High-precision positioning method for combined navigation type underwater operation platform

Abstract

The application relates to the technical field of underwater navigation and positioning, and discloses a high-precision positioning method for a combined navigation type underwater operation platform. The method comprises the following steps: acquiring inertial navigation data, sonar positioning data and satellite positioning data to form a multi-dimensional navigation data source, performing data fusion processing to generate a navigation data warehouse, further extracting a navigation feature matrix, establishing a feature linkage graph to represent the correlation of the multi-source data, and screening a core positioning feature sequence to control platform positioning. The method realizes stable positioning in a complex underwater environment. The introduction of the feature linkage graph optimizes the weight distribution mechanism of the multi-source data, the screening process of the core feature sequence is adaptive to environmental changes, and the negative influence of redundant data on positioning accuracy is reduced. Compared with a traditional fixed-weight fusion method, the application has higher robustness in scenes such as flow disturbance and acoustic signal attenuation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underwater navigation and positioning, in particular to a high-precision positioning method for a combined navigation type underwater operation platform. BACKGROUND

[0002] The positioning technology of underwater operation platforms has always been a key and difficult point in the field of natural fluid engineering and resource exploration. Due to the complex and changeable underwater environment and the serious attenuation of electromagnetic wave signals, traditional single navigation methods often cannot meet the positioning requirements of high precision and high reliability. Although the inertial navigation system can provide continuous autonomous pose information, its error will accumulate over time, resulting in insufficient long-term accuracy. The sonar positioning technology can achieve relative positioning within a certain range, but it is easily affected by water conditions, multipath effects and environmental noise interference, and has a limited coverage range. Satellite positioning can achieve high accuracy on the water surface, but for underwater platforms, the signal cannot penetrate the water body and must rely on a buoy or a water surface unit for transfer, which limits the real-time performance and availability.

[0003] Traditional methods have weak fault tolerance for abnormal data and lack systematic screening and optimization mechanisms for navigation features, which limits the further improvement of the overall positioning performance. Therefore, it is urgent to develop a high-precision positioning method that can deeply integrate multi-source navigation information and has feature correlation analysis and optimization capabilities to meet the long-time, stable and accurate operation requirements of underwater operation platforms in complex underwater environments.

[0004] In the prior art, multi-sensor data fusion methods have been tried to improve positioning accuracy, but simple weighted fusion or Kalman filtering is usually used, without fully considering the feature correlation of different navigation data sources. For example, the dynamic characteristics of inertial navigation data and the spatial geometric constraints of sonar positioning lack deep coupling, resulting in insufficient positioning robustness in complex underwater environments. In addition, the data source screening mechanism of traditional methods is fixed and cannot adapt to dynamic changes in the underwater environment (such as water flow disturbance and terrain obstruction), which easily leads to positioning jumps or failures.

[0005] Some studies try to introduce machine learning to optimize the data fusion process, but the model training relies on a large amount of historical data, and the spatio-temporal alignment problem of multi-source data is not solved. The differences in sampling frequency of sonar and inertial navigation data, the intermittent loss of satellite signals, etc. will still cause the positioning results after fusion to have lag or mutation. Therefore, there is an urgent need for a high-precision underwater positioning method that can adaptively select core features and dynamically establish data correlation. SUMMARY

[0006] The purpose of the present application is to provide a high-precision positioning method for a combined navigation type underwater operation platform to solve the problems raised in the background art.

[0007] To achieve the above object, the application provides a high-precision positioning method for a combined navigation type underwater operation platform, which comprises the following steps:

[0008] Obtaining a multi-dimensional navigation data source of the underwater operation platform, wherein the multi-dimensional navigation data source comprises inertial navigation data, sonar positioning data and satellite positioning data;

[0009] Based on the multi-dimensional navigation data source, performing data fusion processing to generate a navigation data warehouse;

[0010] Based on the navigation data warehouse, extracting a navigation feature matrix;

[0011] Based on the navigation feature matrix, determining a feature linkage graph of the multi-dimensional navigation data source;

[0012] Based on the feature linkage graph, screening a core positioning feature sequence;

[0013] Based on the core positioning feature sequence, controlling the positioning process of the underwater operation platform.

[0014] Preferably, the step of extracting the navigation feature matrix based on the navigation data warehouse comprises separating navigation data of each dimension from the navigation data warehouse.

[0015] Performing feature extraction processing on the navigation data of each dimension to generate a dimension feature vector; integrating the dimension feature vectors of all dimensions to construct the navigation feature matrix.

[0016] Preferably, the step of determining the feature linkage graph of the multi-dimensional navigation data source based on the navigation feature matrix comprises selecting a target dimension in the navigation feature matrix and obtaining a dimension feature vector of the target dimension.

[0017] Calculating the feature correlation degree between navigation features in the dimension feature vector; based on the feature correlation degree, generating a dimension linkage graph of the target dimension.

[0018] Iterating through all dimensions to repeatedly generate the dimension linkage graph step.

[0019] Based on the dimension linkage graphs of all dimensions, connecting the feature correlation degrees of key navigation features to generate the feature linkage graph of the multi-dimensional navigation data source.

[0020] Preferably, the step of screening the core positioning feature sequence based on the feature linkage graph comprises extracting a dimension entropy value corresponding to each navigation feature in the navigation feature matrix.

[0021] Extracting the feature correlation degree of each navigation feature from the feature linkage graph.

[0022] Combining the dimension entropy value and the feature correlation degree, calculating a core entropy value of each navigation feature.

[0023] Filtering, as the core positioning features, the navigation features higher than the preset threshold based on the core entropy values;

[0024] Generating a core positioning feature sequence by aggregating all the core positioning features.

