

[0032] Furthermore, the steps of adapting the raw data from low-Earth orbit satellite multi-source sensors to the mission requirements, and obtaining effective data and their corresponding sensors to be fused based on the adaptability screening results, include: obtaining preset observation matching degree evaluation parameters, including time matching degree threshold, spatial matching degree threshold, and observation angle matching degree threshold; determining the mission target data and corresponding target observation parameters based on mission requirements, including the mission-related time range data (clarifying the start and end times of the mission execution, time intervals, and other key time constraints to match the acquisition time of the raw sensor data), the mission-covered spatial range data (covering the geographical boundaries of the geographical areas of interest to the mission, spatial resolution requirements, and other spatial constraints, corresponding to the geographical location information of the raw sensor data acquisition), and the core observation object data required by the mission (i.e., specific target-related information that the mission needs to focus on monitoring or acquiring data for). The process involves several steps: First, the target data is analyzed. This includes identifying target type and target characteristic parameters to preliminarily determine whether the raw data contains the observation objects of interest to the mission. Target observation parameters include mission time data, mission spatial data, and mission angle data. Second, the raw data from multiple low-Earth orbit satellite sensors is initially screened based on the mission target data, removing data that does not conform to the mission target data range to obtain valid mission data. Third, based on the target observation parameters, the actual observation matching degree parameters for each valid mission data are calculated. These parameters include time matching degree, spatial matching degree, and observation angle matching degree. Fourth, the actual observation matching degree parameters for each valid mission data are compared with their corresponding observation matching degree evaluation parameters. If the actual observation matching degree parameters of any valid mission data all meet the corresponding observation matching degree evaluation parameters, the valid mission data is marked as valid data; otherwise, it is marked as invalid data. Finally, the sensors corresponding to the valid data are marked as sensors to be fused.

[0033] In this embodiment, the present invention provides a standardized and quantifiable basis for data screening by clearly defining the three core dimensions of the observation matching degree evaluation parameters and setting preset thresholds. This replaces the fuzzy judgment mode that relies on experience in traditional screening, completely solving the problem of the lack of a unified standard for whether data meets task requirements. This makes the screening results highly objective and reproducible, effectively avoiding the omission of valid data or the misselection of invalid data due to differences in subjective judgment. By anchoring the target data and observation parameters with task requirements as the starting point, a task-oriented screening logic is constructed, so that data screening is no longer limited to the performance of the sensor itself, but closely revolves around the actual needs of core tasks such as on-orbit target identification. The requirements were outlined to ensure that the selected valid data closely matched the mission objectives, preventing redundant data irrelevant to the mission from entering subsequent processing and significantly improving the accuracy of data-mission compatibility. A dual-screening mechanism of initial screening and precise comparison formed an efficient hierarchical data purification path. Initial screening quickly eliminated raw data completely inconsistent with the mission objectives, significantly narrowing the scope of subsequent processing and reducing the consumption of limited onboard computing and storage resources. Precise comparison based on actual observation matching parameters further refined the control of data quality, ensuring the final data quality through comprehensive verification of time synchronization, spatial overlap, and the rationality of observation angles. The high-quality attributes of valid data provide a solid input foundation for subsequent data fusion. By clearly labeling valid data and corresponding sensors to be fused, precise correlation between data and sensors is achieved. This allows subsequent operations such as data compression and weight adjustment to directly focus on the sensors to be fused and their data, avoiding invalid processing of non-target sensors and improving the targeting and efficiency of the entire data processing flow. This screening step, by eliminating invalid data and corresponding sensors, not only reduces the computational load of subsequent data transmission and fusion but also avoids interference from invalid data with the fusion results. Noise, biases, and other issues that may be contained in invalid data are completely isolated, ensuring the reliability of the fusion results from the data input level. At the same time, the clear identification of the sensors to be fused provides a clear scope for subsequent dynamic adjustment of fusion weights, enabling weight resources to be concentrated on qualified sensors with valid data, further optimizing the fusion effect. The entire screening process forms a complete link from task requirements to data output, transforming abstract task requirements into specific data screening indicators. This achieves deep coupling between task requirements and data processing, allowing the low-orbit satellite multi-source sensor system to flexibly adjust screening criteria according to different tasks, enhancing the system's task adaptability and environmental responsiveness, and providing solid data support for accurate target identification in complex on-orbit scenarios.

