Satellite communication terminal fast deployment and directional locking method for natural disaster rescue

Abstract

The application discloses a satellite communication terminal rapid deployment and directional locking method for natural disaster rescue, which comprises the following steps: acquiring sensor data and preprocessing, performing magnetic field interference mode classification, attitude estimation and filter-cluster double-loop feedback control, generating a compensation parameter set and feeding back to the magnetic field interference mode classification and the attitude estimation, and circularly executing until the accuracy threshold is met; based on the terminal attitude estimation value meeting the accuracy threshold and satellite position information, satellite direction positioning and locking are performed, and satellite pointing angle and pointing control instructions are outputted. The application solves technical problems such as large attitude estimation error, discontinuous parameter adjustment and lack of differentiated compensation of traditional methods in a magnetic field interference environment, and improves the deployment efficiency and communication reliability of the satellite communication terminal in a disaster area environment.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of satellite communication technology and automatic control, and particularly relates to a satellite communication terminal rapid deployment and directional locking method for natural disaster rescue. BACKGROUND

[0002] In a natural disaster rescue operation, the communication infrastructure is often severely damaged, and satellite communication as a key means of connecting the disaster area with the outside world becomes particularly important. The rapid deployment and accurate directional locking capability of the satellite communication terminal is directly related to the rescue efficiency and communication quality. Rescue personnel often face the challenge of establishing a stable communication link in extreme environments within a short period of time, so developing a lightweight satellite communication terminal attitude estimation and directional locking technology that adapts to complex magnetic field environments is of great significance to improving disaster response capabilities and rescue efficiency.

[0003] Currently, satellite communication terminal attitude estimation mainly uses complementary filtering or extended Kalman filtering methods to fuse multi-sensor data. The mainstream technology usually uses an accelerometer to provide gravity reference, a gyroscope to provide angular velocity, and an electronic compass to provide geomagnetic north reference, and achieves high-precision attitude estimation in a static environment. Many systems use low-pass filters to process accelerometer data and high-pass filters to process gyroscope data to reduce the influence of their respective noises. In commercial systems, the common AHRS (Attitude Heading Reference System) fuses sensor data through a nonlinear extended Kalman filter, constructs an attitude state equation and an observation equation, and can achieve accurate attitude estimation in a static environment.

[0004] However, in the complex magnetic field environment of the disaster area, the existing method faces many technical challenges. First, the traditional attitude estimation algorithm simply treats magnetic field interference as an abnormal value, lacks the ability to distinguish different types of magnetic field interference, and causes the azimuth angle estimation error to increase significantly in the area of steel reinforced concrete building debris or damaged power facilities. Second, the existing filtering and clustering methods usually use a one-way dependency relationship, and the filtering result is difficult to feedback to adjust the clustering process, causing inconsistent judgments in complex environments. Third, traditional parameter adjustment mostly uses step or linear changes, which easily causes attitude estimation to jump when the magnetic field interference mode switches, affecting the terminal stable pointing. In addition, the fixed weight sensor fusion method cannot dynamically adjust the strategy according to different magnetic field interference modes, resulting in poor performance of the system in different situations such as persistent environmental drift, sudden disturbance, and periodic disturbance. These problems seriously restrict the deployment efficiency and communication reliability of the satellite communication terminal in the disaster area with varying magnetic fields. SUMMARY

[0005] The application aims to provide a satellite communication terminal rapid deployment and directional locking method for natural disaster rescue, in order to solve at least one technical problem existing in the prior art.

[0006] The technical scheme is a satellite communication terminal rapid deployment and directional locking method for natural disaster rescue, comprising the following steps:

[0007] Obtaining various sensor data, preprocessing, and forming data packets;

[0008] Receiving the data packets, performing magnetic field interference mode classification to obtain a magnetic field interference mode identifier, and performing attitude estimation based on the magnetic field interference mode identifier to generate a terminal attitude estimation value and an uncertainty;

[0009] Based on the magnetic field interference mode identifier, the terminal attitude estimation value, and the uncertainty, performing filter-clustering dual-loop feedback control to generate a compensation parameter set containing a filter parameter adjustment set and a clustering parameter adjustment set; feeding back the clustering parameter adjustment set to the magnetic field interference mode classification and feeding back the filter parameter adjustment set to the attitude estimation, and performing a loop until a precision threshold is met;

[0010] Based on the terminal attitude estimation value meeting the precision threshold and satellite position information, performing satellite direction positioning and locking to output a satellite pointing angle and a pointing control instruction.

[0011] The present application solves the problems of large attitude estimation error, discontinuous parameter adjustment, and lack of differentiated compensation of traditional methods in a magnetic field interference environment, breaks the one-way dependence of traditional filter and clustering algorithms, establishes a bidirectional feedback mechanism, improves the deployment efficiency and communication reliability of satellite communication terminals in harsh environments, and provides reliable communication support for disaster rescue. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 A step flowchart of a satellite communication terminal rapid deployment and directional locking method for natural disaster rescue provided by the embodiment of the present application.

[0013] Figure 2 A step flowchart of performing magnetic field interference mode classification provided by the embodiment of the present application.

[0014] Figure 3 A step flowchart of using a core-satellite structure progressive incremental clustering algorithm to process enhanced magnetic field feature vectors provided by the embodiment of the present application.

[0015] Figure 4 A step flowchart of determining the minimum distance and the corresponding cluster provided by the embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to make the person skilled in the art better understand the technical scheme of the present application, the technical scheme in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the scope of protection of the present application.

[0017] It should be particularly pointed out that, in order to clearly show the step flow of the present application, serial numbers are marked for each step in the description. These serial numbers are only for the convenience of description, and do not limit the execution order of the steps. In actual operation, according to the technical needs of specific implementation scenarios, the steps can be executed in different order from that shown in the description, and in some cases, parallel processing between steps can also be realized.

[0018] As shown in Figure 1 A satellite communication terminal rapid deployment and directional locking method for natural disaster rescue, comprising the following steps:

[0019] S1, obtaining each sensor data, preprocessing, forming a data packet;

[0020] Specifically, each sensor data includes gyroscope raw data, accelerometer raw data and electronic compass raw data. The gyroscope raw data can measure the angular velocity (i.e. the speed of rotation) of the object; the accelerometer raw data can measure the acceleration (such as the change of direction and force) of the object; the electronic compass raw data can be the direction and orientation.

[0021] S2, receiving the data packet, classifying the magnetic field interference mode, and obtaining the magnetic field interference mode identifier;

[0022] Specifically, the magnetic field will be disturbed by the surrounding environment (such as electronic devices, magnetic materials), and these disturbances may cause inaccurate direction measurement. The system analyzes the data packet and finds out the characteristic pattern of these disturbances (for example, identifies the source or type of disturbance). Once the disturbance pattern is identified, it will be given a "label" or "identifier" to represent which type of disturbance (for example: is it from electronic devices, or due to environmental magnetic field anomalies).

[0023] S3, based on the magnetic field interference mode identifier, performing attitude estimation, generating terminal attitude estimation value and uncertainty;

[0024] Specifically, according to the information of the magnetic field interference, the system begins to calculate the current attitude of the terminal, such as the direction (facing which direction) and the angle (inclination, etc.). The core of the attitude estimation is to analyze the sensor data and compensate the influence of the magnetic field interference, so as to obtain the most accurate result as possible. While calculating the attitude estimation value, the system will evaluate the accuracy of the result. For example, it will generate an "uncertainty" value to indicate how much error the estimation may have.

[0025] S4, according to the magnetic field interference mode identification, the terminal attitude estimation value and the uncertainty, performing filter-cluster double-loop feedback control to generate a compensation parameter set containing a filter parameter adjustment set and a cluster parameter adjustment set; feeding back the cluster parameter adjustment set to the magnetic field interference mode classification and feeding back the filter parameter adjustment set to the attitude estimation, and performing the loop until the accuracy threshold is met;

[0026] Specifically, the filter-cluster double-loop feedback control is performed, in which the filter control focuses on removing the interference or noise in the data to make the attitude estimation result more accurate; the cluster control focuses on classifying and identifying the mode of the magnetic field interference to make the classification more reliable; according to the analysis, a set of adjustment parameters is generated, in which the filter parameter adjustment set is used to optimize the attitude estimation process; the cluster parameter adjustment set is used to improve the accuracy of the magnetic field interference classification. The adjusted parameters of the filter will be fed back to the attitude estimation module to continuously improve the accuracy of the direction estimation; the adjusted parameters of the cluster will be fed back to the magnetic field interference classification module to further improve the accuracy of the classification. This process will continue to loop until the results of the attitude estimation and the interference classification meet the preset accuracy threshold.

[0027] S5, based on the terminal attitude estimation value meeting the accuracy threshold and the satellite position information, performing satellite direction positioning and locking to output the satellite pointing angle and pointing control instruction.

[0028] Specifically, the system combines the attitude information of the device and the position of the satellite to calculate the direction to which the device needs to be adjusted to accurately align the satellite, and obtains the angle adjustment value and the pointing control instruction, which are used to drive the device to complete the physical direction adjustment and lock to the satellite.

[0029] The embodiment realizes real-time data acquisition and processing, enables the terminal to quickly adapt to complex magnetic field environment, improves the system response speed, continuously improves the accuracy and reliability of the terminal attitude estimation, ensures that the finally output satellite pointing angle has high precision, and thus improves the communication quality; realizes dynamic adjustment of different magnetic field interference modes and attitude estimation errors, enhances the self-adaptation ability of the system, and ensures that the system can flexibly cope with different environmental conditions. The embodiment has the characteristics of real-time, adaptability and high precision, and is particularly suitable for satellite communication demand in complex environment under natural disaster rescue scene.

