Airspace planning and optimization decision method and device based on historical flight knowledge base

Abstract

The embodiment of the application provides an airspace planning and optimization decision method and device based on a historical flight knowledge base, through innovatively designing a flight path processing system, through a meta-model and feature extraction, effective utilization of data is realized. A knowledge graph mechanism is constructed, combined with correlation analysis and weight calculation, a reliable matching model is established. A scoring screening is introduced, through candidate evaluation and path combination, the rationality of planning is ensured. The method effectively solves the deficiencies of traditional technologies in flight path processing, knowledge modeling and decision optimization, and provides technical support for flight path planning.

Description

Technical Field

[0001] This application relates to the field of data processing, specifically to a method and apparatus for airspace planning and optimization decision-making based on a historical flight knowledge base. Background Technology

[0002] Existing flight path planning methods have significant shortcomings. Traditional systems perform poorly in track processing and feature extraction, and fail to effectively utilize historical data, thus affecting planning results.

[0003] Furthermore, existing technologies suffer from bottlenecks in scene association and knowledge representation. Most systems lack robust association analysis mechanisms and knowledge modeling strategies, resulting in inaccurate path matching.

[0004] Existing systems have technical shortcomings in decision optimization. The lack of in-depth analysis of flight segment characteristics makes it difficult to achieve efficient route generation through scoring optimization, thus affecting planning quality. Solving these problems is crucial for improving flight path planning capabilities. Summary of the Invention

[0005] To address the problems in existing technologies, this application provides an airspace planning and optimization decision-making method and apparatus based on a historical flight knowledge base, which can effectively solve the shortcomings of traditional technologies in flight path processing, knowledge modeling and decision optimization, and provide technical support for flight path planning.

[0006] To solve at least one of the above problems, this application provides the following technical solution: Firstly, this application provides an airspace planning and optimization decision-making method based on a historical flight knowledge base, including: Collect historical flight track data streams, construct an airspace planning meta-model that includes flight segment boundary parameters and splitting rules, perform filtering preprocessing on the historical flight track data streams based on the airspace planning meta-model, determine flight segment switching points based on preset segmenting thresholds, calculate segmenting parameters to generate flight segment feature models, record the associated scene features of the flight segment feature models, and establish a flight segment dataset. Construct a multidimensional association matrix, calculate the membership coefficient between the flight segment feature model and the associated scene features, generate a weight index based on the similarity distribution, combine the membership coefficient and the weight index to form an association strength value, establish a knowledge graph containing flight segment features, scene attributes, and association strength, and construct a flight segment feature matching model. The system receives flight path planning requests, extracts planning constraint parameters, retrieves segment feature models with matching degrees exceeding a threshold based on the knowledge graph, calculates candidate segment scores using the segment feature matching models, sorts and filters according to the association strength values, combines them to generate complete routes, outputs flight path planning results, and writes the flight path planning results into the planning decision database.

[0007] Furthermore, it also includes: accessing multi-source historical track data, extracting track point position coordinates, heading angle, speed, and altitude parameters, performing time-series alignment on the data, constructing a track parameter sequence, generating data filtering rules based on a preset noise threshold, and forming a standardized track dataset; Based on the flight track dataset, establish flight segment classification boundary conditions, construct an airspace planning meta-model that includes location distribution, speed variation, and altitude variation, write the data filtering rules into the airspace planning meta-model, and generate a flight segment classification and preprocessing rule library.

[0008] Furthermore, it also includes: setting a segment switching judgment threshold based on the airspace planning meta-model, performing sliding window analysis on the trajectory data, calculating the parameter change rate between adjacent windows, marking segment switching points according to the comparison result of the parameter change rate and the judgment threshold, and generating a segment segmentation marking sequence; The flight track data is divided into multiple sub-segments using the segment segmentation marker sequence. The feature parameters of each sub-segment are calculated, a segment feature vector is constructed, the scene attributes associated with the segments are extracted, and a segment feature model data structure is established.

[0009] Furthermore, it also includes: constructing an association matrix for the flight segment feature model according to spatial attributes, temporal attributes, and scene attributes; calculating the data distribution difference value between each flight segment feature and the associated scene feature; generating a multi-dimensional feature vector based on the difference value; and constructing a feature mapping space. The feature mapping space is normalized, the spatial distribution density of the multidimensional feature vector is calculated, a membership calculation function is generated based on the density distribution, and the correlation coefficient matrix between the flight segment features and the scene features is generated using the membership calculation function.

[0010] Furthermore, it also includes: constructing a combination function of membership coefficient and weight index, calculating the correlation strength value for each pair of segment features and scene features, generating a correlation strength matrix, and establishing a mapping relationship table between segment features and scene features; Based on the mapping relationship table, a knowledge graph node set is constructed, and the association strength matrix is ​​written into the connection weights between nodes to generate a segment feature matching model that includes node attributes, edge attributes, and weight parameters.

[0011] Furthermore, it also includes: receiving flight path planning request messages, parsing the coordinates of the planning start and end points, flight time windows, and scenario constraints, converting the parsing results into planning parameter vectors, retrieving flight segment nodes in the knowledge graph based on the planning parameter vectors, and generating a candidate flight segment set; The matching coefficients between the segment feature model and the planning parameter vector in the candidate segment set are calculated. The matching coefficients are then normalized using the segment feature matching model to generate a segment score sequence. Valid segments are then selected according to the score threshold.

[0012] Furthermore, it also includes: prioritizing candidate flight segments based on their correlation strength values, constructing flight segment connection constraints, sequentially combining the sorted flight segments according to the connection constraints to generate multiple candidate route schemes, and calculating the comprehensive score for each candidate route; The candidate route with the best score is selected as the planning result. The waypoint sequence, altitude sequence, and speed sequence of the planning result are extracted to generate a standard format flight path planning document, which is then written into the planning decision database.