[0034] like Figure 2The diagram shows a dynamic data compression process flowchart for the low-Earth orbit satellite multi-source sensor data fusion processing method provided in this application embodiment. The steps for dynamically compressing effective data based on bandwidth efficiency dynamic evaluation results and link bandwidth occupancy include: matching the system operating mode with a preset link bandwidth threshold mapping table to obtain corresponding link bandwidth threshold parameters. These parameters include a first link bandwidth occupancy threshold and a second link bandwidth occupancy threshold. The link bandwidth threshold mapping table defines the link bandwidth threshold parameters corresponding to different system operating modes. If the current link bandwidth occupancy of the low-Earth orbit satellite does not reach the first link bandwidth occupancy threshold, then the link resources are deemed sufficient, and the link bandwidth is compressed. The difference between the current link bandwidth utilization rate of the low-Earth orbit satellite and the first threshold is marked as the link bandwidth margin value. This value is matched against a preset transmission mode mapping table, and the data transmission mode of the communication link is adjusted to the matched target data transmission mode. The transmission mode mapping table defines different data transmission modes corresponding to different link bandwidth margin value ranges. If the current link bandwidth utilization rate of the low-Earth orbit satellite reaches the first threshold but does not exceed the second threshold, the link resource is considered to be under normal load, and the current data transmission mode and data compression ratio are maintained without additional processing. If the current link bandwidth utilization rate of the low-Earth orbit satellite exceeds the second threshold, the link resource is considered to be under normal load, and the current data transmission mode and data compression ratio are maintained without additional processing. If the second threshold of bandwidth occupancy is reached, the link resources are determined to be under strain. The coupling processing result of the individual bandwidth occupancy rate of the second type of sensor and the preset bandwidth margin value is marked as the corresponding bandwidth allocation value, and the bandwidth allocated to that sensor is adjusted to the bandwidth allocation value. The difference between the current link bandwidth occupancy rate of the low-orbit satellite and the second threshold of link bandwidth occupancy rate is marked as the link bandwidth overload rate. Based on the sensor priority level and the link bandwidth overload rate, the data compression ratio of the sensors to be fused is dynamically adjusted, and the effective data corresponding to the sensors to be fused is compressed based on the adjusted data compression ratio. First, for each sensor to be fused, the adjusted compression ratio is combined with the type characteristics of its effective data (such as images, etc.). For spectral and structured monitoring data, select appropriate compression algorithms: For image data, algorithms that support controllable distortion, such as JPEG2000, can be used; for structured data, lossless or low-loss compression algorithms, such as LZ77 / LZ78, can be used to achieve precise matching between the algorithm and data attributes. During compression, set algorithm parameters based on the adjusted compression ratio. For example, for effective data from high-priority sensors, if the adjusted compression ratio is low, set the algorithm parameters to high-fidelity mode to retain core information such as target contours and feature parameters. For data from low-priority sensors with a high compression ratio, balance compression efficiency and basic data integrity in parameter settings to avoid the loss of key related information.Simultaneously, the system collects real-time data volume changes and distortion indicators during the compression process, comparing the compressed data volume with the target link bandwidth usage. If the expected results are not met, algorithm parameters (such as quantization step size and encoding depth) are fine-tuned without exceeding the core constraints of the adjusted compression ratio. If the distortion exceeds the threshold corresponding to the sensor data accuracy requirements, the current compression is immediately terminated, and the rationality of the compression ratio is reviewed. The compressed data undergoes integrity and feature consistency verification to ensure that there are no packet losses or garbled characters, and that key observation features (such as target coordinates and monitoring value fluctuation ranges) remain consistent with the valid data before compression. Ultimately, compressed data that meets bandwidth requirements and satisfies subsequent fusion quality needs provides reliable support for the efficient transmission and accurate fusion of multi-source sensor data from low-orbit satellites.

[0035] In this embodiment, the present invention associates the system operation mode with the link bandwidth threshold mapping table, deeply binding the threshold for determining link bandwidth occupancy with the satellite resource status. This avoids the problem of traditional fixed threshold determination being out of touch with actual resource scenarios, providing an adaptability premise for the formulation of compression strategies under different resource conditions, and ensuring that the determination results are more in line with the actual operating conditions of the satellite. For scenarios with ample link resources, the data transmission mode is adjusted by calculating the bandwidth margin value and matching it with the transmission mode mapping table. This breaks the extensive mode of indiscriminate transmission when resources are abundant, and can select a better transmission mode (such as high-fidelity transmission or high-efficiency transmission) based on the margin size. This improves transmission efficiency while ensuring data integrity and avoids the waste of idle bandwidth resources. When the link resources are under normal load, the current status quo is maintained, which avoids the computational power consumption caused by unnecessary adjustments to the transmission mode and compression ratio, and ensures the stability of the data processing flow, achieving the goal of optimal high-efficiency operation without interference. In the case of tight link resources, the bandwidth allocation value is first adjusted through the bandwidth coupling processing results of the second type of sensor, accurately releasing the redundant bandwidth of compliant sensors to free up resource space for critical data transmission, and then based on the link... The compression ratio is dynamically adjusted based on bandwidth overload rate and sensor priority. This approach quantifies compression demand through overload rate to effectively alleviate bandwidth pressure, while prioritizing data from high-level sensors to ensure the compression ratio does not exceed limits. This prevents feature loss due to over-compression of core data, solving the problem of critical data corruption caused by a one-size-fits-all approach in traditional compression. The entire processing flow is based on both real-time link status and prior bandwidth performance assessments, upgrading data compression from static preset to dynamic response. It can quickly adapt to real-time fluctuations in link bandwidth. For example, compression adjustment can be initiated rapidly when the link shifts from ample to strained, and excessive intervention can be stopped promptly when it returns to normal. This significantly improves the system's adaptability to complex link environments. Through precise responses to different link states and dynamic optimization of the compression ratio, the system effectively controls link bandwidth usage while maximizing the preservation of core features of valid data. This provides a reliable guarantee for high-quality fusion of subsequent sensor data to be fused, avoiding the problem of decreased fusion accuracy due to improper compression. It also achieves efficient utilization of onboard link resources and dual protection for core mission data transmission, significantly improving the reliability and resource adaptability of the low-Earth orbit satellite multi-source sensor data processing system.

[0036] Furthermore, the steps for dynamically adjusting the data compression ratio of the sensors to be fused based on sensor priority level and link bandwidth overload rate include: obtaining the data accuracy requirements corresponding to each sensor priority level; determining the data compression ratio threshold for each sensor to be fused based on the data accuracy requirements; matching the link bandwidth overload rate with a preset compression ratio adjustment mapping table to obtain the corresponding compression ratio adjustment coefficient, wherein the compression ratio adjustment mapping table defines the compression ratio adjustment coefficient corresponding to different link bandwidth overload rate ranges; multiplying the current data compression ratio of each sensor to be fused using the compression ratio adjustment coefficient to obtain the target data compression ratio for each sensor to be fused; for each sensor to be fused, determining whether its target data compression ratio exceeds its corresponding data compression ratio threshold; if so, adjusting the data compression ratio of the sensor to its corresponding data compression ratio threshold; otherwise, adjusting the data compression ratio of the sensor to its corresponding target data compression ratio.