[0030] According to an aspect of the present application, the step of preprocessing comprises:

[0031] S11, collect multi-sensor parallel data. Obtain gyroscope raw data at a sampling rate of 500 Hz, record three-axis angular velocity measurements; obtain accelerometer raw data at a sampling rate of 100 Hz, record three-axis acceleration measurements; obtain electronic compass raw data at a sampling rate of 20 Hz, record three-axis magnetic field intensity measurements; record the collection time stamp of each group of data to ensure time correlation in subsequent processing.

[0032] S12, perform sensor data calibration and noise removal. Read the sensor calibration parameters including the bias vector and the scale coefficient matrix, which are calibrated in advance; apply a temperature compensation model to the gyroscope raw data to generate temperature-compensated angular velocity data; apply the calibration parameters to the accelerometer raw data to generate calibrated acceleration data; apply hard iron and soft iron calibration to the electronic compass raw data to generate calibrated magnetic field data; apply a band-pass filter to all calibrated data to generate denoised sensor data.

[0033] S13, perform multi-sampling rate data synchronization and alignment. Design a time window alignment algorithm to establish a unified time reference based on a 100 Hz base sampling rate; downsample the temperature-compensated angular velocity data to generate 100 Hz angular velocity data; perform interpolation calculation on the calibrated magnetic field data to generate 100 Hz magnetic field data; integrate the sensor data streams to form time-aligned multi-sensor data packets containing angular velocity, acceleration and magnetic field data. Calculate the quality evaluation index of each time-aligned multi-sensor data packet, and mark the outliers.

[0034] As shown in Figure 2 According to an aspect of the present application, the step of classifying the magnetic field interference pattern to obtain the magnetic field interference pattern identification comprises:

[0035] S21, receive the data packet, the initial or feedback cluster parameter adjustment set, extract the time domain features of the magnetic field data, and construct the enhanced magnetic field feature vector;

[0036] S22, use the incremental clustering algorithm with core-satellite structure to process the enhanced magnetic field feature vector, and dynamically adjust the clustering threshold based on the cluster parameter adjustment set to obtain the magnetic field interference pattern identification and the clustering confidence score.

[0037] S23, analyze the magnetic field interference pattern identification and the clustering confidence score, evaluate the current magnetic field environment state, generate the magnetic field reliability index and the interference compensation suggestion; wherein the magnetic field interference pattern identification, the magnetic field reliability index and the interference compensation suggestion are used for subsequent adaptive parameter adjustment of the attitude estimation.

[0038] Specifically, the 100Hz magnetic field data in the time-aligned multi-sensor data packet is received, a 500ms sliding window is constructed, time domain features including mean, variance, maximum fluctuation rate and frequency features are extracted from the magnetic field data in the window; compared with the historical reference model, a magnetic field anomaly index is calculated to quantify the degree of deviation of the current magnetic field from the normal value; combined with the magnetic field measurement value and the extracted features, an enhanced magnetic field feature vector is constructed.

[0039] The enhanced magnetic field feature vector and the clustering state information of the previous period are read, the adaptive distance measure of the current feature vector and the known clustering pattern is calculated, and the most matched clustering pattern is identified. According to the matching degree, one of the following operations is performed: if the matching degree is high: update the core parameters of the corresponding cluster, and record the pattern matching identifier; if the matching degree is moderate: take the feature vector as a satellite point of the cluster, and update the cluster boundary; if the matching degree is low: temporarily store as a potential new pattern, and create a new cluster after accumulating enough samples; output the identified magnetic field interference pattern identifier and the clustering confidence score.

[0040] The features of the identified magnetic field interference pattern identifier are analyzed, including duration, intensity change trend and directionality. The current pattern features are matched with the predefined interference type templates to determine which type of interference (such as persistent environmental offset, sudden interference, etc.); according to the clustering confidence score and the pattern stability, a magnetic field reliability index is calculated; a disturbance compensation suggestion is generated, including an adjustment scheme for the filtering weight.

[0041] The embodiment constructs a core-satellite structure clustering model, solves the problem that traditional clustering methods cannot adapt to the characteristics of time series data, enables the clustering model to continuously adapt to changing environments without the need to store and process all historical data, and improves the clustering accuracy.

[0042] As shown in Figure 3 According to one aspect of the present application, the steps of using a progressive incremental clustering algorithm with a core-satellite structure to process the enhanced magnetic field feature vector to obtain the magnetic field interference pattern identifier and the clustering confidence score include:

[0043] For the enhanced magnetic field feature vector, the adaptive distance of the current feature vector and the existing clusters is calculated to determine the minimum distance and the corresponding cluster, which is recorded as the most matched cluster;

[0044] According to the minimum distance and the matching threshold in the cluster parameter adjustment set, the core-satellite structure update is performed, the current sample is assigned as a core update, a satellite point or a potential new pattern, and the cluster core parameters, the cluster boundary information or the potential new pattern buffer are updated accordingly;

[0045] When the preset conditions are met, the clustering maintenance and optimization are performed, the cluster set is updated according to the clustering activity and the similarity, the updated cluster set is generated, and the magnetic field interference pattern identifier of the current most matched cluster is determined.

[0046] calculating a cluster confidence score based on the updated set of cluster states.

[0047] Specifically, the calculation process of the adaptive distance in the progressive incremental clustering is: adaptive distance D(ct, ci) = ωd·Dd(ct, ci) + ωm·Dm(ct, ci) + ωt·Dt(ci); wherein Dd(ct, ci) = 1-cos(θ) is a direction-sensitive distance; θ is the included angle between the direction of the current feature vector ct and the direction of the cluster core ci; Dm(ct, ci) = |log(Mt / Mi)| is an intensity-sensitive distance; Mt is the current feature vector magnetic field strength; Mi is the cluster core magnetic field strength; Dt(ci) = 1-exp(-λt·ΔT) is a time attenuation weight coefficient; ΔT is the difference between the current time and the last update time of the cluster ci; λt is a time attenuation coefficient; ωd, ωm, ωt are the weights of the direction, intensity and time components respectively, and ωd+ωm+ωt=1.

[0048] As shown in FIG. 6, according to one aspect of the present application, the step of calculating the adaptive distance of the current feature vector and the existing clusters, and determining the minimum distance and the corresponding cluster includes: Figure 4

[0049] performing normalization processing on the enhanced magnetic field feature vector to generate a standardized feature vector containing a direction component and an intensity component;

[0050] calculating the difference between the current time and the last update time of each cluster based on the standardized feature vector, and applying an exponential decay function to generate a time attenuation weight coefficient;

[0051] calculating a direction-sensitive distance according to the cosine value of the included angle between the direction component and the direction of each cluster core;

[0052] calculating an intensity-sensitive distance according to the logarithmic ratio of the intensity component and the intensity of each cluster core;

[0053] combining the direction-sensitive distance, the intensity-sensitive distance and the time attenuation weight coefficient using a preset weight coefficient to generate an adaptive distance metric value;

[0054] comparing the adaptive distance metric values of all clusters to determine the minimum distance and the corresponding cluster.

[0055] According to one aspect of the present application, the step of performing core-satellite structure updating and updating the cluster core parameters, cluster boundary information or the potential new mode buffer accordingly includes:

[0056] comparing the minimum distance with a matching threshold to determine whether to perform a core update operation, a satellite point update operation or a potential new mode processing; ​

[0057] When performing the core update operation, the adaptive learning rate is calculated and the cluster core parameters, statistics and active timestamp are updated;

[0058] When performing the satellite point update operation, the current sample is recorded as a satellite point, the cluster boundary information and satellite point set are updated;

[0059] When performing the potential new pattern processing, the current sample is added to the temporary buffer, and when the samples in the temporary buffer meet the preset condition for creating a new cluster, a new cluster core is generated, and the potential new pattern buffer is updated.

[0060] Specifically, the core-satellite structure update mechanism is as follows: the core update learning rate a(ci) = β·exp(-γ·N(ci))·(1-Sc(ci)); wherein a(ci) is the adaptive learning rate of the cluster core ci; β is a basic learning rate constant; γ is a decay factor; N(ci) is the cumulative sample number of the cluster ci; Sc(ci) is the stability index of the cluster ci, with a value range of [0, 1]; the core parameter update formula is: Pnew(ci) = (1-a(ci))·Pold(ci) + a(ci)·Pcur; Pnew(ci) is the updated cluster core parameter; Pold(ci) is the cluster core parameter before updating; Pcur is the current sample feature parameter.

[0061] According to an aspect of the present application, the step of determining the magnetic field interference pattern identifier of the current most matched cluster in the cluster maintenance and optimization includes:

[0062] Performing cluster aging detection, calculating the inactivity time of each cluster, and marking the cluster whose inactivity time exceeds a predetermined threshold as a dormant state; performing cluster merging detection, calculating the cluster core distance based on the updated cluster core parameters, and when the distance between two cluster cores is less than a preset merging threshold, calculating a weighted average to generate a new core and merging the statistics and satellite point set; performing cluster structure rebalancing, periodically optimizing the cluster boundary and core distribution, reducing overlap and redundancy; and outputting the updated cluster set;

[0063] Based on the updated cluster set, the overall stability index of the current cluster model is calculated to determine the magnetic field interference pattern identifier of the current most matched cluster.

[0064] According to an aspect of the present application, the step of calculating the cluster confidence score based on the state of the updated cluster set includes:

[0065] Based on the updated cluster set, the sample number, internal variance and inactivity time are extracted, and the stability index of the matched cluster is calculated;

[0066] Based on the stability index and the minimum distance, a cluster confidence score quantifying the reliability of the current cluster identification result is generated and the interference mode features (key feature parameter set) of the matching cluster are extracted, including the duration, intensity change rate and direction stability, to generate the cluster feature parameter set.