[0013] Secondly, this application provides an airspace planning and optimization decision-making device based on a historical flight knowledge base, comprising: The segment determination module is used to collect historical flight track data streams, construct an airspace planning meta-model containing segment boundary parameters and splitting rules, perform filtering preprocessing on the historical flight track data streams based on the airspace planning meta-model, determine segment switching points based on preset segmenting thresholds, calculate segmenting parameters to generate a segment feature model, record the associated scene features of the segment feature model, and establish a segment dataset. The feature matching module is used to construct a multi-dimensional association matrix, calculate the membership coefficient between the flight segment feature model and the associated scene features, generate a weight index based on the similarity distribution, combine the membership coefficient and the weight index to form an association strength value, establish a knowledge graph containing flight segment features, scene attributes, and association strength, and construct a flight segment feature matching model. The route planning module is used to receive flight route planning requests, extract planning constraint parameters, retrieve segment feature models with matching degrees exceeding a threshold based on the knowledge graph, calculate candidate segment scores using the segment feature matching models, sort and filter according to the association strength values, combine to generate a complete route, output flight route planning results, and write the flight route planning results into the planning decision database.

[0014] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the airspace planning and optimization decision-making method based on a historical flight knowledge base.

[0015] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the airspace planning and optimization decision-making method based on a historical flight knowledge base.

[0016] Fifthly, this application provides a computer program product, including a computer program / instruction that, when executed by a processor, implements the steps of the airspace planning and optimization decision-making method based on a historical flight knowledge base.

[0017] As can be seen from the above technical solution, this application provides an airspace planning and optimization decision-making method and apparatus based on a historical flight knowledge base. Through an innovative design of the trajectory processing system, and by using meta-models and feature extraction, it achieves effective data utilization. A knowledge graph mechanism is constructed, and a reliable matching model is established by combining association analysis and weight calculation. A scoring and screening process is introduced, and the rationality of the planning is ensured through candidate evaluation and path combination. This method effectively solves the shortcomings of traditional technologies in trajectory processing, knowledge modeling, and decision optimization, providing technical support for flight path planning. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the airspace planning and optimization decision-making method based on historical flight knowledge base in the embodiments of this application; Figure 2 This is a structural diagram of the airspace planning and optimization decision-making device based on historical flight knowledge base in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the electronic device in the embodiments of this application.

[0020] Figure label: Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver storage unit 9144, antenna 9111, speaker 9131, microphone 9132. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] The acquisition, storage, use, and processing of data in this application comply with relevant laws and regulations.

[0023] In view of the problems existing in the prior art, this application provides an airspace planning and optimization decision-making method and apparatus based on a historical flight knowledge base. Through an innovative design of the trajectory processing system, and by using meta-modeling and feature extraction, it achieves effective data utilization. A knowledge graph mechanism is constructed, and a reliable matching model is established by combining association analysis and weight calculation. A scoring and screening process is introduced, and the rationality of the planning is ensured through candidate evaluation and path combination. This method effectively solves the shortcomings of traditional technologies in trajectory processing, knowledge modeling, and decision optimization, providing technical support for flight path planning.

[0024] To effectively address the shortcomings of traditional technologies in flight path processing, knowledge modeling, and decision optimization, and to provide technical support for flight path planning, this application provides an embodiment of an airspace planning and optimization decision-making method based on a historical flight knowledge base. See [link to relevant documentation]. Figure 1 The airspace planning and optimization decision-making method based on historical flight knowledge base specifically includes the following: Step S101: Collect historical flight track data streams, construct an airspace planning meta-model containing flight segment boundary parameters and splitting rules, perform filtering preprocessing on the historical flight track data streams according to the airspace planning meta-model, determine flight segment switching points based on preset segmenting thresholds, calculate segmenting parameters to generate flight segment feature models, record the associated scene features of the flight segment feature models, and establish a flight segment dataset. This embodiment uses historical flight track data streams as input. The data sources cover four basic quantities of the positioning point: latitude and longitude, altitude, heading angle, and speed, and include a timestamp field.

[0025] To ensure comparability in subsequent segmentation, the data is first reconstructed into an equally spaced sequence under a unified time scale. Missing data points are filled in using neighborhood interpolation, and transition points are eliminated based on the physical continuity constraints of velocity and altitude. After forming a trajectory parameter sequence on the same time axis, filtering rules are generated based on the noise threshold and anomaly criteria recorded in the airspace planning meta-model. A forward prediction and residual backoff filtering process is then executed, outputting the smoothed trajectory and synchronized derived values, including mileage increment and climb rate.

[0026] The airspace planning meta-model is centered on flight segment boundary parameters and splitting rules. Boundary parameters include the range of location distribution, speed variation range, and altitude variation range. The splitting rules are composed of threshold conditions and persistence requirements. The model is bound to a scale version during loading to ensure compatibility between the threshold conditions and the data sampling rate.

[0027] Based on this model, a sliding window analysis is performed on the smoothed trajectory parameter sequence. Within the window, the average speed, speed direction change, sea height increment, and climb rate stability are statistically analyzed to form a window feature vector, which provides direct input for subsequent switching decisions.

[0028] The segment switching point determination adopts a multi-indicator consistent triggering strategy. To reduce the impact of a single outlier, a comparison of the rate of change between adjacent windows and a minimum number of sustained points constraint are set. When a speed level change transition or altitude change crosses the upper limit and the climb rate stability enters a new state range, a potential switching point is marked; if the criteria for the previous state are no longer met and the new state persists for a set length, a switch is confirmed. To provide an executable calculation method, this embodiment uses the climb rate as the dominant quantity in the profile dimension, and the climb rate is calculated using the following formula: r=(H_n-H_{nm}) / (D_n-D_{nm}), Where H is sea altitude, D is cumulative mileage, n is the current point index, and m is the configurable lag step size. The numerator reflects altitude change, and the denominator reflects horizontal advancement; both together define the ascent trend and avoid amplification of single-point noise.