[0037] In this embodiment, the present invention directly links sensor priority levels with data accuracy requirements and clearly defines compression ratio thresholds, constructing a rigid constraint that prioritizes accuracy. This completely breaks the drawback of traditional compression strategies that prioritize bandwidth over accuracy, ensuring that high-priority sensors always have compression guarantees no lower than the threshold, avoiding the loss of key data features due to over-compression, and fundamentally safeguarding the data quality baseline of core tasks. By matching the link bandwidth overload rate with the compression ratio adjustment mapping table to obtain quantitative adjustment coefficients, the abstract bandwidth pressure is transformed into a calculable adjustment basis, so that the adjustment of the compression ratio no longer relies on experience judgment, but achieves precise quantification based on real-time bandwidth load. For example, a high overload rate corresponds to a high adjustment coefficient to strengthen the compression force, and a low overload rate corresponds to a low adjustment coefficient to weaken the intervention, ensuring the targeted and effective relief of bandwidth pressure. The present invention adopts a process of calculating the target compression ratio by multiplying the adjustment coefficients and verifying the threshold, realizing the organic combination of flexible adjustment and rigid constraints: flexible adjustment enables it to quickly adapt to bandwidth load fluctuations, while threshold verification... This invention sets a ceiling for the compression ratio for each sensor to prevent the target compression ratio from exceeding the accuracy limit due to excessive bandwidth overload, thus providing dual protection for core data. The invention implements differentiated compression strategies for different priority levels and bandwidth load scenarios, providing dedicated accuracy-first guarantees for high-priority sensors while offering flexible bandwidth adjustment space for low-priority sensors. This avoids the dual problems of core data damage or insufficient bandwidth optimization caused by traditional one-size-fits-all compression. The entire adjustment process is centered on data accuracy requirements and guided by real-time bandwidth status, deeply coupling the sensor task value with the link resource status. This upgrades compression ratio adjustment from passive response to active adaptation, effectively alleviating link bandwidth pressure and finding the optimal balance between bandwidth optimization and data quality. This provides crucial support for the integrity and reliability of subsequent data to be fused, while reducing fusion errors caused by improper compression ratios, further improving the operational efficiency and task adaptability of the entire low-orbit satellite multi-source sensor data processing system.

[0038] Furthermore, the steps for dynamically adjusting the fusion weights based on the historical accuracy and adaptability screening results of the sensors to be fused include: obtaining the observation matching degree parameters of the corresponding sensors based on the actual observation matching degree parameters of the effective data; classifying and assigning each effective data point to the corresponding sensor to be fused according to its acquisition source, forming an effective dataset within each sensor to be fused; then directly assigning the actual observation matching degree parameters corresponding to each effective data point to the observation matching degree parameters of the sensor to be fused corresponding to that effective data point, thus obtaining the observation matching degree parameters of the sensors to be fused; evaluating and obtaining the corresponding fusion weight adjustment values ​​based on the observation matching degree parameters of each sensor to be fused; obtaining... The historical data accuracy of each sensor to be fused is used to match the historical data accuracy with a preset weight adjustment coefficient mapping table to obtain the corresponding fusion weight adjustment coefficient. The weight adjustment coefficient mapping table defines the fusion weight adjustment coefficient corresponding to different historical data accuracy ranges. The corresponding baseline fusion weight is obtained based on the sensor priority level of each sensor to be fused. The fusion weight adjustment value of each sensor to be fused is multiplied by the fusion weight adjustment coefficient of each sensor to obtain the final adjustment value of each sensor to be fused. Then, the final adjustment value of each sensor to be fused is used to process the corresponding baseline fusion weight to obtain the fusion weight of each sensor to be fused.

[0039] In this embodiment, the present invention obtains sensor observation matching parameters based on the actual observation matching parameters of effective data and derives fusion weight adjustment values, deeply binding weight adjustment with the task adaptability of the sensor's current data. This solves the problem of traditional fixed weights being detached from real-time data quality. For example, sensors with higher time, space, and angle matching are assigned higher adjustment values, ensuring that weight allocation is tilted towards data with stronger current task adaptability, thus improving the fit between weight and real-time task requirements. The invention also introduces the historical data accuracy of sensors and converts it into quantitative adjustment coefficients through a weight adjustment coefficient mapping table, incorporating the sensor's past performance into the weight calculation system. This avoids the one-sidedness of determining weights solely based on single data performance: sensors with high historical accuracy can still obtain relatively reasonable weights through adjustment coefficients even if the single data matching is slightly lower, while sensors with poor historical performance will have their weights appropriately reduced, ensuring the reliability of weight allocation from a long-term performance perspective. Finally, the invention establishes a baseline fusion weight based on sensor priority, constructing the core bottom line for weight allocation. This ensures that high-priority sensors undertaking critical tasks maintain their basic weight advantage in multi-factor adjustments and do not lose their due weight ratio due to real-time matching fluctuations or minor deviations in historical accuracy, effectively protecting the core data. The system plays a dominant role in data fusion. The final adjustment value is obtained by multiplying the adjustment coefficient and the adjustment value, and then the baseline weight is adjusted. This transforms abstract sensor performance indicators into quantifiable weight parameters, replacing the traditional experience-based weight setting method. This ensures that the weight of each sensor has a clear calculation basis, enabling differentiated and personalized weight configuration. It avoids the problems of insufficient contribution from high-quality data and excessive interference from low-quality data caused by a one-size-fits-all approach. The entire adjustment process, through the quantitative coupling of multi-dimensional parameters and threshold constraints, allows the fusion weight to dynamically respond to the combined changes in real-time sensor performance, historical performance, and task priority. When sensor data adaptability improves and historical accuracy is optimized, the weight is reasonably increased; conversely, it is appropriately decreased, ensuring that the fusion weight is always in a dynamic balance that matches the actual performance of the sensors. Precise weight allocation directly optimizes the subsequent data fusion effect, allowing highly adaptable, highly reliable, and high-priority sensor data to play a dominant role in the fusion process. This minimizes the interference of low-quality data, significantly improving the overall quality and reliability of the fused data. This provides more accurate data support for core tasks such as low-orbit satellite on-orbit target identification, while also enhancing the intelligence and adaptability of the entire multi-source sensor data processing system.