[0067] Specifically, the process of calculating the cluster confidence score is: cluster confidence score Cfid(ci) = ws·Ss(ci) + wv·(1-Sv(ci)) + wa·Sa(ci); where Cfid(ci) is the confidence score of the cluster ci, the value range [0, 1]; Ss(ci) is the sample number standardization score of the cluster ci; Sv(ci) is the internal variance standardization score of the cluster ci; Sa(ci) is the activity score of the cluster ci; ws, wv, wa are the weights of the three components of sample number, internal variance and activity, and ws+wv+wa=1.

[0068] In an embodiment of the present application, an adaptive distance metric is calculated. Read the current enhanced magnetic field feature vector F t and the existing cluster set C = {C1, C2,..., C k}; apply normalization processing to each feature dimension to generate the standardized feature vector F t norm ; calculate the time decay weight coefficient w t = e -λ(tcurrent-tlast) , giving higher weight to recent samples. Define the direction-sensitive distance function: d dir (v_1, v_2) = 1 - |cos(v_1, v_2)|, capturing the change in magnetic field direction. Define the intensity-sensitive distance function: d mag (m_1, m_2) = |log(||m_1|| / ||m_2||)|, capturing the change in magnetic field intensity. Combine to build an adaptive distance metric: d(F t , C i ) =α⋅d dir (F t dir , C i dir ) +β⋅d mag (F t mag , C i mag ) +γ⋅d temporal (F t temp , C i temp ). Calculate the distance d(F t , C i), find the minimum distance d min and the corresponding cluster C min . Where e is the base of the natural logarithm, λ is the parameter of time decay, t current is the current time; t last is the last time point; v_1 and v_2 are the magnetic field direction feature vectors; m_1 and m_2 are the magnetic field intensity feature vectors; C_i is the i-th cluster in the existing cluster set; α is the weight coefficient of the direction-sensitive distance function; F t dir is the direction feature part in the current enhanced magnetic field feature vector; C i dir is the direction feature part of the i-th cluster in the cluster set; β is the weight coefficient of the intensity-sensitive distance function; F t mag is the intensity feature part in the current enhanced magnetic field feature vector; C i mag is the intensity feature part of the i-th cluster in the cluster set; γ is the weight coefficient of the time-sensitive distance function; d temporal is the time-sensitive distance function; F t temp is the time feature part in the current enhanced magnetic field feature vector; C i temp is the time feature part of the i-th cluster in the cluster set.

[0069] Core-satellite structure update. Determine the relationship between the minimum distance d min and the matching threshold θ match : if d min < θ match : execute core update operation; if θ match ≤ d min < θ satellite : execute satellite point update operation; if d min ≥ θ satellite : execute potential new mode processing, where θ satellite is the satellite point threshold. When executing the core update operation, calculate the learning rate η = η0⋅(1−e −Ni / τ ), where η0 is the initial learning rate, τ is the time constant, and N_i is the number of samples of the cluster C min ; update the cluster core parameters: C min core = (1−η)⋅C min core + η⋅F t norm ; update the cluster statistics: increase the sample count N_i, update the variance σ i 2 ; update the most recent active timestamp tlast When satellite point update operation is performed, record current enhanced magnetic field feature vector F t As a satellite point of cluster C min , update cluster boundary information: B min =max(B min , d min ); update satellite point set: S min =S min ∪{F t}; only keep the last K satellite points to prevent infinite memory growth. When potential new pattern processing is performed, add current enhanced magnetic field feature vector F t to temporary buffer T buffer , if the average mutual distance of points in temporary buffer T buffer is less than new pattern threshold θ new and the number exceeds minimum number threshold n min : create a new cluster C new , the initial core is the average of points in temporary buffer T buffer ; set initial sample count, variance and timestamp; add new cluster to cluster set: C=C∪{C new}; empty temporary buffer T buffer .

[0070] Carry out cluster maintenance and optimization. Perform cluster aging detection: calculate the inactivity time t inactive =t current −t last of each cluster; if t inactive >T max : mark the cluster as dormant state, not participating in regular matching but retained in the library, where T max is the maximum threshold of cluster inactivity time. Perform cluster merging detection: if the distance between the cores of two clusters is less than merging threshold θ merge : calculate the weighted average to get a new core; merge statistics and satellite point set; delete the original two clusters and add the merged cluster. Calculate the overall stability index S model of the current cluster model; output the magnetic field interference pattern identification M i of the current best matching cluster.

[0071] Carry out incremental cluster result evaluation. Calculate the stability index of the matching cluster: Stab i =f(Ni, σ i 2 , t inactive ); based on the stability index and matching distance, calculate the cluster confidence score: Conf i =g(Stab i , d min); analyze clustering features, extract key feature parameters of interference patterns: duration, intensity variation rate, direction stability, etc., generate a set of clustering feature parameters, which are used for subsequent interference type identification. The magnetic field interference pattern identification and clustering confidence score are passed to the subsequent processing module. Where f is a function of the stability index of the matching cluster, and g is a function of the cluster confidence score.

[0072] In another embodiment of the present application, in the core algorithm of incremental clustering, input the new sample x_t and the existing cluster set C = {C_1, C_2,..., C_k}, perform feature extraction, calculate the time window feature F_t, construct the enhanced data vector D_t = [x_t, F_t, ΔF_t], where ΔF_t is the change amount of the current time window feature. Calculate the adaptive distance d(D_t, C_i) for each cluster C_i, find the minimum distance d_min and the corresponding cluster C_min. If the minimum distance d_min < match threshold θ_match: assign the enhanced data vector D_t to the cluster C_min, incrementally update the core and statistics of the cluster C_min; otherwise, if the minimum distance d_min < satellite point threshold θ_satellite: the enhanced data vector D_t is taken as a satellite point of the cluster C_min, and the boundary information of the cluster C_min is updated; otherwise: if the available resources are sufficient and D_t is stable: create a new cluster C_new with the enhanced data vector D_t as the core; update the cluster set C' = C∪{C_new}; otherwise: temporarily store the enhanced data vector D_t as an unclassified sample. Perform cluster maintenance: merge similar clusters; remove aged clusters; periodically rebalance the cluster structure; output the updated cluster set C' and sample classification.

[0073] Perform cluster-compensation strategy mapping: for the identified cluster C_i: extract feature vectors: duration, intensity, directionality, rate of change...; calculate the similarity with the predefined patterns; select the most matching pattern type M_j; apply the corresponding compensation strategy S_j: if the pattern type M_j is "persistent environmental magnetic field offset", recalibrate the magnetic field reference model. Adjust the azimuth angle estimation bias; if the pattern type M_j is "sudden short-term interference", temporarily increase the magnetometer noise covariance. Increase the gyroscope integration weight; if the pattern type M_j is "periodic oscillation interference", apply a narrowband notch filter to suppress the corresponding frequency component in the frequency domain; if the pattern type M_j is "random noise enhancement", increase the smoothing filter strength and reduce the overall weight of the magnetometer.

[0074] According to an aspect of the present application, the step of performing attitude estimation to generate terminal attitude estimation value and uncertainty includes:

[0075] S31, multi-sensor data fusion preprocessing. Receive denoised sensor data, magnetic field interference mode identification and magnetic field reliability index; dynamically calculate the sensor fusion weight according to the magnetic field reliability index, reduce the magnetometer weight when the magnetic field is unreliable; adjust the filter parameters by applying interference compensation suggestions, generate adjusted filter configuration.

[0076] S32, implement adaptive extended Kalman filter. Perform state prediction step based on current attitude state estimation (quaternion representation) and 100Hz angular velocity data; calculate the prediction state uncertainty, update the system noise covariance matrix; adjust the magnetometer related components in the measurement noise covariance matrix according to the adjusted filter configuration; perform measurement update step using calibrated acceleration data and 100Hz magnetic field data; output updated attitude state estimation and estimation uncertainty.

[0077] S33, conduct attitude stability enhancement processing. Detect the mutation of attitude state estimation, trigger smooth transition processing if mutation is found; increase the gyroscope integration weight when the magnetic field reliability is low, reduce the dependence on magnetic field measurement; apply exponentially weighted moving average to smooth short-term fluctuations while retaining fast response capability; convert the quaternion representation to Euler angles to generate terminal attitude angles (pitch angle, roll angle and azimuth angle); combine the estimation uncertainty to generate attitude quality score and provide reliability indication.

[0078] According to one aspect of the present application, the steps of generating compensation parameter set by conducting filter-clustering dual-loop feedback control include:

[0079] S41, calculate the anomaly detection statistics based on filter residual, compare it with clustering confidence score, generate consistency score representing the coordination degree of filter and clustering results;

[0080] S42, determine the current state according to the consistency score, trigger deep adjustment when the consistency score is lower than the preset threshold, generate adjustment parameters including clustering threshold adjustment parameters and filter weight adjustment parameters;

[0081] S43, based on the magnetic field interference mode identification and adjustment parameters, select the corresponding compensation strategy, parameterize the selected compensation strategy parameters to generate compensation parameter set, including filter parameter adjustment set and clustering parameter adjustment set.

[0082] According to one aspect of the present application, the steps of generating compensation parameter set by selecting the corresponding compensation strategy include:

[0083] Receive magnetic field interference mode identification, clustering feature parameter set and adjustment parameters;

[0084] The current interference mode is matched with a predefined interference type library using a feature matching algorithm, a mapping relationship matrix including a correspondence relationship between the interference mode and a compensation strategy is established, and an optimal compensation strategy is selected by combining the interference mode features and a preset strategy effect score to generate a compensation parameter set.

[0085] Specifically, the attitude state estimation, estimation uncertainty and magnetic field interference mode identification are read, a filter residual-based anomaly detection statistic is calculated, and the consistency of the measurement and the prediction is evaluated; the anomaly detection statistic is compared with a clustering confidence score, and the consistency of the filtering and clustering results is evaluated; a consistency score is generated to quantify the coordination degree of the filtering and clustering results.