[0029] After the switching point sequence is determined, the flight path is divided into several sub-segments. Feature parameters are calculated for each sub-segment: horizontally, average speed and turning rate are extracted for straight and turning segments; in the profile dimension, start altitude, end altitude, and average rate of climb are extracted, with terminal speed and spatial start and end positions added as needed. These quantities are organized into segment feature vectors using fixed fields, forming the core data block of the segment feature model. To ensure cross-task comparability, all parameters undergo unit normalization and value range pruning before being written; samples exceeding reasonable ranges are retained with low-confidence labels and are not included in the default statistics.

[0030] The associated scene features are recorded synchronously in the same feature extraction process, including five categories of discrete and continuous mixed quantities: aircraft type, takeoff and landing airport identifier, mission type, time window, and weather summary. To avoid scene noise interfering with flight segment features, scene quantities are not involved in segmentation determination, but are only attached to the flight segment feature model as foreign keys, and the event timestamp and source identifier are retained to facilitate subsequent membership calculation and weight learning.

[0031] The flight segment dataset is built accordingly, and the data organization adopts a dual-index structure of segment primary key and scene foreign key. The segment primary key consists of task identifier and switching point index, corresponding to a unique flight segment feature model; the scene foreign key points to the scene feature table, which is used to build a multi-dimensional association matrix during the knowledge graph construction stage. The dataset also retains the meta-model version and the filtering rule version to ensure consistent segmentation results during subsequent replays.

[0032] The above-mentioned products are transported downstream in two types of intermediate quantities.

[0033] One type is the structured segment feature model, which will participate in the calculation of difference and membership degree when constructing the correlation matrix; Another type is the original record of switching points and window statistics, which will be called back when parameter thresholds are updated or segmented errors are corrected, in order to verify the adaptability of boundary conditions.

[0034] By completing the closed loop of preprocessing, threshold condition determination and feature modeling in this step, subsequent steps can directly perform similarity measurement and candidate segment selection in a stable segment space, reducing the propagation of uncertainty across stages.

[0035] Step S102: Construct a multidimensional association matrix, calculate the membership coefficient between the flight segment feature model and the associated scene features, generate a weight index based on the similarity distribution, combine the membership coefficient and the weight index to form an association strength value, establish a knowledge graph containing flight segment features, scene attributes, and association strength, and construct a flight segment feature matching model. This embodiment performs relationship modeling based on an existing flight segment dataset. The input objects are the version-annotated flight segment feature model and its corresponding scene feature set.

[0036] To facilitate calculation and comparison, continuous quantities in the flight segment feature model are first normalized by field, while discrete quantities are converted into numerical representations using one-hot encoding. Scene features retain five fields: time window, aircraft type, departure and arrival airports, mission type, and meteorological summary. The time window is mapped to two values: intraday location and seasonal phase. The meteorological summary is broken down into three items: wind direction zone, wind speed level, and visibility level. All fields are homogenized to form comparable feature vectors on both sides, serving as the basis for subsequent matrix construction.

[0037] The multidimensional association matrix is ​​constructed with flight segments as rows and scenes as columns. To characterize the influence of different dimensions, a family of difference functions is first defined in the feature space. For continuous quantities, a weighted Mahalanobis distance is used; for discrete quantities, an indicator distance is used, and a parent-class consistency constraint is introduced within the same origin level. For time-series quantities, a phase difference metric is used, constrained to take the shortest arc in the periodic space. These difference values ​​are normalized and concatenated into a multidimensional feature vector. Then, the spatial distribution density is calculated to obtain the similarity distribution. Similarity is interpreted as the prior matching degree and used to generate the membership function and weight index.

[0038] Membership coefficients are characterized using a data-driven family of piecewise monotonic functions. To provide an executable approach, let the quantization value of the k-th behavior be X_k, and the weight parameter be W_k. The membership degree between a feature of a flight segment and a feature of a single scene is defined as: μ = [∑_k W_k·X_k] / [max_i ∑_k W_k·X_{k}^{(i)}], where the numerator represents the weighted response of the current combination, and the denominator is the maximum response in the same comparison group, used to normalize to between zero and one. X_k is derived from the combination value of similarity distribution and usage behavior, which includes the number of times the model is used and the duration of dwell time, both of which are converted on the same time basis. W_k is taken from the strategy configuration on the system side, satisfying the condition that the sum is one, avoiding single-dimensional dominance.

[0039] The weighting metric reflects the influence of different scenario dimensions on a specific flight segment category and is derived from two parts. One part is a measure of the stability of the similarity distribution, expressed as the inverse of the variance of that dimension in its neighborhood; the other part is the sensitivity of historical selection preferences to that dimension, expressed as the magnitude of candidate ranking changes caused by changes in that dimension. These two parts are linearly synthesized after scale alignment, outputting a dimensionalized weight vector, which serves as a modulation term for subsequent association strength. This setting reduces uncertain contributions in sparse sample dimensions and appropriately improves resolution in information-dense dimensions.

[0040] The association strength value is obtained by combining the membership coefficient and the weight index. The calculation method is to multiply each dimension and then sum them. The result still falls between zero and one, which is convenient for direct comparison with the threshold condition.

[0041] To suppress the influence of extreme values, a soft truncation is applied to the single-dimensional membership degree before synthesis, causing it to exhibit decreasing gain in the high-value region. The association strength matrix is ​​generated accordingly, with each element of the matrix corresponding to the association strength between a pair of segment features and scene features, which subsequently serves as the numerical source for the edge weights of the knowledge graph.

[0042] The nodes in the knowledge graph consist of two types of entities: segment feature nodes, which carry segment feature vectors and version tags; and scene attribute nodes, which carry scene feature vectors and time stamps. Edges are bidirectional, and edge attributes include association strength, participating dimension weights, construction time, and data source identifiers. To improve local connectivity during retrieval, weak connections are established between segments under the same task type. Edge weights are derived from the similarity of adjacent segments in the feature space, facilitating cross-segment supplementation when candidates are insufficient.