[0040] Furthermore, the step of obtaining the corresponding fusion weight adjustment value based on the observation matching degree parameters of each sensor to be fused includes: obtaining preset observation matching degree weight parameters, which include time matching degree weight, spatial matching degree weight, and observation angle matching degree weight; performing weighted coupling processing on the observation matching degree parameters of each sensor to be fused using the observation matching degree weight parameters to obtain the comprehensive observation matching degree index of each sensor to be fused, which is used to quantitatively evaluate the comprehensive evaluation value of the overall fit between the sensor data and the current task; matching the comprehensive observation matching degree index of each sensor to be fused with a preset weight adjustment value mapping table to obtain the corresponding fusion weight adjustment value, which defines the fusion weight adjustment value corresponding to different ranges of comprehensive observation matching degree indexes.

[0041] The comprehensive index of observation matching degree of each sensor to be fused is obtained as follows:

[0042] ;

[0043] In the formula, This represents the comprehensive index of the observation matching degree of the i-th sensor to be fused. , and These represent the weights for time matching, spatial matching, and observation angle matching, respectively. , and Let represent the time matching degree, spatial matching degree, and observation angle matching degree of the i-th sensor to be fused, respectively, where i is the number of each sensor to be fused, i=1,2,3,...,N, and N is the total number of sensors to be fused.

[0044] In this embodiment, the invention decomposes the observation matching degree into three core dimensions: time, space, and observation angle. Each dimension has a preset basic weight, including the basic weight for time matching degree, the basic weight for space matching degree, and the basic weight for observation angle matching degree. The core of generating the observation matching degree weight parameters in this invention is to make targeted and flexible adjustments to the above-mentioned preset basic weights based on the characteristics of the on-orbit mission, forming observation matching degree weight parameters that are deeply consistent with the mission requirements. The specific adjustment logic is as follows: guided by the core objectives of the mission, the matching dimensions that play a key role in the achievement of the mission are first identified, and then the precise adaptation is achieved by increasing the proportion of the preset basic weight of the dimension and optimizing the weight allocation of other dimensions. Different missions have clear adjustment procedures. For example, target tracking tasks, with ensuring data temporal continuity as their core requirement, dynamically adjust the basic weight ratio of time matching based on a preset temporal continuity adjustment ratio. Then, the processed basic weight of time matching is normalized to other basic weights to obtain the final observation matching weight parameter, ensuring temporal consistency. Regional monitoring tasks, with ensuring accurate geographical coverage as their core requirement, normalize the basic weight of spatial matching to other basic weights based on a preset accuracy adjustment ratio to obtain the observation matching weight parameter, thereby improving geographic positioning accuracy. This differentiated weight adjustment method based on task characteristics solves the problem that the traditional one-size-fits-all evaluation model cannot adapt to diverse on-orbit tasks. It deeply binds the observation matching evaluation logic with the core requirements of different tasks, significantly improving the relevance and effectiveness of the evaluation.By weighting and coupling the matching parameters of each dimension, scattered single-dimensional indicators are integrated into a unified observation matching index, enabling a quantitative assessment of the fit between sensor data and the task. Qualitative judgments such as time synchronization and spatial overlap are transformed into calculable numerical indicators, completely avoiding the bias of subjective experience in traditional assessments and ensuring strong objectivity and comparability in the matching assessment. The construction of the comprehensive index effectively balances the influence of matching degrees in each dimension, preventing sensors with excellent performance in a single dimension but low overall matching degree from being misjudged. For example, if a sensor has extremely high time matching degree but excessive spatial deviation, a lower index will be obtained through comprehensive calculation, ensuring that the assessment results reflect the overall adaptation level of the sensor data. A weight adjustment value mapping table is used to convert the comprehensive index into corresponding fusion weight adjustments. This approach seamlessly integrates evaluation results with weight adjustments, transforming abstract comprehensive indices into concrete parameters directly applicable to weight calculations. This provides a precise and clear quantitative basis for the dynamic adjustment of subsequent fusion weights. The entire process, through standardized weight presets and mapping rules, ensures that the weight adjustment values ​​of different sensors are evaluated under a unified standard, avoiding imbalances in weight allocation caused by inconsistent evaluation scales. Simultaneously, the quantitative evaluation method allows the weight adjustment values ​​to be dynamically updated in real-time with changes in sensor observation matching, ensuring that weight adjustments always keep pace with fluctuations in data fit. This lays a solid foundation for subsequent high-quality data fusion, ultimately achieving a precise guidance where higher data fit corresponds to larger weight adjustment values. This directs fusion resources towards the sensors best suited to the task, significantly improving the relevance and reliability of data fusion.