[0025] Preferably, the controlling of the positioning process of the underwater operation platform based on the core positioning feature sequence comprises: analyzing the common mode in the core positioning feature sequence, and extracting reusable positioning parameters;

[0026] analyzing the difference mode in the core positioning feature sequence, and extracting specific scene positioning parameters;

[0027] Adjusting the positioning parameters of the underwater operation platform based on the reusable positioning parameters and the specific scene positioning parameters.

[0028] Preferably, the method further comprises: classifying the positioning state types of the underwater operation platform based on the navigation data warehouse, wherein the positioning state types include stable positioning state and unstable positioning state;

[0029] Attaching state priority labels based on the positioning state types;

[0030] Classifying the resource consumption types based on the multi-dimensional navigation data sources, wherein the resource consumption types include computing resource consumption and communication resource consumption;

[0031] Attaching resource type labels based on the resource consumption types.

[0032] Preferably, triggering the positioning parameter adjustment signal based on the positioning state types and the resource type labels comprises: generating a first positioning parameter adjustment signal when an unstable positioning state is detected;

[0033] Generating a second positioning parameter adjustment signal when a resource consumption type change is detected;

[0034] Generating a positioning parameter replacement node based on the first positioning parameter adjustment signal and the second positioning parameter adjustment signal;

[0035] The positioning parameter replacement node comprises a parameter adjustment node within the same stable positioning state and a parameter reset node when switching between different positioning state types.

[0036] Preferably, the calculation of the positioning offset parameter based on the positioning parameter replacement node comprises: extracting the parameter adjustment node within the same stable positioning state, and marking it as the same positioning parameter replacement node;

[0037] Extracting the parameter reset node when switching between different positioning state types, and marking it as a different positioning parameter replacement node;

[0038] The parameter offset in the stable positioning state is calculated based on the same positioning parameter replacement node, and is recorded as a first positioning offset parameter;

[0039] The parameter offset when the positioning state type is switched is calculated based on the different positioning parameter replacement node, and is recorded as a second positioning offset parameter;

[0040] The first positioning offset parameter and the second positioning offset parameter are combined to generate a positioning offset parameter.

[0041] Preferably, the optimization of the positioning control based on the positioning offset parameter comprises: analyzing the reusable mode in the positioning offset parameter, and extracting a migration positioning parameter;

[0042] The specific mode in the positioning offset parameter is analyzed, and an independent positioning parameter is extracted;

[0043] A global positioning reference is set based on the migration positioning parameter;

[0044] The parameter of a specific positioning scene is dynamically adjusted based on the independent positioning parameter;

[0045] The adaptive positioning control is realized by combining the global positioning reference and the dynamically adjusted parameter.

[0046] Preferably, the method further comprises: setting a pre-warning mechanism and an exit mechanism of the high-precision positioning state;

[0047] When the pre-warning mechanism is triggered, the underwater operation platform enters the high-precision positioning state, and all new navigation data enters a priority sequence processing;

[0048] The sequence processing order is optimized based on the state priority label and the resource type label;

[0049] When the exit mechanism is triggered, the underwater operation platform exits the high-precision positioning state, and the new navigation data enters a normal processing sequence.

[0050] Compared with the prior art, the beneficial effects of the present application are:

[0051] The positioning precision of the underwater operation platform is significantly improved by the fusion of the multi-dimensional navigation data source and the construction of the feature linkage graph. The short-term stability of the inertial navigation data and the spatial constraint of the sonar positioning form a complement, and the satellite positioning data provide absolute position calibration, so that the three work together to reduce the inherent defects of a single data source. The feature linkage graph dynamically reflects the correlation strength of different data sources, and avoids the error amplification problem caused by fixed weight fusion.

[0052] The screening mechanism of the core positioning feature sequence can adapt to environmental changes. When the sonar signal is disturbed by multipath effect, the system automatically reduces the weight of the sonar data and enhances the short-term prediction of the inertial navigation; when the satellite signal is lost, the positioning continuity is maintained through the deep coupling of sonar and inertial data. This dynamic adjustment characteristic effectively suppresses the positioning jump phenomenon commonly seen in traditional methods.

[0053] The structured storage mode of the navigation data warehouse supports fast retrieval and comparison of historical data, providing real-time reference for updating the feature matrix. In the data fusion process, the time and space alignment algorithm eliminates the sampling frequency difference of multi-source data, ensuring the time sequence consistency of feature extraction. This method shows stronger adaptability in complex underwater terrain, such as in steep slopes or obstacle-dense areas, the matching accuracy of the geometric features of sonar data and the inertial trajectory is higher.

[0054] The construction process of the feature linkage map introduces a topology optimization strategy to reduce the consumption of computing resources. Compared with the traditional machine learning method which relies on offline training mode, the online feature screening mechanism of the present invention reduces the dependence on historical data and is more suitable for dynamic scenes in actual operations. The synergistic effect of multi-dimensional data sources prolongs the effective working time of the system and maintains the positioning accuracy in long-time deep water operations. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 The working principle diagram of the high-precision positioning method of the combined navigation type underwater operation platform described in the present invention;

[0056] Figure 2 The flowchart for extracting navigation feature matrix;

[0057] Figure 3 The flowchart for screening core positioning feature sequence;

[0058] Figure 4 The flowchart for classifying and labeling the positioning state type and resource consumption type. DETAILED DESCRIPTION

[0059] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present invention.

[0060] Please refer to Figure 1 The present invention provides a high-precision positioning method for a combined navigation type underwater operation platform, which comprises:

[0061] The precise position control of the underwater operation platform is achieved by integrating multi-dimensional navigation data sources. The multi-dimensional navigation data sources of the underwater operation platform include inertial navigation data, sonar positioning data, and satellite positioning data. These data sources are collected in real time through sensors and receiving equipment. The inertial navigation data is derived from the inertial measurement unit (IMU) on the platform, providing acceleration and angular velocity information. The sonar positioning data is obtained through the underwater sonar system, measuring the distance and direction relative to the reference point. The satellite positioning data is obtained through the global navigation satellite system (GNSS) receiver, providing global position coordinates when the platform approaches the water surface. All data sources are synchronized with time stamps to ensure time consistency.