[0086] Based on the consistency score, it is judged whether the current state needs to be adjusted: if the consistency is high: keep the current parameter configuration; if the consistency is medium: slightly adjust the parameters; if the consistency is low: trigger a deep adjustment. When adjustment is needed, a clustering-oriented clustering threshold adjustment parameter and a filtering-oriented filtering weight adjustment parameter are generated. A smooth transition function is designed to ensure the gradualness of parameter adjustment and avoid introducing new instability.

[0087] The magnetic field interference mode identification and clustering feature parameters are received as inputs, a feature matching algorithm is used to map the current clustering mode to a predefined interference type library, a mapping relationship matrix is constructed, and a correspondence relationship between the interference mode and the best compensation strategy is established; based on historical compensation effect data, the strategy effect score is dynamically updated; according to the interference mode features and the strategy effect score, the optimal compensation strategy is selected, and a detailed compensation strategy parameter set is generated, including filter parameter adjustment, measurement weight correction and model update instruction.

[0088] According to the magnetic field interference mode identification, the corresponding compensation strategy is selected: for persistent environmental bias, a magnetic field reference model is applied for re-calibration; for sudden interference, a short-term magnetic field exclusion strategy is applied; for periodic interference, a frequency domain selective filter is applied. The selected compensation strategy is parameterized into a compensation parameter set, and the compensation parameter set is distributed to the attitude estimation module and the clustering module to form a closed-loop control.

[0089] The clustering analysis process is divided into interruptable micro-task units, the execution order and priority of the micro-tasks are dynamically adjusted based on the attitude quality score and system load; time slice allocation of the filtering main loop and the clustering auxiliary loop is realized to ensure real-time performance; the execution time of each module is monitored and recorded to optimize the task scheduling strategy.

[0090] The consistency score is calculated as follows: Consistency score Con = (1-|Anorm-Cfid|)·exp(-λc·σ 2Con is a consistency score of filtering and clustering results, value range [0, 1]; Anorm is a normalized anomaly detection statistic; Cfid is a confidence score of the current matching cluster; σ 2 est is a pose estimation uncertainty; λ c is an adjustment coefficient, controlling the influence degree of uncertainty on consistency.

[0091] The compensation parameter set generation mechanism is specifically: a filtering parameter adjustment amount ΔRm = Rm0·(1-Con)·f(Mpat); wherein ΔRm is a magnetometer measurement noise covariance adjustment amount; Rm0 is a reference noise covariance matrix; Con is a consistency score; f(Mpat) is an adjustment function based on the magnetic field interference mode Mpat; a clustering parameter adjustment amount ΔTc = Tc0·(1-Con)·g(Anorm); wherein ΔTc is a clustering matching threshold adjustment amount; Tc0 is a reference matching threshold; g(Anorm) is an adjustment function based on the anomaly detection statistic Anorm.

[0092] The embodiment breaks the one-way dependent relationship of the traditional filtering and clustering algorithm, and establishes a bidirectional feedback mechanism; through the consistency evaluation mechanism, mutual verification and compensation of the filtering result and the clustering result are realized; the smooth transition function design ensures the continuity of the pose estimation when the interference mode is switched, and avoids the pose jump in the traditional method.

[0093] According to one aspect of the present application, the step of calculating an anomaly detection statistic based on a filter residual and generating a consistency score includes:

[0094] According to the terminal pose estimation value and the historical state estimation value, a prediction residual is calculated;

[0095] A standardized statistic of the prediction residual is calculated using the Mahalanobis distance formula to generate an anomaly detection statistic;

[0096] The anomaly detection statistic and the clustering confidence score are weighted and fused to evaluate the consistency of the filtering and clustering results; a consistency score quantifying the consistency degree is generated.

[0097] According to one aspect of the present application, judging the current state according to the consistency score includes the following steps:

[0098] The consistency score is compared with a preset threshold to determine the adjustment level currently required to be executed;

[0099] According to the adjustment level, a clustering threshold adjustment parameter for the clustering algorithm is generated, including adjusting the matching threshold and the satellite point threshold;

[0100] According to the adjustment level, a filtering weight adjustment parameter for the filtering algorithm is generated, including adjusting the magnetometer-related components in the measurement noise covariance matrix.

[0101] According to an aspect of the present application, the steps of performing satellite direction positioning and locking, outputting satellite pointing angle and pointing control instruction include:

[0102] S51, calculate the satellite pointing angle. Receive the terminal attitude angle and satellite position information (azimuth and elevation), calculate the theoretical target pointing angle based on the current geographic position and satellite position; combine the attitude quality score to calculate the pointing angle uncertainty; optimize the pointing strategy to generate antenna pointing control parameters,

[0103] S52, directional mechanism control and feedback. Convert the antenna pointing control parameters into rudder control signals; monitor the rudder position feedback, calculate the pointing actual error; realize closed-loop position control, fine-tune the antenna pointing; when the pointing is stable, lock the mechanical structure to reduce energy consumption.

[0104] S53, signal quality monitoring and optimization. Monitor the satellite signal strength and signal-to-noise ratio, calculate the communication quality index; perform fine-tuning scanning to find the signal optimum point, record the optimal pointing offset; feedback the optimal pointing offset to the attitude estimation module as the system calibration parameter; continuously evaluate the communication stability, if necessary, trigger the reorientation process.

[0105] The following will be further illustrated by specific cases:

[0106] Case 1, in rugged mountainous environment, the rescue team needs to quickly deploy satellite communication terminal to establish communication connection with the command center. There are complex magnetic field environment such as geomagnetic anomaly and metal equipment interference in the field, and the traditional satellite orientation system relying on electronic compass cannot accurately position the satellite direction. In view of the communication demand of the first batch of rescue team in mountainous earthquake disaster area, a satellite communication terminal rapid deployment and orientation locking method for natural disaster rescue is proposed, which is as follows:

[0107] Step 1, system composition and data acquisition.

[0108] The system of the embodiment includes: three-axis gyroscope (model MPU-9250, sampling rate 500Hz); three-axis accelerometer (model MPU-9250, sampling rate 100Hz); three-axis magnetometer (model HMC5883L, sampling rate 20Hz); single board computer (with ARM Cortex-A72 processor); two-axis electrically adjustable antenna support (horizontal direction 360° rotation, vertical direction 0°-90° adjustment).

[0109] After the system starts, first execute the sensor initialization program to read the preset calibration parameters. Then execute the following data acquisition and preprocessing steps:

[0110] Multi-sensor parallel data acquisition: Gyroscope acquires three-axis angular velocity ω = [ωx, ωy, ωz] with 500 Hz sampling rate, unit rad / s; Accelerometer acquires three-axis acceleration a = [ax, ay, az] with 100 Hz sampling rate, unit m / s2; Magnetometer acquires three-axis magnetic field intensity m = [mx, my, mz] with 20 Hz sampling rate, unit μΤ; Each set of data is attached with a time stamp t with millisecond precision. T T 2 T T is transpose; ωx, ωy, ωz are angular velocities around X, Y, Z axes respectively; ax, ay, az are accelerations along X, Y, Z axes respectively; mx, my, mz are magnetic field intensities along X, Y, Z axes respectively.

[0111] Sensor data calibration and noise removal: Gyroscope data calibration: ω_cal = S_g·(ω_raw - b_g); where S_g is gyroscope scale factor matrix, b_g is bias vector. Gyroscope temperature compensation: ω_comp = ω_cal - k_g·(T - T_ref); where T is current temperature, T_ref is reference temperature, k_g is temperature coefficient. Accelerometer data calibration: a_cal = S_a·(a_raw - b_a); where S_a is accelerometer scale factor matrix, b_a is bias vector. Magnetometer hard iron calibration: m_hard = m_raw - b_m, where b_m is hard iron bias vector, m_raw is raw data measured by magnetometer. Magnetometer soft iron calibration: m_cal = S_m·m_hard, where S_m is soft iron correction matrix.

[0112] Multi-sampling rate data synchronization and alignment: Establish a unified time reference with 100 Hz as the reference sampling rate, gyroscope data down-sampling: take average every 5 samples; magnetometer data linear interpolation: interpolate between adjacent two sampling points to form aligned data packet: Data_packet(t) = {ω_comp(t), a_cal(t), m_cal(t), t}.

[0113] Step two, magnetic field interference pattern recognition and classification.

[0114] 2.1, Time series feature enhanced magnetic field data representation.

[0115] Extract magnetic field data features using sliding window method: Construct a sliding window with window length of 50 samples (500 ms). Extract time domain features of magnetic field data in the window: Magnetic field intensity mean: M_mean = ||m|| avg ​​​​= (Σ||m_i||) / 50; magnetic field direction unit vector: m_dir = m_mean / ||m_mean||; magnetic field intensity standard deviation: M_std = sqrt[(Σ(||m_i|| - M_mean) 2 ) / 50]; magnetic field direction fluctuation: D_var = (Σ(1-cos(θ_i))) / 50, θ_i is the angle between m_i and m_dir; short-time change rate: ΔM_max = max(||m_i+1 - m_i||) / Δt. Construct the enhanced magnetic field feature vector: M_feat = [m_dir, M_mean, M_std, D_var, ΔM_max]. Where m_i is the three-dimensional vector of magnetic field data; ||m|| avg is the average modulus of magnetic field data, that is, the average intensity; m_mean is the average vector of magnetic field in the sliding window; Δt is the time interval.

[0116] 2.2, Progressive incremental clustering algorithm of core-satellite structure.