[0043] A segment feature matching model, named the Segment Matching Scorer, is constructed based on the aforementioned graph. The input is a planning parameter vector. First, the set of closest scene nodes is located on the scene side. Then, using these nodes as bridges, the adjacent segment feature nodes and their corresponding edge weights are aggregated. The scorer calculates a normalized score for candidate segments, with association strength as the primary score and a penalty term for dimensional weights. The output is a candidate set sorted in descending order of score and a score decomposition, used to explain the contribution dimensions. To ensure numerical stability, the scorer uses only the frozen weight configuration and the latest graph snapshot during the inference phase, avoiding online drift.

[0044] The product of this step performs two types of functions downstream.

[0045] First, it provides comparable scores and ranking criteria for the feature models of flight segments whose matching degree exceeds the threshold, directly supporting candidate screening and combination; Secondly, it provides traceable evidence for subsequent knowledge maintenance. The versioned recording of edge weights and weights allows the replay of past decision paths after strategy adjustments.

[0046] By establishing a series of constraints through the correlation matrix, membership degree, and weights, a closed loop is created from data discrepancies to interpretable scoring, ensuring that subsequently generated routes can be verified under the same criteria.

[0047] Step S103: Receive flight path planning request, extract planning constraint parameters, retrieve segment feature models with matching degree exceeding the threshold according to the knowledge graph, calculate candidate segment scores using the segment feature matching model, sort and filter according to the association strength value, combine to generate a complete route, output flight path planning results, and write the flight path planning results into the planning decision database.

[0048] This embodiment focuses on the planning task at the operational end. The input consists of a structured planning request message and the knowledge graph and segment matching scorer established in previous steps. The request message includes constraints such as origin and destination coordinates, desired time window, task type, aircraft type, and safety interval. The system performs semantic parsing on the message, unifying spatial coordinates to the same geodetic datum, mapping the time field to intraday position and seasonal phase values, and converting the task and aircraft type into discrete codes, merging them to form a planning parameter vector. To prevent abnormal parameters from interfering with retrieval, field-level integrity checks and boundary pruning are employed. Missing fields are substituted with the statistical median value of the most recent similar tasks and labeled with a low-confidence tag to avoid empty fields.

[0049] Nearest neighbor retrieval is performed on the scene side of the knowledge graph based on planning parameter vectors. The retrieval strategy employs multi-dimensional threshold condition superposition, with time phase, meteorological summary, and task type as hard constraints, and aircraft type and airport category as soft constraints. First, the set of scene nodes that satisfy the hard constraints is located, and then the neighborhood of the soft constraints is expanded from high to low similarity. The segment feature nodes connected to these scene nodes constitute a candidate segment pool, with each candidate carrying the association strength from the graph edges as well as its own feature vector and version number. If the candidate size is insufficient, it is allowed to supplement across adjacent scene nodes under the same task type, but without breaking the hard constraints.

[0050] In this embodiment, a matching score is calculated for each candidate flight segment. The flight segment matching scorer takes the planning parameter vector and the scenario and feature quantities of the candidate flight segment as input. The main scoring term is the association strength on the edge, and the penalty term characterizes the geometric inconsistency between the flight segment and the current start and end constraints, including the deviation of the connection direction, the distance between the end of the flight segment and the start of the next segment, and the poor continuity of the profile height. The penalty term is deducted from the main score after being weighted according to the dimension, forming a normalized score. The scorer outputs the score and decomposition terms of each candidate for source tracing. Low confidence labels trigger the upper limit of the score to avoid the dominance of abnormal data.

[0051] After candidate sorting is completed, the combination phase begins. Combination constructs a directed graph with the start and end points as anchors. Nodes represent the start and end positions of candidate flight segments, and edge weights represent the comprehensive cost. The comprehensive cost consists of two parts: a negative score from the scorer and a connection cost, including spatial transitions and temporal window deviations. To avoid the exponential complexity of global search, connection constraints are introduced, limiting the continuation of candidate flight segments to meet spatial adjacency thresholds and altitude band overlap thresholds; if these are exceeded, no edge is connected. A layer-by-layer, width-limited search is performed on this directed graph, with the width determined by the task type, resulting in several candidate routes that connect to the destination.

[0052] A comprehensive score is calculated from the candidate route set. The comprehensive score is a weighted sum of the matching scores of each segment on the route, with the weights proportional to the segment duration, and the accumulated connection costs are deducted. Routes that meet the safety separation and altitude constraints are sorted in descending order of score, and the route with the highest score is selected as the planning result. To output executable path information, waypoint sequences, altitude sequences, and speed sequences are extracted from the route. Waypoints are derived from the end points of each segment and interpolated at fixed arc length intervals within the turning segments. The altitude sequence is smoothly spliced ​​according to the profile parameters, and the speed sequence is constructed based on the average speed of the horizontal segment and the average linear speed of the turning segment. All three are resampled on a unified time axis to ensure that downstream systems can directly call them.

[0053] The generated flight path planning results are accompanied by source credentials, including the segment identifiers of the combined flight paths, the corresponding scenario nodes, and the edge weight version numbers, which facilitate future verification.

[0054] Before writing, a consistency check is performed to verify that the time window coverage is complete, the spatial boundaries do not cross known restricted areas, and the profile is continuous without reverse height jumps. After the check passes, the results are stored in the planning decision database in a standard format, with the primary key being the task identifier and generation time, and the value range including the three-sequence route, candidate set summary, and scoring details. The database retains snapshot numbers, corresponding to the map and scorer versions, to ensure consistent output can be reproduced under replay conditions with the same input.