[0062] Data fusion processing is performed to generate a navigation data warehouse. The data fusion uses the Kalman filter algorithm to weight and fuse information from different data sources to reduce noise and errors. The navigation data warehouse is a structured database that stores historical and real-time navigation data, including position, velocity, attitude, and time information. The design of the data warehouse supports efficient query and data retrieval, facilitating subsequent feature extraction. Based on the navigation data warehouse, a navigation feature matrix is extracted. The navigation feature matrix is a multi-dimensional matrix, where each row represents a time point and each column represents a navigation feature. The feature extraction process involves calculating derived indicators from the original data, such as position deviation, velocity change rate, attitude angle, etc. These features are used to capture the dynamic behavior of the navigation data.

[0063] Based on the navigation feature matrix, a feature linkage graph is determined for the multi-dimensional navigation data sources. The feature linkage graph is a graph structure, with nodes representing navigation features and edges representing the degree of association between features. The construction of the graph is achieved by calculating the correlation coefficient or mutual information between features to reveal the internal relationship between data sources. Based on the feature linkage graph, a core positioning feature sequence is selected. The core positioning feature sequence is a group of high-impact features that significantly contribute to positioning accuracy. The selection process is based on the importance score of the features, and features with scores higher than the threshold are selected to form the sequence. Based on the core positioning feature sequence, the positioning process of the underwater operation platform is controlled. The control process involves adjusting the positioning parameters of the platform, such as filter settings, sensor weights, or output frequency, to optimize positioning performance. The control algorithm adjusts the parameters adaptively according to the dynamic changes of the core feature sequence, ensuring that the platform maintains high-precision positioning in various underwater environments.

[0064] Example 1: see Figure 2The navigation feature matrix is extracted from the navigation data warehouse, and a multi-dimensional navigation data source feature linkage graph is constructed. The navigation data warehouse, as a centralized storage system, contains raw data records from inertial navigation units, sonar positioning devices, and satellite receivers. These data are usually stored in time series, with each data point having an accurate timestamp to ensure time consistency. The separation operation needs to be divided according to the data source and type, such as classifying the acceleration and angular velocity data output by the inertial measurement unit as inertial navigation data dimensions, classifying the distance and azimuth angle measured by the sonar device as sonar positioning data dimensions, and classifying the latitude and longitude coordinates provided by the satellite receiver as satellite positioning data dimensions. In this process, special attention needs to be paid to data synchronization problems, as the sampling frequencies of different sensors may differ, and interpolation or sampling rate unification methods are needed to align all data on the time axis. After data separation, each dimension's data set is extracted separately to form a data subset for further processing. These data subsets not only contain raw measurement values, but also may contain some preliminary processed derivative data, such as sensor readings after preliminary filtering or position information after coordinate conversion.

[0065] After data separation, feature extraction processing is needed for each dimension of navigation data. The purpose of feature extraction is to extract quantitative indicators that can represent the characteristics of the data from the original data. For inertial navigation data, statistical features within a moving window may be calculated, such as the mean and variance of acceleration data within a sliding time window. These statistical quantities can reflect the stability of the motion state. Frequency domain features can also be calculated by performing Fourier transform on the acceleration data to obtain the frequency energy distribution, which helps to identify periodic motion patterns. For sonar positioning data, feature extraction may include the calculation of signal quality indicators, such as the signal-to-noise ratio of the echo signal and the strength estimation of the multipath effect. These features can help evaluate the reliability of sonar measurement data. Feature extraction of satellite positioning data focuses on positioning accuracy indicators, such as the calculation of horizontal and vertical dilution of precision, and the trend of satellite signal strength. Each feature extraction process needs to select appropriate algorithms and parameter settings according to the specific data characteristics. All the extracted feature values are organized into dimension feature vectors, each representing the feature set of the dimension data at a specific time point. The length of the dimension feature vector depends on the number of features extracted, and usually tens of feature indicators are extracted for each dimension.

[0066] For all completed dimension feature vectors, they need to be integrated into a navigation feature matrix, and the integration process needs to consider the time alignment problem to ensure that the feature vectors of different dimensions correspond at the time point. The integrated navigation feature matrix is a two-dimensional array structure, with rows representing time points and columns representing feature indicators. The number of columns of the matrix is equal to the total number of all dimension features, and the number of rows is equal to the number of time points. After constructing the matrix, data standardization processing is usually needed, because different features may have different dimensions and numerical ranges. Standardization processing can convert all feature values to the same numerical range, and common methods include minimum maximum standardization and standard deviation standardization. The standardized feature matrix is more conducive to subsequent feature correlation analysis, as it eliminates the influence of dimensional differences. Based on the navigation feature matrix, to determine the feature linkage graph of the multi-dimensional navigation data source, the target dimension in the navigation feature matrix needs to be selected. The selection of the target dimension can be arbitrary, and usually starts with the dimension with the most abundant or most important data. Obtaining the dimension feature vector of the target dimension means extracting all feature columns belonging to this dimension from the feature matrix. For example, if the inertial navigation dimension is selected as the target dimension, all feature columns related to inertial navigation need to be extracted from the matrix. These feature columns constitute the feature vector set of this dimension at all time points.

[0067] The next important step is to calculate the feature correlation degree between the navigation features in the dimension feature vector. The feature correlation degree measures the degree of correlation between different features, which may be linear or nonlinear. Linear correlation coefficients such as Pearson correlation coefficients are usually used to measure the linear correlation strength between two features. For nonlinear relationships, mutual information and other indicators may be used to measure the degree of dependence between features. During the calculation process, all feature pairs need to be traversed, and for a dimension with features, the correlation between features needs to be calculated. The correlation values generated by these calculations are stored in a correlation matrix, which is symmetric, and the elements on the diagonal are usually set to 1, indicating complete correlation of the feature with itself.