[0117] The embodiment constructs a special incremental clustering algorithm with a "core-satellite" structure, which is suitable for dynamic changes of the magnetic field environment. The algorithm flow is as follows: three basic clustering templates (stable environment, periodic disturbance and strong magnetic field interference) are preset during system initialization; for each new enhanced magnetic field feature vector M_feat, the adaptive distance from the existing cluster is calculated: adaptive distance D(M_feat, ci) = 0.6·(1-cos(θi)) + 0.3·|log(M_mean / M_i)| + 0.1·(1-exp(-0.05·ΔT_i)); wherein θi is the included angle between the directions of M_feat and the core of the cluster ci; M_mean is the current average magnetic field strength, M_i is the core strength of the cluster ci; ΔT_i is the difference (seconds) between the current time and the last update time of the cluster ci. Find the minimum distance D_min and the corresponding cluster c_match; perform core-satellite structure update according to the minimum distance D_min: if D_min < Th_core (core threshold, initial value 0.15): belong to core update, calculate adaptive learning rate: α = 0.2·exp(-0.001·N_match)·(1-S_match), wherein N_match is the number of cluster matching, and S_match is the cluster stability index; update the cluster core parameters: c_match_new = (1-α)·c_match + α·M_feat; update the cluster statistics and time stamp. If Th_core ≤ D_min < Th_sat (satellite threshold, initial value 0.3): belong to satellite point, record the enhanced magnetic field feature vector M_feat as the satellite point of the cluster c_match; update the cluster boundary information: radius_max = max(radius_max, D_min); record the satellite point features for possible core splitting in the future. If D_min ≥ Th_sat: belong to potential new mode; add the enhanced magnetic field feature vector M_feat to the temporary buffer; if the number of similar samples in the buffer is ≥5, create a new cluster core.

[0118] Perform clustering maintenance and optimization. Cluster aging detection: calculate the cluster inactivity time, and mark it as dormant if it exceeds 30 seconds; cluster merging detection: if the distance D(ci, cj) between two cluster cores is <0.1, merge them into one; calculate the cluster confidence score: Cfid(c_match) = 0.4·min(1, N_match / 50) + 0.4·(1-min(1, V_match / 0.5)) +0.2·exp(-0.1·ΔT_act); wherein N_match is the cumulative number of samples in the cluster; V_match is the internal variance of the cluster; ΔT_act is the recent active interval time (seconds) of the cluster.

[0119] Output recognition result: magnetic field interference mode identification: Mode_ID = ID(c_match); clustering confidence score: Cfid(c_match); extract mode feature parameters: duration, intensity variation rate, direction stability.

[0120] Step three, adaptive parameter adjustment attitude estimation.

[0121] This embodiment uses an extended Kalman filter (EKF) for attitude estimation, dynamically adjusts the filter parameters according to the magnetic field interference mode.

[0122] 3.1, construct filter state and observation model.

[0123] State vector: x = [q, b_ω] T ; q = [q0, q1, q2, q3] T is the attitude quaternion; b_ω = [bx, by, bz] T is the gyroscope bias estimate. State transition equation: q_k+1 = q_kΘq(ω_k·dt); b_ω_k+1 = b_ω_k; where Θ represents quaternion multiplication, and q(ω_k·dt) represents the quaternion increment obtained by integrating the angular velocity. The measurement update uses acceleration and magnetic field data: gravity direction observation: g_meas = a_cal / ||a_cal||; magnetic field direction observation: m_meas = m_cal / ||m_cal||. Where q_k is the attitude quaternion at the current time, q_k+1 is the attitude quaternion at the next time, b_ω_k is the gyroscope bias estimate vector at the current time, and b_ω_k+1 is the gyroscope bias estimate vector at the next time.

[0124] 3.2, adaptive parameter adjustment mechanism.

[0125] The system dynamically adjusts the EKF parameters according to the magnetic field interference mode identification Mode_ID and the clustering confidence score Cfid: Measurement noise covariance matrix adjustment: R_mag = R_mag0·(2 - Cfid)·f(Mode_ID); where R_mag0 is the magnetometer baseline noise covariance; f(Mode_ID) is the adjustment function based on the interference mode: f = 1.0 for "stable environment" mode; f = 1.5 for "periodic disturbance" mode; f = 3.0 for "strong magnetic field interference" mode. Process noise covariance matrix adjustment: when Cfid < 0.3, increase the gyroscope noise covariance and reduce the dependence on magnetic field measurements; when Mode_ID is "strong magnetic field interference" and the duration > 10 seconds, switch to "gyroscope + gravity only" mode. Pose stability enhancement: detect pose mutations: if the estimated angle difference for two consecutive times > 20°, trigger smooth transition; apply adaptive smoothing: output pose q_out = SLERP(q_prev, q_curr, β), where SLERP is the spherical linear interpolation function, β is the smoothing coefficient, which is dynamically adjusted according to Mode_ID and Cfid. Output terminal pose estimation value: Euler angle representation: [roll, pitch, yaw], where roll is roll, pitch is pitch, and yaw is yaw; estimate uncertainty: based on the diagonal elements of the state covariance matrix P.

[0126] Step four, filter-clustering double-loop feedback control.

[0127] This embodiment realizes closed-loop control through the double-loop feedback mechanism between filtering and clustering algorithms.

[0128] 4.1, filter-clustering result consistency evaluation.

[0129] Calculate filter residual statistics: S_res = (z - h(x)) T ·R -1 ·(z - h(x)); where z is the actual measurement value; h(x) is the state-based predicted measurement value; R is the measurement noise covariance matrix. Normalize residual statistics: Anorm = min(1, S_res / χ 2 α), where χ 2 α is the 95% critical value of the chi-square distribution with 5 degrees of freedom. Calculate consistency score: Con = (1 - |Anorm - Cfid|)·exp(-0.5·tr(P) / 3), where Cfid is the confidence score of the current matching cluster, tr(P) is the trace of the state covariance matrix P, representing the estimation uncertainty.

[0130] 4.2, generate double-loop feedback parameters.

[0131] Based on the consistency score Con, the system generates two sets of feedback parameters:

[0132] Filter parameter adjustment set: AR_mag = R_mag0 · (1 - Con) · f(Mode_ID); AQ_gyro = Q_gyro0 · (1 - Con) · (2 - Cfid); where R_mag0 is the magnetometer reference noise covariance; Q_gyro0 is the gyroscope reference process noise covariance; f(Mode_ID) is the adjustment function based on interference mode. Cluster parameter adjustment set: ATh_core = Th_core0 · (1 - Con) · (0.5 + 0.5 · Anorm); ATh_sat = Th_sat0 · (1 - Con) · (0.5 + 0.5 · Anorm); where Th_core0 is the core threshold reference value 0.15; Th_sat0 is the satellite threshold reference value 0.3. Parameter smoothing transition: apply first-order low-pass filter to smooth parameter changes; parameter change rate limiter to prevent abrupt changes.

[0133] 4.3, Dynamic compensation strategy implementation.

[0134] For different magnetic field interference modes, the system implements specific compensation strategies: "Stable environment" mode (Mode_ID = 1): normal weight configuration, trust magnetic field measurement; R_mag = R_mag0, standard measurement update process. "Periodic disturbance" mode (Mode_ID = 2): apply frequency domain selective filtering, filter periodic interference; adjust the magnetic field measurement update weight: R_mag = 1.5 · R_mag0; shorten the sliding window length, improve short-term response ability. "Strong magnetic field interference" mode (Mode_ID = 3): execute magnetic field exclusion strategy, mainly rely on gyroscope and gravity; significantly increase the magnetic field measurement noise: R_mag = 3.0 · R_mag0; use historical magnetic field reference model to assist orientation. "Unknown interference" mode (new cluster): conservative strategy, increase system uncertainty; reduce filter convergence speed, maintain stability.

[0135] Step five, satellite direction positioning and locking.

[0136] 5.1, Calculate satellite pointing angle.

[0137] Based on the GPS position and target satellite parameters, calculate the theoretical satellite azimuth Az_sat and elevation El_sat. Combine the current terminal attitude estimation [roll, pitch, yaw] to calculate the azimuth and elevation that the antenna needs to adjust: Az_adjust = Az_sat - yaw + Az_offset; El_adjust = El_sat cos(roll) - pitch sin(Az_adjust) + El_offset; where Az_offset and El_offset are calibration offsets. Calculate the pointing angle uncertainty according to the attitude estimation uncertainty: σ_Az = sqrt(σ_yaw 2 + (σ_roll·sin(El_sat)) 2 ); σ_El = sqrt(σ_pitch 2 +(σ_roll·cos(Az_adjust)) 2 ), where σ_Az is the azimuth uncertainty, σ_El is the elevation uncertainty, σ_yaw is the uncertainty of the yaw estimation, σ_roll is the uncertainty of the roll estimation, and σ_pitch is the uncertainty of the pitch estimation.

[0138] 5.2, Perform directional mechanism control and locking.

[0139] Convert the pointing angle to servo control signals: horizontal servo angle = map(Az_adjust, -180°, 180°, 0°, 360°); vertical servo angle = map(El_adjust, 0°, 90°, 0°, 90°). Implement a two-stage locking strategy: coarse adjustment: approach the target direction at a higher speed; fine adjustment: fine-tune to the optimal receiving position at low speed. Adaptive search algorithm: when the pointing uncertainty σ_Az > 10°, perform a scanning search; the search range is an elliptical region with ±2σ; the step size is proportional to the uncertainty, gradually reducing. Signal peak locking: monitor the received signal strength RSSI and record the peak position; when RSSI > threshold and relatively stable, lock the current position; start the mechanical lock to fix the antenna pointing.