[0045] Furthermore, the steps for self-optimizing fusion weights based on the fusion effect include: calculating the average deviation between the data of each sensor to be fused and the fused data; matching the average deviation with a preset deviation threshold; if the average deviation exceeds the deviation threshold, monitoring the fluctuation of the fusion weights of each sensor; if the fluctuation frequency of the fusion weight of any sensor exceeds a preset first fluctuation frequency threshold, marking the difference between the weight fluctuation frequency and the first fluctuation frequency threshold as the first deviation fluctuation frequency; matching the first deviation fluctuation frequency with a preset first adjustment ratio mapping table to obtain the corresponding adjustment coefficient adjustment ratio; and multiplying the fusion weight adjustment coefficient of the sensor based on the adjustment coefficient adjustment ratio. The first adjustment ratio mapping table defines the adjustment coefficients corresponding to different first deviation fluctuation frequency ranges. The adjustment ratio is calculated as follows: If the fluctuation frequency of the fusion weights of each sensor does not exceed the preset fluctuation frequency threshold, the fluctuation of the data of each sensor is monitored. If the fluctuation frequency of any sensor data exceeds the preset second fluctuation frequency threshold, the difference between the data fluctuation frequency and the second fluctuation frequency threshold is marked as the second deviation fluctuation frequency. The second deviation fluctuation frequency is matched with the preset second adjustment ratio mapping table to obtain the corresponding adjustment value adjustment ratio. Based on the adjustment value adjustment ratio, the fusion weight adjustment value of the sensor is multiplied. The second adjustment ratio mapping table defines the adjustment value adjustment ratio corresponding to different second deviation fluctuation frequency ranges. Otherwise, a data fusion anomaly prompt is issued, and the preset personnel are notified to check the sensor equipment. If the average deviation value does not exceed the deviation threshold, no additional processing is performed.

[0046] In this embodiment, the present invention uses the average deviation between the data from each sensor and the fused data as the core quantitative indicator of the fusion effect. This transforms the abstract fusion quality into a calculable and comparable numerical basis, replacing the traditional, crude method of relying on manual judgment of the fusion effect. This makes the fusion effect evaluation highly objective and accurate, providing a clear trigger condition for weight optimization and avoiding problems such as unfounded optimization or missed detection of poor-quality fusion. The present invention establishes a hierarchical investigation logic for deviation threshold verification and fluctuation source tracing. When the average deviation exceeds the threshold, weight fluctuations are monitored first, followed by data fluctuations, accurately locating the root cause of poor fusion effect—whether it is due to dynamic imbalance in weight allocation or insufficient stability of the sensor data itself. This solves the drawback of blindly adjusting weights in traditional optimization, ensuring that optimization measures directly address the core problem. For cases where weight fluctuations exceed the limit, the first deviation fluctuation frequency is calculated and matched with a first adjustment ratio mapping table, converting the degree of weight fluctuation into a quantitative adjustment coefficient ratio. This allows for precise correction of the fusion weight adjustment coefficient, effectively suppressing fusion instability caused by excessive weight fluctuations and avoiding rigid weight imbalance caused by a one-size-fits-all correction. The system is designed to better reflect actual fluctuations. When data fluctuations exceed a threshold, the system uses a second deviation fluctuation frequency to match a second adjustment ratio mapping table to obtain an adjustment value. This allows for targeted optimization of the fusion weight adjustment value, indirectly mitigating the interference of unstable data on the fusion results and achieving the effect of offsetting data fluctuations through weight optimization. In scenarios where both weights and data fluctuations are compliant but deviations still exceed limits, an anomaly alert is issued. This organically combines autonomous system processing with manual intervention, preventing the continuous deterioration of fusion anomalies caused by hidden problems such as sensor hardware failures, thus improving the system's fault response capability and reliability. When the deviation value is compliant, the system maintains the status quo, avoiding unnecessary weight adjustments that consume onboard computing power and achieving the goal of energy-efficient optimization on demand. The entire self-optimization process upgrades the fusion weights from static configuration to dynamic adaptation, enabling continuous response to changes in fusion effects and sensor states. It continuously optimizes the weight allocation scheme, significantly improving the stability, accuracy, and reliability of the fused data. This provides adaptive data quality assurance for core tasks such as low-orbit satellite on-orbit target identification, further enhancing the intelligence level and environmental adaptability of the entire multi-source sensor data processing system.

[0047] like Figure 3The diagram shows the structure of a low-Earth orbit satellite multi-source sensor data fusion processing system provided in this application embodiment. It includes: a performance evaluation module, a data filtering module, a data compression module, and a data fusion module. The performance evaluation module dynamically evaluates bandwidth performance based on the remaining resources of the low-Earth orbit satellite, according to the sensor characteristic parameters and individual bandwidth occupancy rates of the multi-source sensors. This dynamic bandwidth performance evaluation indicates the dynamic assessment of the bandwidth occupancy of each sensor under satellite resource conditions, providing a performance basis for subsequent data compression strategies. The data filtering module performs adaptability filtering on the raw data from the low-Earth orbit satellite multi-source sensors based on mission requirements, and obtains valid data and their corresponding sensors to be fused based on the adaptability filtering results. Adaptability filtering indicates the selection of data that meets the requirements. The effective data of the mission objective and the corresponding sensors to be fused establish a high-quality input foundation for subsequent data compression strategies. The data compression module is used to dynamically process the effective data based on the dynamic evaluation results of bandwidth efficiency and link bandwidth utilization. The dynamic data compression processing represents a real-time processing mechanism that dynamically adjusts the data compression strategy to match the current low-Earth orbit satellite communication channel conditions and mission requirements. The data fusion module is used to dynamically adjust the fusion weights based on the adaptability parameters and adaptability screening results of the sensors to be fused. It then performs data fusion based on the adjusted sensor fusion weights and effective data, and self-optimizes the fusion weights based on the fusion effect. The dynamic adjustment of fusion weights represents the dynamic adjustment of the fusion weights of the sensors to be fused to optimize the overall quality and reliability of the fused data.