[0068] Based on the calculated feature correlation, a dimension linkage graph for the target dimension can be generated. The dimension linkage graph is a graph structure where nodes represent features and edges represent the correlation between features. The weight of the edge is determined by the feature correlation. Only edges with correlation above a certain threshold are usually kept, so as to avoid keeping accidental correlations that might be caused by noise. The graph can be constructed using graph theory algorithms, with each feature as a node and the correlation as the edge weight, forming a weighted undirected graph. This graph structure can visually demonstrate the closeness of the relationship between features in the dimension. It is necessary to repeat the above steps for all dimensions, as each dimension needs to generate its own dimension linkage graph. For the sonar positioning dimension, the feature columns of this dimension need to be extracted, the correlation between these features is calculated, and the linkage graph of the sonar positioning dimension is constructed. Similarly, the same operation is performed for the satellite positioning dimension. Each dimension will produce an independent dimension linkage graph, and these graphs reflect the relationship structure between the features in their respective dimensions.

[0069] Based on the dimension linkage graphs of all dimensions, a feature linkage graph atlas of the multi-dimensional navigation data source is generated by connecting the feature correlations of key navigation features. Key navigation features are features that appear in multiple dimensions or have special importance. For example, the location estimation feature may exist in both the inertial navigation and satellite positioning dimensions. Such features become the bridge connecting different dimensions. The connection process is to identify these cross-dimensional features and connect the linkage graphs of different dimensions through these common features. The resulting feature linkage graph atlas is a comprehensive graph structure that shows the correlation between all features in all dimensions, including intra-dimension and inter-dimension correlations. This graph atlas can help understand the internal relationship between different data sources and provide a basis for subsequent feature screening and analysis. The storage of the graph atlas usually adopts the form of adjacency list or adjacency matrix, while retaining the metadata information of all features, such as feature name, dimension to which it belongs, etc.

[0070] Embodiment 2: see Figure 3, based on the feature linkage graph, and use the sequence to realize the accurate control of the positioning process of the underwater operation platform. The dimension entropy value is an index for measuring the uncertainty of the distribution of the characteristic value, which is calculated based on the concept in information theory. Specifically, for each feature column in the navigation feature matrix, the system will construct a probability distribution model of the feature value at different time points. This model can be realized by histogram statistics or kernel density estimation. Based on the probability distribution, the information entropy formula is used for calculation, which can quantify the chaos degree or randomness of the feature value. For example, if the value of a feature remains relatively stable in the time series, the change amplitude is small, and the entropy value is low, indicating that the feature contains relatively limited and predictable information; on the contrary, if the feature value fluctuates violently and presents high randomness, the entropy value is high, indicating that the feature may carry more dynamic information but may also introduce more noise. In actual operation, this calculation process needs to traverse the navigation feature matrix column by column, model the probability of each column of feature value sequence, and calculate the entropy value. The result is stored as an entropy vector corresponding to the feature column, which will be used for subsequent comprehensive evaluation.

[0071] On the basis of obtaining the dimension entropy value, the system further extracts the feature correlation degree of each navigation feature from the feature linkage graph. As a graph structure, the feature linkage graph has nodes representing each navigation feature and edges representing the correlation between features. The weight of the edge represents the numerical value of the feature correlation degree. The feature correlation degree reflects the mutual influence strength between different features. The higher the value, the stronger the collaboration or dependence between features. The extraction process involves traversing all nodes in the feature linkage graph. For each node corresponding to a feature, collect the weight values of all adjacent edges, and calculate a representative correlation degree score through weighted average or maximum pooling aggregation operation. For example, if a feature has strong connections with multiple other features, its correlation degree score will be higher, indicating that the feature is in a core position in the overall data relationship; on the contrary, if a feature has fewer connections or lower connection weights, its correlation degree score will be lower, indicating that the feature may be more independent or have less influence. The output of this step is a correlation degree vector, where each element corresponds to the correlation degree score of a navigation feature.

[0072] The system needs to combine the dimension entropy value and the feature correlation degree to calculate the core entropy value of each navigation feature. The core entropy value is a comprehensive index that aims to balance the information content of the feature itself and its influence in the overall graph. Its calculation formula is:

[0073]

[0074] where: represents the core entropy value, denotes the dimension entropy value, denotes the feature correlation degree, and a and b are weight coefficients for adjusting the relative importance of entropy value and correlation degree in comprehensive evaluation. The setting of weight coefficients needs to be adjusted according to specific application scenarios. For example, in a scenario emphasizing data stability, a higher weight may be given to the feature correlation degree to preferentially select those features with strong synergy with other features; while in a scenario focusing on information richness, a higher weight may be given to the dimension entropy value to retain those features carrying more dynamic information. After calculation, each navigation feature will obtain a core entropy value score, which is used for subsequent feature screening. Based on the core entropy value, the system screens out navigation features higher than a preset threshold as core positioning features. The preset threshold is usually determined according to historical data or experimental experience, for example, it can be set to the upper quartile of the core entropy value sequence or a dynamically adjusted statistical quantile. The screening process traverses all navigation features, compares their core entropy values with the threshold, and selects those features with higher scores into the core set. These core positioning features usually represent key factors that significantly contribute to positioning accuracy, such as high-precision position estimation features or stable speed measurement features. The system aggregates all screened core positioning features, sorts them from high to low according to core entropy values, and forms a core positioning feature sequence. This sequence serves as an ordered set to provide input for subsequent positioning control.