[0055] To illustrate the feasibility of the aforementioned route, consider a starting point located north of an airport, an endpoint in adjacent operational airspace, and a time window around sunset. The system locates scene nodes with similar seasonal phases and weather conditions in the map, retrieving a candidate set including takeoff, turn-climb, and level cruise segments. The scorer assigns a higher primary score due to the consistency of time phase and mission type, while penalizing a sharp turn at the junction, ultimately combining to obtain two through routes. The route with the higher overall score has fewer turns and a continuous profile, and is selected as the planning result and written into the database for direct reference in subsequent approval and distribution stages.

[0056] As described above, the airspace planning and optimization decision-making method based on a historical flight knowledge base provided in this application can effectively utilize data through an innovative design of the trajectory processing system and through meta-modeling and feature extraction. It constructs a knowledge graph mechanism, combining association analysis and weight calculation to establish a reliable matching model. A scoring and screening process is introduced, using candidate evaluation and path combination to ensure the rationality of the planning. This method effectively addresses the shortcomings of traditional technologies in trajectory processing, knowledge modeling, and decision optimization, providing technical support for flight path planning.

[0057] In one embodiment of the airspace planning and optimization decision-making method based on historical flight knowledge base of this application, it may further include the following: Step S201: Access multi-source historical track data, extract track point position coordinates, heading angle, speed, and altitude parameters, perform time-series alignment on the data, construct a track parameter sequence, generate data filtering rules based on a preset noise threshold, and form a standardized track dataset; Step S202: Establish segment classification boundary conditions based on the flight track dataset, construct an airspace planning meta-model that includes position distribution, speed change, and altitude change, write the data filtering rules into the airspace planning meta-model, and generate a segment classification and preprocessing rule library.

[0058] This embodiment focuses on the input standardization and model storage preparation for historical data, using multi-source historical track data from ground surveillance systems and mission recording systems. The original records contain six fields: timestamp, longitude, latitude, sea height, heading angle, and speed. The sampling periods are inconsistent, and packet loss occurs. To standardize the timescale, a target timeline is first constructed based on whole seconds. Time alignment is then performed on each track: when adjacent sampling intervals are less than twice the target period, linear interpolation is used to fill in the gaps; when the interval exceeds the threshold and the speed and heading within the segment are unstable, the segment is marked as a discontinuous area and will not be included in subsequent continuous statistics.

[0059] After the coordinates are transformed to the same geodetic datum, the cumulative mileage, horizontal distance increment and heading change are calculated in the order of the trajectory to form a sequence of trajectory parameters. The sequence elements include four principal quantities: position, velocity, heading and altitude, and their first-order differences.

[0060] Noise characteristics vary significantly across different data sources, and directly applying a uniform filtering strength can easily lead to loss of detail. Therefore, data filtering rules are generated based on preset noise thresholds. These rules consist of three sets of criteria: velocity mutation threshold, heading mutation threshold, and altitude spike threshold, corresponding to the boundaries of the first-order difference in velocity, the circumferential difference in heading, and the first-order difference in altitude, respectively; these are further combined with short-window variance as a stability constraint.

[0061] For each track, the empirical quantiles of these three differences are estimated within a short window. If the difference exceeds the corresponding threshold, it is treated as a bad point and replaced with a neighborhood fit. If the variance of the short window continues to exceed the limit, the segment is marked as a disturbance region and the confidence level is reduced. The filtering rules are saved in parameterized form, including the threshold set and window length, for reuse in subsequent model loading, thus forming a standardized track dataset.

[0062] After obtaining a high-quality flight track dataset, boundary conditions for flight segment classification are established based on flight motion characteristics. The classification targets cover five categories: takeoff segment, takeoff route segment, airspace segment, landing route segment, and landing segment. The boundary conditions employ a multi-quantity collaborative representation: location distribution constraints are used to limit the geographical range of takeoff and landing related segments; speed change patterns are used to identify acceleration, deceleration, and constant speed intervals; and altitude change patterns are used to distinguish between climb, level flight, and descent. Taking the profile dimension as an example, the climb rate *r* is defined as the increment of altitude relative to mileage. r=(H_n-H_{nm}) / (D_n-D_{nm}), Where H is sea height, D is cumulative mileage, n is the current point index, and m is the lag step size. By combining the sign and stability of r, as well as the continuity of heading changes, the state within the window can be determined. Position distribution constraints are expressed through airport reference points and their buffer zones; the buffer zone radius is related to stage attributes and is not a fixed constant.

[0063] The aforementioned boundary conditions and filtering rules are assembled into a spatial planning meta-model. This meta-model consists of three parts: a parameter dictionary to record the velocity, altitude, and position ranges for each stage; a decision flowchart to constrain the segmentation order and switching prerequisites, ensuring that the state machine does not make illegal jumps; and a set of execution operators to call specific differencing, smoothing, and windowing statistical methods. To ensure consistency across different sampling rates, the meta-model reads the temporal resolution of the dataset during loading and performs scale adaptation, scaling the lag step size and window length proportionally, and comparing thresholds within the normalization domain.

[0064] After the metamodel is solidified, the aforementioned data filtering rules are written into the preprocessing unit to form a segment classification and preprocessing rule base. This rule base is managed at the version level, with each version containing threshold configurations, window parameters, state machine transition matrices, and position buffer templates, supporting version-by-version playback when new data is accessed. The rule base provides two types of functions: first, it performs consistent preprocessing and segmentation on any historical track to obtain a reproducible segmentation sequence; second, it outputs the boundary hit evidence used during segmentation, including the triggered threshold conditions and window statistics, facilitating the subsequent use of the original basis when calculating membership degrees on the knowledge graph side.

[0065] The standardized track dataset and segment classification and preprocessing rule base generated in this embodiment are directly called downstream. The former serves as input for calculating the segment feature model, providing a stable parameter sequence and confidence annotations; the latter provides standardized boundary conditions during the segmentation and verification stages, ensuring that the segmentation results of different batches of data are comparable. Through the integrated organization of data alignment, noise suppression, and boundary modeling, an executable foundation is laid for subsequent multidimensional correlation matrix construction and matching scoring.