[0048] The various features and processes described above can be used independently of each other or can be combined in various ways. All possible combinations and sub-combinations are intended to fall within the scope of this disclosure. Furthermore, certain method or process blocks may be omitted in some embodiments. The methods and processes described herein are not limited to any particular order, and the blocks or states associated with them may be performed in other suitable orders. For example, the described blocks or states may be performed in an order different from the order specifically disclosed, or multiple blocks or states may be combined in a single block or state. Example blocks or states may be performed serially, in parallel, or in some other manner. Blocks or states may be added to or removed from the disclosed example embodiments. The exemplary systems and components described herein may be configured differently from those described. For example, elements may be added to, removed from, or rearranged compared to the disclosed example embodiments.

[0049] The various operations of the example methods described herein can be performed at least in part by an algorithm. This algorithm can be contained in program code or instructions stored in memory (e.g., the aforementioned non-transitory computer-readable storage medium). Such an algorithm may include a machine learning algorithm. In some embodiments, the machine learning algorithm may not be explicitly programmed into the computer to perform the function, but can learn from training data to create a predictive model that performs the function.

[0050] The various operations of the example methods described herein can be performed, at least in part, by one or more processors that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors can constitute the engine of a processor implementation that operates to perform one or more of the operations or functions described herein.

[0051] Similarly, the methods described herein can be implemented at least in part by a processor, where one or more specific processors are examples of hardware. For example, at least some operations of a method can be performed by one or more processors or an engine implemented by a processor. Furthermore, one or more processors can also be operated to support the performance of related operations in a “cloud computing” environment or as “Software as a Service” (SaaS). For example, at least some operations can be performed by a set of computers (as an example of a machine including processors), where these operations are accessible via a network (e.g., the Internet) and via one or more suitable interfaces (e.g., application programming interfaces (APIs)).

[0052] The performance of certain operations can be distributed across processors, residing not only within a single machine but also deployed across multiple machines. In some example embodiments, the processor or processor-implemented engine may reside in a single geographic location (e.g., within a home environment, office environment, or server cluster). In other example embodiments, the processor or processor-implemented engine may be distributed across multiple geographic locations.

[0053] In this specification, multiple instances may implement components, operations, or structures described as single instances. Although individual operations of one or more methods are shown and described as separate operations, one or more of the separate operations may be performed simultaneously and do not need to be performed in the order shown. Structures and functions presented as separate components in the example configuration may be implemented as composite structures or components. Similarly, structures and functions presented as single components may be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of this document.

[0054] While an overview of the subject matter has been described with reference to specific example embodiments, various modifications and changes can be made to these embodiments without departing from the broader scope of embodiments of this disclosure. Such embodiments of the subject matter are referred to herein, individually or collectively, by the term "invention," and are used for convenience only and are not intended to limit the scope of this application to any single disclosure or concept, should more than one disclosure or concept be disclosed in fact.

[0055] The embodiments described herein have been described in sufficient detail to enable those skilled in the art to practice the disclosed teachings. Other embodiments may be used and derived therefrom, such that structural and logical substitutions and changes may be made without departing from the scope of this disclosure. Therefore, the detailed description should not be construed as limiting, and the scope of the various embodiments is defined only by the appended claims and the full scope of their equivalents.

Claims

1. A method for fusing and processing multi-source sensor data from low-orbit satellites, characterized in that, Includes the following steps: Based on the remaining resources of low-orbit satellites, bandwidth efficiency is dynamically evaluated according to the sensor characteristic parameters and individual bandwidth occupancy rates of multi-source sensors. The dynamic evaluation of bandwidth efficiency means dynamically evaluating the bandwidth occupancy of each sensor under satellite resource conditions, providing an efficiency basis for subsequent data compression strategies. Based on mission requirements, the raw data from multiple sources of low-orbit satellite sensors are subjected to adaptability screening, and the effective data and their corresponding sensors to be fused are obtained based on the adaptability screening results. The adaptability screening means screening effective data and corresponding sensors to be fused that meet the mission objectives, thus establishing a high-quality input foundation for subsequent data compression strategies. The steps of performing adaptability screening on the raw data from low-Earth orbit satellite multi-source sensors based on mission requirements, and obtaining valid data and their corresponding sensors to be fused based on the adaptability screening results, include: Obtain preset observation matching degree evaluation parameters, which include time matching degree threshold, spatial matching degree threshold and observation angle matching degree threshold; Based on the task requirements, the task target data and corresponding target observation parameters are determined. The target observation parameters include task time data, task space data, and task angle data. Based on the mission objective data, the raw data from the multi-source sensors of the low-orbit satellite were initially screened to obtain effective mission data; Based on the target observation parameters, the actual observation matching degree parameters of each effective task data are calculated. The actual observation matching degree parameters include time matching degree, spatial matching degree and observation angle matching degree. The actual observation matching degree parameter of each valid task data is compared with its corresponding observation matching degree evaluation parameter. If the actual observation matching degree parameter of any valid task data reaches the corresponding observation matching degree evaluation parameter, the valid task data is marked as valid data; otherwise, it is marked as invalid data. Mark the sensors corresponding to valid data as the sensors to be fused; Based on the dynamic evaluation results of bandwidth efficiency and the link bandwidth utilization rate, the effective data is dynamically compressed. The dynamic data compression process refers to a real-time processing mechanism that dynamically adjusts the data compression strategy to match the current low-orbit satellite communication channel conditions and mission requirements. The fusion weights are dynamically adjusted based on the adaptability parameters and adaptability screening results of the sensors to be fused. Data fusion is performed based on the adjusted sensor fusion weights and effective data. The fusion weights are then self-optimized based on the fusion effect. The dynamic adjustment of the fusion weights means dynamically adjusting the fusion weights of the sensors to be fused to optimize the overall quality and reliability of the fused data.