[0075] After obtaining the core positioning feature sequence, the system begins to control the positioning process of the underwater operation platform based on the sequence. This control process first involves common mode analysis of the features in the sequence, aiming to extract reusable positioning parameters. Common mode refers to stable behavior patterns that appear in multiple core features, for example, in a stable positioning state, multiple features may simultaneously exhibit low variance or high autocorrelation. Pattern recognition usually uses clustering analysis or principal component analysis to divide the feature value sequence into different mode categories, and extracts representative parameters such as average offset or standard deviation from each category. These reusable positioning parameters form a basic parameter set suitable for common positioning scenarios, providing a stable benchmark configuration for the system. At the same time, the system also analyzes the difference mode in the core positioning feature sequence to extract specific scene positioning parameters. Difference mode refers to unique feature behavior that appears under specific conditions or abnormal situations, for example, when sonar signals are disturbed, related features may exhibit abnormal fluctuations or mutations.

[0076] The difference pattern detection typically employs anomaly detection algorithms or change-point detection techniques to identify outliers or abrupt changes in the feature sequence and extract corresponding parameter adjustment values such as temporary filter coefficients or sensor weight adjustment amounts. These scenario-specific positioning parameters are targeted at atypical or emergent situations, providing the system with flexible configurations for special environments. The system dynamically adjusts the positioning parameters of the underwater operation platform based on reusable positioning parameters and scenario-specific positioning parameters. Parameter adjustment is achieved through control algorithms, for example, using a proportional-integral-derivative controller or a model predictive control algorithm, which calculates output parameter adjustment instructions based on real-time input core positioning feature values. Adjustment operations may involve modifying filter cutoff frequencies, adjusting sensor data fusion weights, or changing positioning output frequencies. The entire control process forms a closed-loop feedback, with the system continuously monitoring changes in core positioning feature sequences and dynamically updating parameter settings to ensure high-precision positioning of the underwater operation platform in various operating environments. During implementation, computational efficiency may need to be considered, with incremental learning or sliding window techniques used to optimize real-time performance and ensure that the system can quickly respond to environmental changes.

[0077] Example 3: refer to Figure 4 The positioning state types and resource consumption types of the underwater operation platform are classified, and positioning parameter adjustment signals are triggered based on these classifications. The classification process is achieved by analyzing historical and real-time positioning data stored in the navigation data warehouse, including position coordinates, velocity vectors, attitude angles, and other multivariate time series information. The system uses sliding time window technology to statistically analyze positioning data within each time window, calculating indicators such as the variance of position data, the amplitude of velocity changes, and the fluctuation range of attitude angles within the window. For example, for a 30-second time window, the system calculates the three-dimensional coordinate variance of all position points within the window. If the variances in the three coordinate axes are all below the preset threshold, the positioning state during this time period is classified as stable positioning; otherwise, if the variance of any one direction exceeds the threshold, it is classified as unstable positioning. In addition to variance analysis, the system also examines the continuity and smoothness of the data by calculating the position change rate between adjacent time points to detect abnormal jumps, which often indicate that the positioning system is disturbed or malfunctioning. The classification algorithm can use a rule-based method or a machine learning classifier such as a support vector machine or a random forest to establish a classification model by training historical data. Regardless of the method used, the classification results are stored in the form of time series labels, with each time point corresponding to an identification of the positioning state type.

[0078] After completing the classification of positioning state types, the system will attach a state priority label to each state. The state priority label is a quantitative indicator representing the importance or urgency of the state, and its value is based on the nature of the state type itself and factors such as the duration of the state. For example, unstable positioning states are usually assigned a high priority label, as such states indicate that there may be problems with the positioning system that need to be addressed promptly; while stable positioning states are assigned a normal priority label. The calculation of priority can use the formula:

[0079]

[0080] where Ps represents the state priority, represents the state type indicator (e.g. 1 for unstable state, 0 for stable state), D represents the state duration, and are weight coefficients. The meaning of this formula is to consider the influence of state type and duration on priority, for example, a long-lasting unstable state will get a higher priority score than a short unstable state. The value of the weight coefficient needs to be adjusted according to the actual application scenario, for example, in applications with high security requirements, the state type indicator may be given a greater weight. The addition of the priority label allows the system to distinguish the urgency of different states, providing a basis for subsequent resource scheduling and parameter adjustment.

[0081] The system classifies resource consumption types based on multi-dimensional navigation data sources, including two main categories: computing resource consumption and communication resource consumption. Computing resource consumption specifies the positioning system's demand for computing resources in data processing, algorithm operation, etc., and its quantitative indicators include CPU utilization, memory usage, computing delay, etc.; communication resource consumption refers to the system's demand for network resources in data transmission, communication interaction, etc., and its quantitative indicators include bandwidth occupancy, data transmission volume, communication delay, etc. The classification process is achieved by monitoring system resource usage, and the system will regularly collect various resource usage indicators and classify them into high consumption or low consumption based on pre-set thresholds. For example, if the CPU utilization exceeds 70% for a continuous period of time, the current computing resource consumption will be classified as high computing resource consumption; if the bandwidth occupancy is less than 20%, the communication resource consumption will be classified as low communication resource consumption. The classification results are also stored in the form of time series labels, each time point with the corresponding resource consumption type identifier.

[0082] Based on the classification results of resource consumption types, the system will attach resource type labels to each resource consumption type. Resource type labels are identifiers that describe the characteristics of resource consumption. They not only identify the consumption type, but also the consumption level. For example, the "high computing resource" label indicates that the computing resource consumption is at a high level, and the "low communication resource" label indicates that the communication resource consumption is at a low level. The addition of labels enables the system to quickly identify the current resource status, providing information support for subsequent resource management decisions.