[0066] In one embodiment of the airspace planning and optimization decision-making method based on historical flight knowledge base of this application, it may further include the following: Step S301: Set a segment switching judgment threshold based on the airspace planning meta-model, perform sliding window analysis on the trajectory data, calculate the parameter change rate between adjacent windows, mark the segment switching point according to the comparison result of the parameter change rate and the judgment threshold, and generate a segment segmentation marking sequence. Step S302: Divide the track data into multiple sub-segments using the segment segmentation marker sequence, calculate the feature parameters of each sub-segment, construct the segment feature vector, extract the scene attributes associated with the segment, and establish the segment feature model data structure.

[0067] This embodiment performs segmentation and feature modeling based on the standardized track dataset and segment classification and preprocessing rule base obtained previously. The input consists of four principal quantities under a unified time scale: latitude and longitude, sea height, speed, and heading, along with their first-order differences and confidence labels.

[0068] To avoid scale dependence, all quantities are transferred to the normalization domain before entering the decision-making process, and the window length and lag step are mapped to the current sampling rate according to the rule base version.

[0069] Segment switching decisions employ a sliding window analysis, statistically analyzing the mean speed, heading phase change, sea height increment, climb rate stability, and short-window variance within each window. The rate of change of parameters between adjacent windows is calculated and compared with the decision thresholds set for each stage in the airspace planning meta-model. Thresholds are categorized into entry and exit thresholds, forming a hysteresis-based state machine to reduce boundary jitter. The dominant quantity in the profile dimension is the climb rate, calculated using the following method: r=(H_n-H_{nm}) / (D_n-D_{nm}), Where H is sea altitude, D is cumulative mileage, n is the current point index, and m is the lag step size. The sign and variance of r characterize the candidate states of climb, level flight, and descent; meanwhile, heading phase change and velocity gradient are used to identify straight sections and turning sections, as well as acceleration and deceleration processes. A potential switching point is marked only when at least two types of indicators simultaneously exceed the corresponding entry threshold and the exit threshold corresponding to the previous state has been met; a potential switching point must be maintained within a minimum number of duration points to be confirmed as a valid switch and written into the segment segmentation marker sequence.

[0070] To improve usability, the segmentation marker sequence undergoes a consistency check after generation. First, it checks if the distance between adjacent switching points is too short; if it's shorter than the minimum segment length, it's merged into a state with a higher score. Second, low-confidence segments are rolled back according to their parent state to prevent erroneous segmentation from spreading to subsequent feature statistics. The validated marker sequence is aligned with the original timeline, forming a boundary set that can be directly used for slicing.

[0071] The flight path data was thus divided into multiple sub-segments. Each sub-segment had its characteristic parameters calculated in both horizontal and profile dimensions: the horizontal dimension outputs average horizontal velocity and azimuth stability for straight segments, and average linear velocity, average rate of turn, and turning radius for turning segments; the profile dimension outputs start altitude, end altitude, average rate of climb, and level flight length. To eliminate endpoint effects, the ends of the sub-segments were corrected for boundary length based on the window length, and endpoint values ​​were obtained through interpolation. Segments with discontinuous markers were not included in the average-type statistics; only interval summaries and warning markers were retained for later filtering.

[0072] Feature parameters are written into the segment feature vector in a fixed field order. The fields include six sets of quantities: spatial origin and end location, duration, mileage, speed statistics, turning statistics, and profile statistics. Each set of quantities is accompanied by a calculation method and version label. Scene attributes of the same sub-segment are extracted simultaneously at this stage, including aircraft type, take-off and landing airport identification, mission type, time window, and meteorological summary.

[0073] Scene attributes are not involved in numerical aggregation but are linked by reference to maintain a one-to-one correspondence with feature vectors. For sub-segments that cross airport buffer zones or meteorological zone boundaries, they are split into two records according to the coverage ratio, with the ratio information recorded in an extended field to avoid ambiguity during subsequent association.

[0074] At the data structure level, a segment feature model is established. The primary key consists of the task identifier and the sub-segment number, and the value range includes the segment feature vector and scene attribute references. The model internally retains two types of auxiliary indexes: one organized by stage type for easy horizontal comparison within the same type of segment; and the other organized by time window for easy seasonal phase retrieval and matching. Each record carries a confidence label from the source track and a preprocessing version number to ensure consistency in subsequent playback.

[0075] The aforementioned segment feature model serves as the direct input for constructing the downstream multidimensional association matrix. Feature vectors are used for difference calculation, and scene attributes are used for the behavioral quantification branch of the membership function. The segmentation label sequence provides a replay basis during policy updates and reviews. Through threshold-based hysteresis state machines and collaborative determination using two-dimensional parameters, the segmentation results maintain sensitivity while possessing sufficient stability, providing a reliable intermediate layer for subsequent matching scoring and route combination.

[0076] In one embodiment of the airspace planning and optimization decision-making method based on historical flight knowledge base of this application, it may further include the following: Step S401: Construct an association matrix for the flight segment feature model according to spatial attributes, temporal attributes, and scene attributes; calculate the data distribution difference value between each flight segment feature and the associated scene feature; generate a multi-dimensional feature vector based on the difference value; and construct a feature mapping space. Step S402: Normalize the feature mapping space, calculate the spatial distribution density of the multidimensional feature vector, generate a membership calculation function based on the density distribution, and use the membership calculation function to generate the correlation coefficient matrix between the flight segment features and the scene features.