2. The low-orbit satellite multi-source sensor data fusion processing method as described in claim 1, characterized in that: The sensor characteristic parameters include data dimension, data transmission volume, and sensor priority level; The steps for dynamically evaluating bandwidth efficiency based on the remaining resources of low-Earth orbit satellites, according to the sensor characteristic parameters of multi-source sensors and individual bandwidth occupancy rates, include: The sensor priority level of each sensor is matched with the preset standard occupancy mapping table to obtain the corresponding baseline bandwidth occupancy threshold. The remaining resources of low-orbit satellites are matched with a preset operation mode mapping table to obtain the corresponding system operation mode, and the corresponding preset bandwidth adjustment ratio is obtained based on the system operation mode. The data dimensions and data transmission volume of each sensor are matched with the preset first bandwidth adjustment mapping table and second bandwidth adjustment mapping table respectively to obtain the corresponding first bandwidth utilization rate adjustment value and second bandwidth utilization rate adjustment value. The bandwidth utilization rate adjustment values ​​of the first and second adjustment values ​​are weighted using the bandwidth adjustment ratio to obtain the bandwidth utilization rate adjustment values ​​of each sensor. The bandwidth occupancy rate adjustment value of each sensor is used to process the corresponding baseline bandwidth occupancy rate threshold to obtain the individual occupancy rate threshold of each sensor. Dynamic evaluation of bandwidth performance is performed based on the bandwidth occupancy adjustment values ​​of each sensor and the individual occupancy threshold.

3. The low-orbit satellite multi-source sensor data fusion processing method as described in claim 2, characterized in that: The steps for dynamically evaluating bandwidth performance based on the bandwidth occupancy adjustment values ​​of each sensor and the individual occupancy threshold include: If the individual bandwidth utilization rate of any sensor exceeds the individual utilization rate threshold, it is marked as a Class I sensor; If the individual bandwidth occupancy rate of any sensor does not exceed the individual occupancy rate threshold, it is marked as a second-class sensor.

4. The low-orbit satellite multi-source sensor data fusion processing method as described in claim 1, characterized in that: The steps for dynamically compressing effective data based on bandwidth efficiency dynamic evaluation results and link bandwidth utilization include: The system operating mode is matched with a preset link bandwidth threshold mapping table to obtain the corresponding link bandwidth threshold parameters, which include a first threshold for link bandwidth utilization and a second threshold for link bandwidth utilization. If the current link bandwidth utilization rate of the low-orbit satellite does not reach the first threshold of link bandwidth utilization rate, the difference between the first threshold of link bandwidth utilization rate and the current link bandwidth utilization rate of the low-orbit satellite is marked as the link bandwidth margin value. The link bandwidth margin value is matched with the preset transmission mode mapping table, and the data transmission mode of the communication link is adjusted to the matched target data transmission mode. If the current link bandwidth utilization rate of a low-orbit satellite reaches the first threshold of link bandwidth utilization rate but does not exceed the second threshold of link bandwidth utilization rate, no additional processing will be performed. If the current link bandwidth occupancy rate of a low-orbit satellite exceeds the second threshold of link bandwidth occupancy rate, the coupling processing result of the individual bandwidth occupancy rate of the second type of sensor and the preset bandwidth margin value will be marked as the corresponding bandwidth allocation value, and the bandwidth allocated to the sensor will be adjusted to the corresponding bandwidth allocation value. The difference between the current link bandwidth occupancy rate of the low-orbit satellite and the second threshold of the link bandwidth occupancy rate is marked as the link bandwidth overload rate. The data compression ratio of the sensor to be fused is dynamically adjusted based on the sensor priority level and the link bandwidth overload rate, and the effective data corresponding to the sensor to be fused is compressed based on the adjusted data compression ratio.

5. The low-orbit satellite multi-source sensor data fusion processing method as described in claim 4, characterized in that: The step of dynamically adjusting the data compression ratio of the sensors to be fused based on sensor priority and link bandwidth overload rate includes: Obtain the data accuracy requirements corresponding to the priority level of each sensor, and determine the data compression ratio threshold of each sensor to be fused based on the data accuracy requirements. The link bandwidth overload rate is matched with the preset compression ratio adjustment mapping table to obtain the corresponding compression ratio adjustment coefficient. The current data compression ratio of each sensor to be fused is processed using the compression ratio adjustment coefficient to obtain the target data compression ratio of each sensor to be fused. For each sensor to be fused, determine whether its target data compression ratio exceeds its corresponding data compression ratio threshold; If so, adjust the data compression ratio of the sensor to its corresponding data compression ratio threshold; Otherwise, adjust the data compression ratio of the sensor to its corresponding target data compression ratio.