[0083] Based on the positioning state type and resource type label, the system will trigger corresponding positioning parameter adjustment signals. When an unstable positioning state is detected, the system generates a first positioning parameter adjustment signal. The detection process is achieved by real-time monitoring of the positioning state label. The system will continuously check the current state label, and as soon as it finds that the label has become unstable, it will immediately trigger signal generation. The first positioning parameter adjustment signal contains information about the unstable state, such as the start time, duration, severity, etc. These information helps subsequent parameter adjustment decisions. When a change in resource consumption type is detected, the system generates a second positioning parameter adjustment signal. Resource consumption type change refers to the transition of resource consumption level from high to low or from low to high, which may affect the performance and stability of the system. The detection mechanism is achieved by comparing the resource type labels of adjacent time points. If a change in label is found, signal generation is triggered. The second positioning parameter adjustment signal contains specific information about the resource type change, such as the resource consumption level before and after the change, the time point of the change, etc.

[0084] Based on the first positioning parameter adjustment signal and the second positioning parameter adjustment signal, the system will generate positioning parameter replacement nodes. Positioning parameter replacement nodes are specific points on the time axis, indicating the time when positioning parameter adjustment is needed. The generation of nodes is based on the time information and content information of the signals. The system will determine the timing of node generation according to the severity and urgency of the signals. Nodes include two types: parameter adjustment nodes within the same stable positioning state and parameter reset nodes when switching between different positioning state types. Parameter adjustment nodes within the same stable positioning state are used to fine-tune and optimize parameters while maintaining the current positioning state; parameter reset nodes when switching between different positioning state types are used to make large adjustments to parameters when the state changes fundamentally. The generation logic of nodes considers multiple factors, such as the importance of signals, the current load of the system, historical adjustment effects, etc., to ensure that the timing and amplitude of parameter adjustment are just right.

[0085] Example 4: Computing positioning offset parameters based on positioning parameter replacement nodes and using these parameters to optimize positioning control. Positioning parameter replacement nodes are specific time points or event points derived from parameter adjustment signals, which mark the critical moments when parameter adjustment is needed. The system extracts from the node sequence those parameter adjustment nodes that occur within the same stable positioning state. These nodes typically correspond to moments when fine-tuning of parameters is needed while maintaining the current positioning state unchanged. The extraction process is based on matching of timestamps and state labels. The system checks the timestamp of each node to confirm whether the corresponding positioning state label is a stable state and the duration of this stable state has exceeded a minimum threshold to ensure the stability of the state. Nodes that meet these conditions are marked as the same positioning parameter replacement node, meaning that these nodes are all parameter adjustment requirements that occur in the same positioning state. The system extracts those parameter reset nodes that occur at the switching of different positioning state types. These nodes correspond to moments when parameter resetting is needed when the positioning state fundamentally changes. The extraction process is achieved by detecting the change points of state labels. The system identifies those time points where the state type changes and marks parameter reset nodes at these time points. These nodes are marked as different positioning parameter replacement nodes, indicating that these nodes are associated with the transition of the positioning state and the reconfiguration of the parameters. The entire extraction and marking process requires precise time synchronization and state tracking to ensure that each node can be correctly classified and marked.

[0086] After the extraction and labeling of the nodes, the system starts to calculate the positioning bias parameters. For the same positioning parameter, the system calculates the parameter bias within the stable positioning state, which reflects the magnitude of the parameter adjustment needed to maintain the state stability. The calculation process is based on the parameter value changes on the node time series. The system compares the parameter differences between adjacent nodes, and uses difference calculation or proportional change calculation to determine the bias. For example, for the cutoff frequency parameter of the position filter, the system calculates the absolute difference or relative change rate between the current node value and the last node value. This difference is the bias of the parameter within the current stable state. All these biases are recorded as the first positioning bias parameters, which reflect the natural drift or gradual adjustment needs of the parameters in the stable state. For different positioning parameter replacement nodes, the system calculates the parameter bias when the positioning state type switches, which reflects the magnitude of the parameter adjustment needed when the state transitions. The calculation process is based on the comparison of parameter values before and after the state switch. The system extracts the parameter value at the last time point before the state switch and the parameter value at the first time point after the state switch, and calculates the difference between the two. For example, when switching from a stable state to an unstable state, the system may need to make a large adjustment to the process noise parameter of the Kalman filter. The magnitude of this adjustment is the bias of the parameter when the state switches. All these biases are recorded as the second positioning bias parameters, which reflect the sudden adjustment needs of the parameters when the state transitions. The system combines the first positioning bias parameters and the second positioning bias parameters to form a complete set of positioning bias parameters. The combination method can be simple vector splicing or weighted fusion, depending on the importance and reliability of different bias parameters. The combined positioning bias parameters provide a panoramic view of the system parameter changes, including gradual changes in the stable state and sudden changes in the state transition. See Table 1.

[0087] Table 1: Parameter replacement node data as follows.

[0088] Node ID Timestamp Node Type Positioning State Parameter Type Old Parameter Value New Parameter Value Offset N001 2025-06-1008:30:15 Same Node Stable State Filter Cutoff Frequency 1.5 Hz 1.8 Hz 0.3 Hz N002 2025-06-1008:35:22 Same Node Stable State Data Fusion Weight 0.6 0.7 0.1 N003 2025-06-1008:40:18 Different Node Stable -> Unstable Process Noise Parameter 0.01 0.05 0.04 N004 2025-06-1008:45:30 Different Node Unstable -> Stable Measurement Noise Parameter 0.02 0.01 -0.01