[0077] This embodiment builds upon the previously established flight segment feature model, constructing a correlation matrix around three categories of quantities: spatial attributes, temporal attributes, and scene attributes. On the input side, spatial attributes are derived from the spatial start and end positions, mileage, azimuth stability, and profile statistics in the flight segment feature vector; temporal attributes are obtained by mapping the flight segment start and end times to intraday position and seasonal phase values; scene attributes are encoded representations of aircraft type, arrival and departure airport identifiers, mission type, and meteorological summary. To avoid computational bias caused by different units, continuous quantities are first range-scaled, and discrete quantities are converted into strict one-hot encoding or hierarchical parent-class encoding, ensuring that subsequent differences can be compared under a unified standard.

[0078] The correlation matrix uses flight segments as rows and scenarios as columns. Matrix elements need to characterize the degree of fit between the flight segment and the scenario across various attribute dimensions. To this end, difference values ​​are defined dimension by dimension: the spatial dimension uses weighted Mahalanobis distance, with weights derived from the robust inverse variance of each component within the feature vector; the temporal dimension calculates the phase difference within the ring space, taking the shortest arc length and measuring it separately for day / night and seasonal cycles; the discrete difference for the scenario dimension is primarily indicated by zeros and ones, while a parent-class consistency rule is introduced to impose a smaller penalty on near-class components at the aircraft type or task level. The differences in each dimension are scaled to between zero and one, concatenated in a fixed order to form a corresponding multi-dimensional feature vector, serving as the original representation of that row-column pair.

[0079] The feature mapping space is defined by the set of multidimensional feature vectors of all row and column pairs. This space preserves the metric attributes and initial weights of each dimension, and stores the empirical distribution function of each dimension for subsequent estimation of density and membership.

[0080] To enhance the distinguishability of sparse regions, highly correlated subquantities are orthogonalized in the mapping space to reduce the interference of redundant dimensions on density estimation. For a small number of missing subquantities, the mean is not filled in, but the distribution quantile anchor points of that dimension are used instead, and a visibility mask is marked on the vector to avoid incorrectly including the filled values ​​in the density peak.

[0081] After the feature mapping space is established, normalization is required to ensure that the influence of each dimension on the density estimation is on the same scale. Specifically, a robust scaling estimate is used to linearly scale each dimension so that the median falls at zero and the interquartile range is mapped to a uniform range. At the same time, angular continuity is preserved for the ring space dimensions to prevent jumps introduced by boundary breaks.

[0082] The normalized vector set is then subjected to kernel density estimation. The kernel function is a Gaussian kernel with adaptive bandwidth in each dimension. The bandwidth is given as a function of the sample size and dimension to avoid over-smoothing in high dimensions. The output of the density estimation serves as an indicator of the probability of the vector appearing in the mapping space, reflecting the degree of concentration of similar combinations.

[0083] A membership function is constructed based on the density results. The intuitive meaning of membership is "the suitability of this flight segment in a given scenario," and its value should monotonically decrease as the difference increases and increase as the local density increases. This embodiment uses a piecewise monotonic function to map the difference vector and local density to values ​​between zero and one. To provide an executable approach, the core term of membership can be set as the ratio of the weighted response to the maximum response in the same group, in the form: μ=[∑_k W_k·X_k] / [max_i ∑_k W_k·X_k^{(i)}], Where X_k is the fusion index of the k-th behavior and similarity, W_k is the weight of this index, and the sum is one. The fusion index X_k is linearly synthesized from two parts: one part comes from the inverse function of the difference, and the other part comes from the kernel density value of the point. The missing dimension is ignored according to the visibility mask to ensure that the function still has a stable output in the sparse region of the sample.

[0084] Applying the membership function to all combinations of flight segments and scenarios yields a correlation coefficient matrix. The matrix elements are membership values, accompanied by two types of metadata: first, dimensional contribution decomposition, recording the positive and negative contributions of each dimension to the value; second, version identifiers for bandwidth and weights, used for replay and tracing. At the numerical level, soft truncation is applied to abnormally high and low values ​​to avoid disproportionately influencing subsequent rankings from extreme samples; at the structural level, row and column blocks of similar scenarios within the same task type are smoothed to reduce accidental disturbances.

[0085] The final correlation coefficient matrix is ​​input into the downstream segment feature matching model, serving as the direct source of graph edge weights. During the retrieval phase, the downstream model reads the scene node column corresponding to a specific planning parameter vector, filters candidate segment rows according to membership thresholds, and interprets the reasons for the candidates based on dimensional contribution information.

[0086] Since this step provides a continuous mapping from multidimensional differences to membership, the same criteria can be maintained in the subsequent scoring and combination stages, reducing the uncertainty caused by strategy switching.

[0087] In one embodiment of the airspace planning and optimization decision-making method based on historical flight knowledge base of this application, it may further include the following: Step S501: Construct a combination function of membership coefficient and weight index, calculate the correlation strength value for each pair of segment features and scene features, generate a correlation strength matrix, and establish a mapping relationship table between segment features and scene features; Step S502: Construct a knowledge graph node set based on the mapping relationship table, write the association strength matrix into the connection weights between nodes, and generate a flight segment feature matching model containing node attributes, edge attributes, and weight parameters.

[0088] This embodiment takes the obtained correlation coefficient matrix and dimensional weight vector, and generates a correlation strength that can be used for retrieval and combination, while maintaining consistent numerical caliber. The input includes two parts: The first is a correlation coefficient matrix organized by flight segment rows and scene columns, with elements representing membership degrees; Second, the weight index is jointly estimated by similarity distribution and historical selection sensitivity, and stored as components in spatial, temporal and scene dimensions.

[0089] To avoid amplification of a single dimension, the weight indicators are constrained to sum to one through robust standardization during loading, and soft truncation is implemented for components that exceed the preset upper bound.

[0090] The association strength is calculated using an interpretable combination function. Considering that membership already reflects the matching basis, weights are used to modulate the contributions of different dimensions. The combination caliber is set to summation after multiplication of each dimension, and a compression function with diminishing returns is applied to high-value regions to suppress extreme samples.