[0140] Compared with the traditional method, the average error of satellite pointing accuracy of the traditional method is 8.5°, and the maximum error is 21.3°; the average error of satellite pointing accuracy of the embodiment is 3.2°, and the maximum error is 7.8°; the average error is reduced by 62.4%. The positioning time of the traditional method is 73 seconds on average, and the slowest is 126 seconds; the positioning time of the embodiment is 31 seconds on average, and the slowest is 54 seconds; the improvement rate is 57.5% on average. In the strong magnetic field interference area (magnetic field strength change > 50 μT): the magnetic field interference adaptability of the traditional method cannot be reliably oriented (success rate < 30%), and the magnetic field interference adaptability of the embodiment is stable (success rate > 85%). The CPU occupancy of the embodiment is < 40%; the memory occupancy is < 120 MB; the battery endurance time is > 6 hours. The embodiment improves the directional accuracy and deployment speed of the satellite terminal in harsh environments through the incremental clustering algorithm and the filter-clustering double-loop feedback control mechanism, and provides reliable communication support for disaster relief.

[0141] Case two, the embodiment provides a satellite communication terminal rapid deployment and directional locking method for natural disaster rescue, which can realize accurate attitude estimation and satellite directional locking of the satellite communication terminal in a complex magnetic field interference environment. The system adopts a nine-axis sensor combination (three-axis accelerometer, three-axis gyroscope, and three-axis electronic compass) to collect data, solves the attitude estimation error problem caused by magnetic field interference in the disaster area environment through a filter-clustering double-loop feedback control structure and an incremental clustering algorithm of the progressive magnetic field interference mode. Specifically,

[0142] Step one, sensor data collection and preprocessing.

[0143] The terminal device adopts a nine-axis sensor with model MPU9250, and collects data according to the following sampling rate and accuracy requirements: gyroscope: sampling rate 500 Hz, range ± 250° / s, resolution 0.01° / s; accelerometer: sampling rate 100 Hz, range ± 2g, resolution 0.001g; electronic compass: sampling rate 20 Hz, range ± 4800 μT, resolution 0.6 μT.

[0144] The original sensor data is calibrated, and the calibration equation is as follows: gyroscope calibration: ω_calibrated = (ω_raw - b_g)·S_g·T_comp; wherein ω_calibrated is the calibrated angular velocity data, ω_raw is the original angular velocity data, b_g is the gyroscope bias vector, S_g is the proportional coefficient matrix, and T_comp is the temperature compensation coefficient.

[0145] Accelerometer calibration: a_calibrated = (a_raw - b_a) · S_a; where a_calibrated is calibrated acceleration data, a_raw is raw acceleration data, b_a is accelerometer bias vector, S_a is scale factor matrix.

[0146] Electronic compass calibration: m_calibrated = (m_raw - b_m) · S_m · A; where m_calibrated is calibrated magnetic field data, m_raw is raw magnetic field data, b_m is hard iron calibration bias vector, S_m is soft iron calibration matrix, A is azimuth calibration matrix.

[0147] Align sensor data with different sampling rates to 100Hz, design time window alignment algorithm. For gyroscope data, use fourth-order Butterworth low-pass filter (cutoff frequency 45Hz) to reduce noise influence, for electronic compass data, use cubic spline interpolation method to improve sampling rate. Align three kinds of sensor data on the basis of timestamp, generate time-aligned multi-sensor data packet D_aligned = {t, ω, a, m}, where t is timestamp, ω is angular velocity vector, a is acceleration vector, m is magnetic field vector.

[0148] Step two, magnetic field interference pattern recognition and classification.

[0149] 2.1, time series feature enhanced magnetic field data representation.

[0150] Receive time-aligned multi-sensor data packet, construct 500ms sliding window, extract time-domain features of magnetic field data. Feature calculation as follows: magnetic field intensity feature: M_mag = ||m|| = sqrt(m_x 2 + m_y 2 + m_z 2 ); direction feature: M_dir = [arctan(m_y / m_x), arctan(m_z / sqrt(m_x 2 + m_y 2 ))] Time-domain statistical features: mean: μ_m = (1 / N)·∑m_i; variance: σ 2 _m = (1 / N)·∑(m_i - μ_m) 2 ; maximum change rate: Δm_max = max(||m_i - m_{i-1}||) / Δt; frequency energy: E_m = ∑|FFT(m)| 2. Relative deviation feature: Δ_ref = ||m- m_ref|| / ||m_ref||; where m_ref is the reference magnetic field strength determined according to the geographical location. The above features are combined to construct the enhanced magnetic field feature vector F = [M_mag, M_dir, μ_m, σ 2 _m, Δm_max, E_m, Δ_ref]. Where m is the magnetic field vector; m_x, m_y, m_z represent the components of the magnetic field vector on the x-axis, y-axis and z-axis, respectively; N is the number of data points in the sliding window; m_i is the magnetic field vector data at the i-th time in the current window; Δt is the time difference; FFT represents the fast Fourier transform.

[0151] 2.2, Incremental clustering of progressive magnetic field interference patterns.

[0152] The following detailed description of the key calculation process of the core-satellite structure progressive incremental clustering algorithm:

[0153] Adaptive distance metric calculation: a) Feature normalization processing, generating a normalized feature vector: F_norm = (F - F_min) / (F_max - F_min). b) Time decay weight calculation: w_t = exp(-λ·(t_current - t_last)); where λ is the time decay coefficient (value taken in this embodiment 0.05), t_current is the current timestamp, t_last is the last update timestamp of clustering. c) Direction-sensitive distance calculation: d_dir(v1, v2) = 1 - |cos(v1, v2)| = 1 - |v1·v2| / (||v1||·||v2||); This distance measures the similarity of two magnetic field direction vectors, with a value range of [0, 1]. d) Intensity-sensitive distance calculation: d_mag(m1, m2) = |log(||m1|| / ||m2||)|; This distance captures the relative change proportion of the magnetic field intensity, and the logarithmic function makes the intensity changes of different sizes comparable. e) Temporal feature distance calculation: d_temp(F1_temp, F2_temp) = sqrt(∑w_i)·(F1_temp_i - F2_temp_i) 2where F_temp represents the time-domain statistical feature part, and w_i is the weight of each feature. f) Comprehensive calculation of the adaptive distance metric function: d(F, C_i) = a · d_dir(F_dir, C_i_dir) + β · d_mag(F_mag, C_i_mag) + γ · d_temp(F_temp, C_i_temp); where a, β, and γ are weight coefficients, and in this embodiment, the initial values are set to 0.4, 0.3, and 0.3, respectively, and are dynamically adjusted according to the degree of magnetic field interference. g) Calculate the distance between the current sample and each existing cluster, find the minimum distance d_min and the corresponding cluster C_min. Where F is the feature vector; F_min is the minimum value of each feature in the feature vector; F_max is the maximum value of each feature in the feature vector; v1 and v2 are magnetic field direction vectors; m1 and m2 are magnetic field strength vectors; F_dir is the direction feature of the sample; C_i_dir is the direction feature of the cluster center; F_mag is the magnetic field strength feature of the sample; C_i_mag is the magnetic field strength feature of the cluster center; F_temp is the time feature vector of the sample; C_i_temp is the time feature vector of the cluster center.

[0154] Core-satellite structure update: a) Set two key thresholds: matching threshold θ_match (initial value 0.15 in this embodiment) and satellite point threshold θ_satellite (initial value 0.3 in this embodiment). b) According to the relationship between the minimum distance d_min and the threshold, one of the following three operations is performed: if d_min < θ_match: perform core update operation; if θ_match ≤ d_min < θ_satellite: perform satellite point update operation; if d_min ≥ θ_satellite: perform potential new mode processing. c) Detailed calculation process of core update operation: learning rate calculation: η = η_0 · (1 - exp(-N_i / τ)); where η_0 is the basic learning rate (value 0.1), N_i is the number of samples of cluster C_min, and τ is the scaling parameter (value 50). Core parameter update: C_min_core = (1-η)·C_min_core + η·F_norm; Update cluster statistics: increase sample count: N_i = N_i +1; Update variance: σ_i 2 = (1-η)·σ_i 2 + η·||F_norm - C_min_core|| 2; update the last active timestamp: t last = t current.d) Satellite point update operation detailed calculation process: add the normalized feature vector F norm as a satellite point of the cluster C min: S min = S min U {F norm}; update the cluster boundary information: B min = max(B min, d min); keep the satellite point set size limited: if |S min| > K max, remove the oldest satellite point, the maximum capacity K max of the satellite point set in this embodiment is set to 20.e) Potential new mode processing detailed calculation process: add the normalized feature vector F norm to the temporary buffer: T buffer = T buffer U {F norm}; calculate the average mutual distance of the points in the buffer: d avg = (2 / n(n-1))·∑∑d(F i, F j), where i < j; n is the number of points in the buffer; if the new cluster creation condition is met: d avg < θ new and |T buffer| > n min, create a new cluster C new, the initial core is the average value of the points in the temporary buffer T buffer: C new_core = (1 / |T buffer|)·∑F i; set the initial sample count N new = |T buffer|, the initial variance σ new 2 = d avg, the initial timestamp t new_last= t current; add the new cluster to the cluster set: C = C U {C new}; empty the temporary buffer: T buffer =∅, the distance threshold θ new of the new cluster in this embodiment is 0.2, and the minimum point quantity threshold n min of the new cluster is 5.

[0155] Cluster maintenance and optimization: the system performs the following operations every certain time (this embodiment is set to 30 seconds) or when a specific event is triggered: a) Cluster aging detection: calculate the inactive time of each cluster: t inactive = t current- t last; if t inactive > T max (T max in this embodiment is set to 300 seconds), mark the cluster as dormant.b) Cluster merging detection: calculate the distance between the cores of any two active clusters: d(C i_core, C j_core); if d(C i_core, C j_core) < θ merge (the value in this embodiment is 0.1): calculate the weighted average to get the new core: C new_core = (N i·C i_core + N j·C j_core) / (N i + N j); merge the statistics: N new =N i + N j σ new 2= (N_i·σ_i 2 + N_j·σ_j 2 + N_i·N_j·||C_i_core - C_j_core|| 2 / (N_i + N_j)) / (N_i + N_j); Merge satellite point sets: S_new = S_i ∪ S_j; If |S_new| > K_max, retain the nearest K_max points; Delete the original two clusters and add the merged cluster. Where C_i_core is the core parameter of cluster C_i; θ_merge is the distance threshold for cluster merging.