[0089] Based on the positioning offset parameters, the system optimizes the positioning control, and the optimization process analyzes reusable patterns in the positioning offset parameters, which are parameter variation rules that repeatedly appear in multiple offset instances. The system uses pattern recognition algorithms to analyze the sequence of offset parameters and identify offset patterns that repeatedly appear under similar conditions. For example, the system may find that the process noise parameter needs to be increased by a range of 0.03 to 0.05 every time it switches from a stable state to an unstable state, and this range value is a reusable pattern. From these reusable patterns, the system extracts migration positioning parameters, which are parameter adjustment values that are universally applicable and can be applied to similar scenarios. The extraction of migration positioning parameters is based on statistical analysis methods, such as calculating the mean, variance, or confidence interval of the offset, to determine the most representative parameter value. The system analyzes specific patterns in the positioning offset parameters, which are unique parameter variations that only occur under specific conditions. Specific patterns are usually related to abnormal situations or special environmental conditions, such as special parameter adjustment requirements when encountering strong water flow interference. From these specific patterns, the system extracts independent positioning parameters, which are specialized adjustment values for specific scenarios and do not have universal applicability. Based on the migration positioning parameters, the system sets a global positioning reference, which is a set of basic parameter values that provides a default configuration scheme for the system.

[0090] The setting of the global positioning reference is based on the statistical characteristics of the migration positioning parameters, such as using the mode or median of the offset as the reference value. Based on the independent positioning parameters, the system implements dynamic adjustment of parameters for specific positioning scenarios. This adjustment process is adaptive, and the system will monitor environmental conditions and working states in real time, and automatically apply the corresponding independent positioning parameters when a specific scenario is detected. The system combines the global positioning reference and dynamic adjustment parameters to achieve adaptive positioning control. This combination creates a hierarchical control structure, with the global reference providing a stable foundation and dynamic adjustment providing flexible response, enabling the system to maintain optimal positioning performance under various working conditions. The entire optimization process is continuous, and the system will continuously learn new offset patterns, update migration positioning parameters and independent positioning parameters, and gradually improve the accuracy and adaptability of control.

[0091] Example 5: A warning and exit mechanism for high-precision positioning is established, and the data processing sequence is optimized based on these mechanisms. The warning mechanism is designed based on continuous monitoring of the navigation system's operating status. The system analyzes data streams from multiple sensors in real time, including the output of the inertial measurement unit, readings from the acoustic positioning system, and availability indicators of satellite navigation signals. After preliminary processing, these data generate a series of performance indicators, such as the eigenvalues ​​of the covariance matrix of position estimation, filter residual statistics, and time-varying characteristics of signal strength. The determination of warning conditions depends on a comprehensive evaluation of these performance indicators. When key indicators exceed preset threshold ranges, the system triggers a warning state. The evaluation process uses a weighted summation model:

[0092]

[0093] Where: W represents the comprehensive early warning index, and W represents the first... The standardized values ​​of each performance metric This represents the weighting coefficient of the corresponding indicator. The value range is [0,1], and the larger the value, the more serious the deviation of the indicator from the normal range. The value of reflects the relative importance of different indicators in early warning judgment, and is usually determined by domain experts based on the actual application scenario. This indicates the total number of performance metrics included in the evaluation. When When the value exceeds a preset warning threshold, the system determines that it needs to enter a high-precision positioning state. For example, when the uncertainty of position estimation increases significantly, manifested as the main eigenvalues ​​of the covariance matrix continuously exceeding the normal range, or when the consistency between multi-sensor data suddenly decreases, manifested as abnormal statistical characteristics of filter residuals, the system will determine that the current positioning accuracy is at risk. The warning mechanism not only considers instantaneous indicator values ​​but also analyzes the temporal characteristics of the indicators, using a sliding window statistical method to identify trend changes and avoid false triggering due to brief interference. Once the warning conditions are met, the system immediately initiates a state transition process, switching the underwater operation platform's positioning system to a high-precision positioning mode. This mode is characterized by higher investment in computing and communication resources, aiming to restore and maintain positioning accuracy through enhanced data processing capabilities.

[0094] After entering the high-precision positioning state, all newly arrived navigation data is sent to the priority processing sequence. The management of this sequence adopts a priority-based scheduling strategy, and each data packet is assigned a different priority weight according to the importance of its source sensor, data freshness, and expected contribution to positioning accuracy. Data freshness is determined by the difference between the acquisition timestamp and the current time, with a smaller difference indicating higher freshness. The evaluation of expected contribution relies on the historical performance record of the data source and the reliability estimate under the current environmental conditions. The calculation of priority weight integrates these factors, and the system generates a priority score for each data packet, with high-score data packets being arranged for priority processing. This optimization of processing order ensures that data that is most beneficial to improving positioning accuracy can be utilized in a timely manner, while not completely ignoring low-priority data, but balancing processing efficiency and positioning performance through a reasonable scheduling strategy. The state priority label and resource type label play an important role in this process, providing supplementary information for the priority score. The state priority label indicates the urgency of the current positioning state, and in high-urgency states, the system further shortens the processing delay of high-priority data. The resource type label reflects the demand for computing and communication resources of different data processing tasks, and the system dynamically adjusts the scheduling strategy according to available resources, prioritizing data processing tasks that are efficient in resource usage when resources are scarce. The implementation of the entire priority processing sequence requires efficient queue management algorithms and real-time schedulers to maximize the performance of the positioning system under limited resources.

[0095] The design of the exit mechanism corresponds to the warning mechanism, and its trigger conditions are based on the recovery of positioning performance and the consideration of resource constraints. The system continuously monitors positioning accuracy indicators, and when these indicators remain within the normal range and remain stable, the system determines that the high-precision positioning state can be exited. The exit decision not only considers accuracy indicators, but also evaluates resource usage, and if maintaining a high-precision state leads to long-term high resource consumption, the system may also trigger the exit mechanism to ensure stable operation of the overall system. The exit process adopts a gradual approach, gradually reducing data processing priority and resource allocation, avoiding performance fluctuations caused by sudden state switching. When the exit mechanism is triggered, the underwater operating platform exits the high-precision positioning state, and newly collected navigation data no longer enters the priority processing sequence, but is transferred to the normal processing flow. The normal processing sequence adopts a standard data processing strategy and no longer gives priority processing rights to specific data, but processes according to time sequence or default scheduling strategy. This processing method requires relatively less resources and is more suitable for regular operation state. The implementation of the entire warning and exit mechanism requires careful parameter tuning and state management to ensure that the system can respond to changes in positioning performance in a timely manner, while maintaining a good balance between resource constraints and positioning accuracy.