6. The low-orbit satellite multi-source sensor data fusion processing method as described in claim 1, characterized in that: The step of dynamically adjusting the fusion weight based on the historical accuracy and adaptability screening results of the sensors to be fused includes: The observation matching degree parameters of the corresponding sensor are obtained based on the actual observation matching degree parameters of the effective data; The corresponding fusion weight adjustment value is obtained based on the evaluation of the observation matching degree parameters of each sensor to be fused; Obtain the historical data accuracy of each sensor to be fused, and match the historical data accuracy with a preset weight adjustment coefficient mapping table to obtain the corresponding fusion weight adjustment coefficient. The baseline fusion weight is obtained based on the sensor priority level of each sensor to be fused; The fusion weight adjustment value is processed by the fusion weight adjustment coefficient of each sensor to be fused to obtain the final adjustment value of each sensor to be fused. Then, the final adjustment value of each sensor to be fused is used to process the corresponding benchmark fusion weight to obtain the fusion weight of each sensor to be fused.

7. The low-orbit satellite multi-source sensor data fusion processing method as described in claim 6, characterized in that: The step of obtaining the corresponding fusion weight adjustment value based on the observation matching degree parameters of each sensor to be fused includes: Obtain preset observation matching degree weight parameters, which include time matching degree weight, spatial matching degree weight and observation angle matching degree weight; The observation matching degree parameters of each sensor to be fused are weighted and coupled using the observation matching degree weight parameter to obtain the observation matching degree comprehensive index of each sensor to be fused. The observation matching degree comprehensive index is used to quantitatively evaluate the overall fit between the sensor data and the current task. The observation matching degree comprehensive index of each sensor to be fused is matched with the preset weight adjustment value mapping table to obtain the corresponding fusion weight adjustment value.

8. The low-orbit satellite multi-source sensor data fusion processing method as described in claim 1, characterized in that: The steps for self-optimizing fusion weights based on fusion effect include: Calculate the average deviation between the data from each sensor to be fused and the fused data; The average deviation value is matched with a preset deviation threshold. If the average deviation value exceeds the deviation threshold, the fluctuation of the fusion weight of each sensor is monitored. If the fluctuation frequency of the fusion weight of any sensor exceeds a preset first fluctuation frequency threshold, the difference between the weight fluctuation frequency and the first fluctuation frequency threshold is marked as the first deviation fluctuation frequency. The first deviation fluctuation frequency is matched with a preset first adjustment ratio mapping table to obtain the corresponding adjustment coefficient adjustment ratio. The fusion weight adjustment coefficient of the sensor is processed based on the adjustment coefficient adjustment ratio. If the fluctuation frequency of the fusion weights of each sensor does not exceed the preset fluctuation frequency threshold, the fluctuation of the data of each sensor is monitored. If the fluctuation frequency of any sensor data exceeds the preset second fluctuation frequency threshold, the difference between the data fluctuation frequency and the second fluctuation frequency threshold is marked as the second deviation fluctuation frequency. The second deviation fluctuation frequency is matched with the preset second adjustment ratio mapping table to obtain the corresponding adjustment value adjustment ratio. The fusion weight adjustment value of the sensor is processed based on the adjustment value adjustment ratio. Otherwise, a data fusion abnormality prompt is issued. If the average deviation value does not exceed the deviation threshold, no additional processing is required.

9. A low-Earth orbit satellite multi-source sensor data fusion processing system, employing the low-Earth orbit satellite multi-source sensor data fusion processing method as described in any one of claims 1-8, characterized in that, include: Performance evaluation module, data filtering module, data compression module, data fusion module; The performance evaluation module is used to dynamically evaluate bandwidth performance based on the remaining resources of the low-orbit satellite, according to the sensor characteristic parameters and individual bandwidth occupancy of the multi-source sensors. The dynamic bandwidth performance evaluation represents the dynamic evaluation of the bandwidth occupancy of each sensor under satellite resource conditions, providing a performance basis for subsequent data compression strategies. The data filtering module is used to perform adaptability filtering on the raw data of low-orbit satellite multi-source sensors based on mission requirements, and to obtain effective data and their corresponding sensors to be fused based on the adaptability filtering results. The adaptability filtering means filtering effective data and corresponding sensors to be fused that meet the mission objectives, thus establishing a high-quality input foundation for subsequent data compression strategies. The steps of performing adaptability screening on the raw data from low-Earth orbit satellite multi-source sensors based on mission requirements, and obtaining valid data and their corresponding sensors to be fused based on the adaptability screening results, include: Obtain preset observation matching degree evaluation parameters, which include time matching degree threshold, spatial matching degree threshold and observation angle matching degree threshold; Based on the task requirements, the task target data and corresponding target observation parameters are determined. The target observation parameters include task time data, task space data, and task angle data. Based on the mission objective data, the raw data from the multi-source sensors of the low-orbit satellite were initially screened to obtain effective mission data; Based on the target observation parameters, the actual observation matching degree parameters of each effective task data are calculated. The actual observation matching degree parameters include time matching degree, spatial matching degree and observation angle matching degree. The actual observation matching degree parameter of each valid task data is compared with its corresponding observation matching degree evaluation parameter. If the actual observation matching degree parameter of any valid task data reaches the corresponding observation matching degree evaluation parameter, the valid task data is marked as valid data; otherwise, it is marked as invalid data. Mark the sensors corresponding to valid data as the sensors to be fused; The data compression module is used to perform dynamic data compression processing on effective data based on the dynamic evaluation results of bandwidth efficiency and the link bandwidth utilization rate. The dynamic data compression processing refers to a real-time processing mechanism that dynamically adjusts the data compression strategy to match the current low-orbit satellite communication channel conditions and mission requirements. The data fusion module is used to dynamically adjust the fusion weight based on the adaptability parameters and adaptability screening results of the sensors to be fused, perform data fusion based on the adjusted sensor fusion weights and effective data, and self-optimize the fusion weight based on the fusion effect. The dynamic adjustment of the fusion weight means dynamically adjusting the fusion weight of the sensors to be fused to optimize the overall quality and reliability of the fused data.

Images