[0156] Incremental clustering result evaluation: a) Calculation of clustering stability index: Stab_i = f(N_i, σ_i) 2 , t_inactive) = N_i·exp(-k1·σ_i 2 a) Calculate the clustering confidence score: Conf_i = g(Stab_i, d_min) = Stab_i·exp(-d_min / d_0); where k1 and k2 are weight parameters (in this embodiment, they are 2.0 and 0.5 respectively). b) Calculate the clustering confidence score: Conf_i = g(Stab_i, d_min) = Stab_i·exp(-d_min / d_0); where d_0 is the scaling parameter (in this embodiment, it is 0.2). c) Extract the interference mode feature parameters: duration: T_duration = t_current - t_first_appear; intensity change rate: R_intensity = (||m_current|| - ||m_first||) / T_duration; directional stability: S_direction = 1 - σ(cos(m_i, μ_m)). d) Generate the magnetic field interference mode identifier M_id and the clustering confidence score Conf_i. Where t_first_appear is the timestamp of the first appearance of the magnetic field interference pattern; m_current is the magnetic field strength vector at the current moment; m_first is the magnetic field strength vector when the magnetic field interference pattern first appears; and σ is a statistic in directional stability.

[0157] Step 3: Attitude Estimation with Adaptive Parameter Adjustment

[0158] 3.1 Multi-sensor data fusion preprocessing.

[0159] Based on the magnetic field interference mode identification M_id and the magnetic field reliability index Conf_i, the sensor fusion weight is dynamically adjusted: sensor weight calculation: w_m = w_m0·exp(-k·(1-Conf_i)); wherein w_m0 is the magnetometer basic weight (0.3 in this embodiment), and k is an adjustment parameter (2.0). Measurement noise covariance matrix adjustment: R_m = R_m0 / w_m wherein R_m0 is the magnetometer basic noise covariance matrix, and the covariance is increased at low reliability.

[0160] 3.2, Adaptive extended Kalman filter implementation.

[0161] State vector definition: x = [q, b_g, b_a, b_m] T ; wherein q is the attitude quaternion, b_g is the gyroscope bias, b_a is the accelerometer bias, and b_m is the magnetometer bias. State prediction step: a) quaternion integration: q*_k|k-1 = q_k-1|k-1 Θ q(Δt·(ω_k - b_g_k-1|k-1)); wherein Θ represents quaternion multiplication, q(·) represents the conversion from angular velocity to quaternion increment; q_k-1|k-1 is the attitude quaternion at the last time; ω_k is the gyroscope measurement value at the current time; b_g_k-1|k-1 is the gyroscope bias at the last time. b) bias prediction: b*_g_k|k-1 = b_g_k-1|k-1 b*_a_k|k-1 = b_a_k-1|k-1 b*_m_k|k-1 = b_m_k-1|k-1; wherein b*_a_k|k-1 is the predicted accelerometer bias at the current time; b_a_k-1|k-1 is the accelerometer bias at the last time; b*_m_k|k-1 is the predicted magnetometer bias at the current time; b_m_k-1|k-1 is the magnetometer bias at the last time. c) predicted state uncertainty update: P_k|k-1 = F_k·P_k-1|k-1·F_k T + Q_k; wherein P_k|k-1 is the uncertainty covariance matrix of the predicted state; F_k is the state transition matrix, and Q_k is the process noise covariance matrix. Measurement update step: a) According to the adjusted filter configuration, set the measurement noise covariance matrix R_k. b) Calculate the Kalman gain: K_k = P_k|k-1·H_k T ·(H_k·P_k|k-1·H_k T + R_k) -1; where H_k is the measurement matrix. c) State update: x_k|k = x_k|k-1 + K_k·(z_k - h(x_k|k-1)); where x_k|k-1 is the predicted state vector at current time, z_k is the actual measurement, and h(·) is the measurement function. d) Covariance update: P_k|k = (I - K_k·H_k)·P_k|k-1; where I is the identity matrix. Output the updated attitude state estimation (quaternion representation) and estimation uncertainty P_k|k.

[0162] Attitude stability enhancement processing: Attitude jump detection: if ||q_k|k Θ q_k-1|k-1-1|| > θ_jump (value 0.05 in this embodiment): trigger smooth transition processing. Smooth transition processing: q_smooth = slerp(q_k-1|k-1, q_k|k, s); where slerp is the spherical linear interpolation function, s is the smoothing coefficient, and the calculation formula is: s = 0.5 - 0.5·cos(π·min(t / T_trans, 1)); T_trans is the transition time constant (value 1 second in this embodiment). Attitude angle calculation: convert the quaternion q into the terminal attitude angle [φ, θ, ψ] represented by Euler angles (representing roll angle, pitch angle and azimuth angle respectively). Attitude quality score calculation: Q_attitude = exp(-tr(P_q) / σ_0 2 ); where P_q is the covariance matrix of the quaternion part, tr(·) represents the trace of the matrix, and σ_0 2 is the normalization parameter (value 0.01 in this embodiment).

[0163] Step four, filter-clustering double-loop feedback control.

[0164] 4.1, filter-clustering result consistency evaluation.

[0165] Abnormality detection statistics calculation: a) Prediction residual calculation: r_k = z_k - h(x_k|k-1); where h () is the measurement function. b) Mahalanobis distance calculation: D_M 2 = r_k T ·S_k -1 ·r_k; where S_k = H_k·P_k|k-1·H_k T + R_k is the residual covariance matrix. c) Abnormality detection threshold calculation: if D_M 2 > χ 2 n, α (χ 2 is the threshold of chi-square distribution, n is the measurement dimension, and α is the significance level, which is 0.05 in this embodiment), then an abnormality is detected.

[0166] Consistency score calculation: Consistency = w_filter·exp(-D_M 2 / s_1) + w_cluster·Conf_i; where w_filter and w_cluster are weight coefficients (both take 0.5 in this embodiment), and s_1 is a scaling parameter (takes the value of the residual degree of freedom n).

[0167] 4.2, Generating bidirectional feedback parameters.

[0168] State adjustment judgment: if Consistency > θ_high (0.8 in this embodiment): maintain the current parameter configuration; if θ_low < Consistency ≤ θ_high (θ_low takes 0.5): slightly adjust the parameters; if Consistency ≤ θ_low: trigger deep adjustment.

[0169] Cluster threshold adjustment parameter calculation: a) Matching threshold adjustment: θ_match_new = θ_match·(1 + k_adjust·(θ_mid - Consistency)); where k_adjust is an adjustment intensity parameter (0.5 in value), and θ_mid is an intermediate threshold value (0.65 in value). b) Satellite point threshold adjustment: θ_satellite_new = θ_satellite·(1 + k_adjust·(θ_mid - Consistency)). Filter weight adjustment parameter calculation: a) Measurement noise covariance scaling ratio: scale_R = 1 + k_R·max(0, θ_high - Consistency); where k_R is an amplification coefficient (4.0 in value). b) Update the magnetometer-related components in the R matrix: R_m_new = scale_R·R_m. Smooth transition function implementation: param_smooth = param_old + (param_new - param_old)·(0.5 - 0.5·cos(π·min(t / T_trans, 1))); where param_old is the parameter state before adjustment; param_new is the target new value of the parameter; T_trans is a transition time constant (2 seconds in this embodiment).

[0170] 4.3, Cluster-compensation strategy mapping mechanism.

[0171] Feature matching algorithm: compute the similarity between the current cluster pattern and the predefined interference type library pattern: similarity(M_current, M_ref) = exp(-d(F_current, F_ref) / d_scale); where d(·,·) is the aforementioned adaptive distance metric function, d_scale is the scaling parameter (value 0.3); M_current is the current identified interference pattern; M_ref is the reference pattern in the predefined interference type library; F_current is the feature vector of the current pattern; F_ref is the feature vector of the reference pattern.

[0172] Construct the mapping relationship matrix: according to the interference characteristics, construct the mapping matrix Map, each row represents a type of interference, each column represents a compensation strategy, and the element value represents the applicability. This embodiment supports the following interference types: persistent environmental magnetic field offset; sudden transient interference; periodic oscillation interference; random noise enhancement.

[0173] Strategy effect score update: Score_new(M_i, S_j) = λ·Score_old(M_i, S_j) + (1-λ)·ΔH; where Score_old is the historical effect score; λ is the historical weight coefficient (value 0.8), ΔH is the state uncertainty change before and after the strategy execution. Optimal compensation strategy selection: for the current identified interference pattern M_i, select the strategy with the highest effect score: S_opt = argmax_j(Map[M_i, S_j]·Score(M_i, S_j)). Compensation strategy parameterization: according to the selected compensation strategy, generate the corresponding compensation parameter set, including: filter parameter adjustment instructions; measurement weight correction value; model update instructions.

[0174] 4.4, Dynamic compensation strategy implementation.

[0175] Differentiated processing is implemented for different types of magnetic field interference: persistent environmental magnetic field offset compensation: a) recalibrate the magnetic field reference model: m_ref_new = (1-α_ref)·m_ref_old + α_ref·m_current; where m_ref_old is the old magnetic field reference model; α_ref is the adaptation coefficient (value 0.2). b) Adjust the azimuth angle estimation bias: ψ_bias = arctan2(m_ref_north×m_current, m_ref_north·m_current); where m_ref_north is the north component of the reference magnetic field; azimuth correction: ψ_corrected = ψ - ψ_bias.