[0096] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting; it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and illustrative figures, it should be apparent that the scope of the present application is not limited to these specific embodiments.

[0097] While the embodiments of the application have been shown and described herein, it is to be understood that the scope of the application, jointly pointed out in the appended claims, is not to be limited to the above-described embodiments but can be otherwise variously changed, modified, replaced, and altered within the principles and spirit of the present application.

Claims

1. A method for high-precision positioning of a combined navigation type underwater operation platform, comprising the following steps: obtaining a multi-dimensional navigation data source of the underwater operation platform, the multi-dimensional navigation data source comprising inertial navigation data, sonar positioning data and satellite positioning data; performing data fusion processing based on the multi-dimensional navigation data source to generate a navigation data warehouse; extracting a navigation feature matrix based on the navigation data warehouse; determining a feature linkage graph of the multi-dimensional navigation data source based on the navigation feature matrix; screening a core positioning feature sequence based on the feature linkage graph; controlling a positioning process of the underwater operation platform based on the core positioning feature sequence; the step of determining the feature linkage graph of the multi-dimensional navigation data source based on the navigation feature matrix comprises: selecting a target dimension in the navigation feature matrix, and obtaining a dimension feature vector of the target dimension; calculating a feature correlation degree between navigation features in the dimension feature vector, and generating a dimension linkage graph of the target dimension based on the feature correlation degree; iterating the step of generating the dimension linkage graph; generating the feature linkage graph of the multi-dimensional navigation data source by connecting the feature correlation degrees of key navigation features based on the dimension linkage graphs of all dimensions; the step of screening the core positioning feature sequence based on the feature linkage graph comprises: extracting a dimension entropy value corresponding to each navigation feature in the navigation feature matrix; extracting a feature correlation degree of each navigation feature from the feature linkage graph; calculating a core entropy value of each navigation feature by combining the dimension entropy value and the feature correlation degree; screening navigation features with core entropy values higher than a preset threshold as core positioning features based on the core entropy values; aggregating all the core positioning features to generate a core positioning feature sequence.

2. The method of claim 1, wherein the extracting a navigation feature matrix based on the navigation data repository comprises: separating navigation data of each dimension from the navigation data warehouse; performing feature extraction processing on the navigation data of each dimension to generate a dimension feature vector, and integrating the dimension feature vectors of all dimensions to construct a navigation feature matrix.

3. The method of claim 1, wherein the controlling the positioning process of the underwater operation platform based on the sequence of core positioning features comprises: analyzing common patterns in the core positioning feature sequence to extract reusable positioning parameters; analyzing difference patterns in the core positioning feature sequence to extract specific scene positioning parameters; adjusting positioning parameters of the underwater operation platform based on the reusable positioning parameters and the specific scene positioning parameters.

4. The method of claim 1, further comprising: classifying positioning state types of the underwater operation platform based on the navigation data warehouse, the positioning state types comprising stable positioning states and unstable positioning states; attaching state priority labels based on the positioning state types; classifying resource consumption types based on the multi-dimensional navigation data source, the resource consumption types comprising computing resource consumption and communication resource consumption; attaching resource type labels based on the resource consumption types.

5. The method of claim 4, wherein triggering a positioning parameter adjustment signal based on the positioning state type and the resource type label comprises: generating a first positioning parameter adjustment signal when an unstable positioning state is detected; generating a second positioning parameter adjustment signal when a resource consumption type changes; generating a positioning parameter replacement node based on the first positioning parameter adjustment signal and the second positioning parameter adjustment signal; the positioning parameter replacement node comprises a parameter adjustment node within the same stable positioning state and a parameter reset node when switching between different positioning state types.

6. The method of claim 5, wherein the positioning parameters are replaced based on the nodes, and the positioning offset parameters are calculated based on the positioning parameters. Extract the parameter adjustment nodes within the same stable positioning state, and mark them as the same positioning parameter replacement nodes; Extract the parameter reset nodes when switching between different positioning state types, and mark them as different positioning parameter replacement nodes; Based on the same positioning parameter replacement nodes, calculate the parameter offset within the stable positioning state, and mark it as the first positioning offset parameter; Based on the different positioning parameter replacement nodes, calculate the parameter offset when switching between different positioning state types, and mark it as the second positioning offset parameter; Combine the first positioning offset parameter and the second positioning offset parameter to generate the positioning offset parameter.

7. The method of claim 6, wherein the optimizing positioning control based on the positioning offset parameter comprises: Analyze the reusable mode in the positioning offset parameter, and extract the migration positioning parameter; Analyze the specific mode in the positioning offset parameter, and extract the independent positioning parameter; Based on the migration positioning parameter, set the global positioning reference; Based on the independent positioning parameter, dynamically adjust the parameters of the specific positioning scene; Combine the global positioning reference and the dynamically adjusted parameters to realize adaptive positioning control.

8. The method of claim 1, further comprising: Set the warning mechanism and exit mechanism of the high-precision positioning state; When the warning mechanism is triggered, the underwater operation platform enters the high-precision positioning state, and all new navigation data enters the priority sequence processing; Based on the state priority label and the resource type label, optimize the sequence processing order; When the exit mechanism is triggered, the underwater operation platform exits the high-precision positioning state, and the new navigation data enters the normal processing sequence.

Images