[0176] Transient burst disturbance compensation: a) Temporarily increase the magnetometer noise covariance: R_m_temp = β_temp·R_m where β_temp is the temporary amplification factor (value 8.0). b) Increase the gyro integration weight: w_gyro_temp = min(1.0, w_gyro + Δw_gyro); where w_gyro is the integration weight of the gyro; Δw_gyro is the weight increment (value 0.3). c) Set the recovery time constant T_recover (value 5 seconds) and then gradually restore the normal parameters.

[0177] Periodic oscillation disturbance compensation: a) Spectrum analysis identifies the main disturbance frequency f_disturb. b) Design a narrow-band notch filter: H(z) = (1 - 2·cos(2πf_disturb / fs)·z -1 + z -2 ) / (1 - 2·r·cos(2πf_disturb / fs)·z -1 + r 2 ·z -2 ); where z is the variable in the filter, fs is the sampling frequency, and r is the attenuation coefficient (value 0.9). c) Apply the notch filter to the magnetic field data.

[0178] This embodiment constitutes the core implementation of the satellite communication terminal rapid deployment and directional locking method for natural disaster rescue. Through incremental clustering and filter-clustering double-loop feedback control of gradual magnetic field disturbance mode, the precision and stability of satellite communication terminal attitude estimation in the magnetic field disturbance environment can be effectively solved, and rapid deployment and reliable directional locking can be achieved.

[0179] The application designs a core-satellite structure clustering model, solves the problem that the traditional clustering method cannot adapt to the characteristics of time series data, adopts an incremental updating mechanism, so that the clustering model can continuously adapt to the changing environment without storing and processing all historical data, and an adaptive distance measurement function considers the time correlation and direction characteristics of the magnetic field data, thereby improving the clustering accuracy. The application breaks the one-way dependent relationship of the traditional filtering and clustering algorithm, establishes a two-way feedback mechanism, realizes mutual verification and compensation of the filtering result and the clustering result through a consistency evaluation mechanism, and ensures the continuity of the attitude estimation when the interference mode is switched through the design of a smooth transition function, thereby avoiding the attitude jump in the traditional method. The application designs a differentiated compensation strategy for different types of magnetic field interference, rather than the simple weight adjustment in the traditional method, adopts different processing methods for persistent environmental offset, sudden interference and periodic interference, thereby improving the system adaptability, and parameterizes the compensation strategy design to enable the system to flexibly cope with various complex interference environments. The application decomposes the computationally intensive clustering analysis into interruptable micro-tasks, solves the contradiction between real-time performance and complex computation, dynamically adjusts the processing priority according to the attitude quality, ensures the optimization of the system performance under the condition of limited resources, and ensures that the key processing is not delayed through the time slice allocation mechanism of the filtering main loop and the clustering auxiliary loop.

[0180] The above describes the preferred embodiments of the application, but the application is not limited to the specific details in the above embodiments, and various equivalent transformations can be made to the technical solutions of the application within the technical concept range of the application, and these equivalent transformations all belong to the protection range of the application.

Claims

1. A method for fast deployment and directional locking of satellite communication terminals for natural disaster relief, characterized in that, The method comprises the following steps: Obtaining sensor data, preprocessing, and forming a data packet; Receiving the data packet, classifying the magnetic field interference mode, and obtaining the magnetic field interference mode identifier; Performing attitude estimation based on the magnetic field interference mode identifier, and generating a terminal attitude estimation value and an uncertainty; Performing filter-clustering double-loop feedback control based on the magnetic field interference mode identifier, the terminal attitude estimation value, and the uncertainty, and generating a compensation parameter set comprising a filter parameter adjustment set and a clustering parameter adjustment set; feeding back the clustering parameter adjustment set to the magnetic field interference mode classification and feeding back the filter parameter adjustment set to the attitude estimation, and performing a loop until a precision threshold is met; Performing satellite direction positioning and locking based on the terminal attitude estimation value meeting the precision threshold and satellite position information, and outputting a satellite pointing angle and a pointing control instruction; The step of classifying the magnetic field interference mode to obtain the magnetic field interference mode identifier comprises: Receiving the data packet, the initial or feedback clustering parameter adjustment set, extracting the time domain features of the magnetic field data, and constructing an enhanced magnetic field feature vector; Processing the enhanced magnetic field feature vector using a progressive incremental clustering algorithm with a core-satellite structure, and dynamically adjusting the clustering threshold based on the clustering parameter adjustment set to obtain the magnetic field interference mode identifier and a clustering confidence score; The step of performing filter-clustering double-loop feedback control to generate a compensation parameter set comprises: Calculating an anomaly detection statistic based on the filter residual, comparing it with the clustering confidence score, and generating a consistency score representing the coordination degree of the filter and clustering results; Triggering a deep adjustment when the consistency score is lower than a preset threshold, and generating adjustment parameters, including a clustering threshold adjustment parameter and a filter weight adjustment parameter; Based on the magnetic field interference mode identifier and the adjustment parameters, selecting a corresponding compensation strategy, parameterizing the selected compensation strategy to generate a compensation parameter set, including a filter parameter adjustment set and a clustering parameter adjustment set.

2. The method of claim 1, wherein, The step of processing the enhanced magnetic field feature vector using a progressive incremental clustering algorithm with a core-satellite structure to obtain the magnetic field interference mode identifier and the clustering confidence score comprises: For the enhanced magnetic field feature vector, calculate the adaptive distance between the current feature vector and the existing clusters, determine the minimum distance and the corresponding cluster, and record it as the most matched cluster; According to the minimum distance and the matching threshold in the clustering parameter adjustment set, perform core-satellite structure update, assign the current sample as core update, satellite point or potential new mode, and update the clustering core parameter, clustering boundary information or potential new mode buffer accordingly; When the preset condition is met, perform clustering maintenance and optimization to generate an updated cluster set, and determine the magnetic field interference mode identifier of the current most matched cluster; Calculate the clustering confidence score based on the state of the updated cluster set.

3. The method of claim 2, wherein, The step of calculating the adaptive distance between the current feature vector and the existing clusters, and determining the minimum distance and the corresponding cluster comprises: Performing normalization processing on the enhanced magnetic field feature vector to generate a standardized feature vector comprising a direction component and an intensity component; Based on the standardized feature vector, calculate the difference between the current time and the last update time of each cluster, and apply an exponential decay function to generate a time decay weight coefficient; According to the angle cosine value between the direction component and the core direction of each cluster, calculate the direction sensitive distance; According to the logarithmic ratio of the intensity component and the intensity of each cluster core, the intensity-sensitive distance is calculated; An adaptive distance metric value is generated by combining the direction-sensitive distance, the intensity-sensitive distance, and the time decay weight coefficient using a preset weight coefficient set; The adaptive distance metric values of all clusters are compared to determine the minimum distance and the corresponding cluster.

4. The method of claim 2, wherein, The steps of performing core-satellite structure updating, and correspondingly updating the cluster core parameters, the cluster boundary information, or the potential new mode buffer, include: The minimum distance is compared with the matching threshold to determine whether to perform a core update operation, a satellite point update operation, or potential new mode processing; When the core update operation is performed, an adaptive learning rate is calculated and the cluster core parameters, statistics, and active time stamp are updated; When the satellite point update operation is performed, the current sample is recorded as a satellite point, and the cluster boundary information and the satellite point set are updated; When the potential new mode processing is performed, the current sample is added to the temporary buffer, and when the samples in the temporary buffer meet the preset new cluster creation condition, a new cluster core is generated, and the potential new mode buffer is updated.

5. The method of claim 4, wherein, The steps of performing cluster maintenance and optimization, and determining the magnetic field interference mode identifier of the current most matching cluster, include: Performing cluster aging detection, calculating the inactivity time of each cluster, and marking the cluster whose inactivity time exceeds the predetermined threshold as a dormant state; performing cluster merging detection, calculating the cluster core distance based on the updated cluster core parameters, and when the distance between two cluster cores is less than a preset merging threshold, a new core is generated by calculating the weighted average and merging the statistics and satellite point set; performing cluster structure rebalancing, periodically optimizing the cluster boundary and core distribution, reducing overlap and redundancy; and outputting the updated cluster set; Based on the updated cluster set, the overall stability index of the current cluster model is calculated, and the magnetic field interference mode identifier of the current most matching cluster is determined.

6. The method of claim 2, wherein, The steps of calculating the cluster confidence score based on the state of the updated cluster set include: Based on the updated cluster set, the sample quantity, internal variance, and inactivity time are extracted, and the stability index of the matching cluster is calculated; Based on the stability index and the minimum distance, a cluster confidence score quantifying the reliability of the current cluster identification result is generated, and the interference mode features of the matching cluster, including the duration, intensity change rate, and direction stability, are extracted, and a cluster feature parameter set is generated.

7. The method of claim 1, wherein, The steps of selecting a corresponding compensation strategy and generating a compensation parameter set include: Receiving the magnetic field interference mode identifier, the cluster feature parameter set, and the adjustment parameter; Using a feature matching algorithm to match the current interference mode with a predefined interference type library, establishing a mapping relationship matrix including the correspondence between the interference mode and the compensation strategy, and combining the interference mode features and the preset strategy effect score to select the optimal compensation strategy and generate a compensation parameter set.

8. The method of claim 1, wherein, The steps of calculating the anomaly detection statistics based on the filter residual and generating a consistency score include: According to the terminal attitude estimation value and the historical state estimation value, the prediction residual is calculated; The Mahalanobis distance formula is used to calculate the standardized statistics of the prediction residual to generate the anomaly detection statistics; The anomaly detection statistics and the cluster confidence score are weighted and fused to evaluate the consistency of the filtering and clustering results; a consistency score quantifying the consistency degree is generated.

Images