[0091] The association strength between flight segment i and scene j is denoted as S_{ij}, and its calculation process is as follows: First, the membership subquantities of this row and column pair in the three dimensions are extracted into vectors μ_{ij}, which are then multiplied element-wise with the homomorphic weight vector w to obtain the response vector r_{ij}. Then, r_{ij} is summed and monotonically compressed and mapped to the interval between zero and one. To ensure boundary consistency, when some dimension labels are missing, a visibility mask is used to skip that dimension during summation, and the data is re-normalized according to the effective dimension to avoid unfairly penalizing missing samples.

[0092] The association strength matrix is ​​constructed using S_{ij} for all row and column pairs. After matrix generation, two stabilization processes are performed: first, quantile scaling within rows and columns to weaken extreme tails; second, mild smoothing is applied to blocks of the same task type and approximate time phase to reduce ranking turbulence caused by occasional disturbances. A mapping table is then established, with entries using segment identifiers and scene identifiers as composite primary keys, recording association strength, dimension weight snapshots, membership decomposition, computation time, and rule version number. This table provides replayable evidence for downstream atlas assembly and online scorers, allowing any decision to be traced back to the source of the weights and memberships at that time.

[0093] A knowledge graph is constructed based on a mapping table. The node set consists of two types of nodes: flight segment feature nodes and scene attribute nodes. The former carries the flight segment feature vector, stage type, duration, mileage, and profile summary; the latter carries the aircraft type, departure and arrival airports, mission type, time phase, and meteorological summary. To enhance the usability of the structure, weak adjacency relationships are established between nodes of the same type on the scene side, with edge values ​​based on similarity; adjacency relationships are established between nodes of the same stage on the flight segment side, with edge values ​​based on feature space similarity. These weak connections do not participate in the main scoring, but only provide expansion paths when there are insufficient candidates.

[0094] The association strength matrix is ​​written into the weight attributes of the edges between nodes, forming a weighted bidirectional graph. Each edge from a flight segment node to a scene node contains three types of attributes: first, the edge weight, S_{ij}; second, contribution decomposition, recording the positive and negative contribution ratios of each dimension to the edge weight; and third, version information, including the weight configuration and kernel density bandwidth number. To ensure temporal consistency in retrieval, the graph is published in snapshot format, with a new snapshot number generated for each update and frozen until the end of the scoring period.

[0095] Based on this, a segment feature matching model is generated. This model uses a graph as its foundation, defining a scoring entry point and merging logic. The input is a planning parameter vector, which, after scene encoding, locates a set of scene nodes in the graph. The set of segment nodes directly adjacent to these nodes is then taken, and edge weights are used as the principal score. Contribution decomposition is used for interpretation. The model internally includes connection constraint checks, eliminating candidates that do not meet spatial adjacency and profile continuity requirements; for segments with low-confidence labels, their edge weights are proportionally reduced. The output includes candidate segments, their matching scores, contribution details, and the version number of the source edges, which can be directly used in subsequent route generation.

[0096] This matching model, together with the graph, constitutes an executable retrieval and scoring environment. The association strength generated upstream is transformed into queryable edge weights. Downstream, when combining paths, only the edge weights and connection costs need to be summed and sorted, without needing to repeatedly calculate membership degrees or adjust weights. Through versioned management of the mapping table and graph snapshots, the system can obtain consistent scoring results when replaying historical plans or comparing strategies, ensuring the interpretability and stability of the planning decision-making chain.

[0097] In one embodiment of the airspace planning and optimization decision-making method based on historical flight knowledge base of this application, it may further include the following: Step S601: Access the flight path planning request message, parse the planning start and end point coordinates, flight time window, and scenario constraints, convert the parsing results into a planning parameter vector, and retrieve the flight segment nodes in the knowledge graph based on the planning parameter vector to generate a candidate flight segment set; Step S602: Calculate the matching coefficient between the segment feature model and the planning parameter vector in the candidate segment set, normalize the matching coefficient using the segment feature matching model, generate a segment scoring sequence, and filter valid segments according to the scoring threshold.

[0098] This embodiment focuses on request processing and candidate filtering at the runtime level. The input consists of a planning request message, a frozen knowledge graph snapshot, and a flight segment feature matching model. The message includes fields such as origin coordinates, destination coordinates, flight time window, mission type, aircraft type, and safety interval. The system first corrects the coordinates under a unified geodetic datum, decomposing the time window into two numerical expressions: intraday position and seasonal phase. Mission type and aircraft type are encoded as discrete quantities, and if meteorological conditions exist, they are mapped to three items: wind direction zone, wind speed level, and visibility level. The above results are assembled into a planning parameter vector in a fixed field order. When a field is missing, it is replaced by the median value of a similar mission and labeled as low confidence; this value does not participate in subsequent hard constraint determination.

[0099] Retrieval is performed on the scene side of the knowledge graph based on the planning parameter vector. The retrieval strategy consists of two layers: a hard constraint layer requires that the time phase and task type be consistent, and that airport category and restricted area constraints be satisfied; a soft constraint layer expands the model type and meteorological data based on nearest neighbors. The system selects nodes that satisfy the hard constraints from the scene node set as the core set, and then expands them from high to low according to the soft constraint similarity to a preset upper limit, merging them to obtain the scene node set used for recall. Using this set as a bridge, the system reads its first-order adjacent flight segment nodes in the graph to form a candidate flight segment set; each candidate carries edge weights, flight segment feature vectors, and version identifiers for subsequent scoring.

[0100] Matching coefficients are calculated between candidate flight segments and the planning parameter vectors. The calculation is performed in two parts: one part is derived from the map edge weights, reflecting the prior matching of historical membership and weight modulation; the other part is a geometric consistency term, measuring the connectivity between the start and end positions of the flight segment and the planned start and end points, the azimuth difference, and the overlap of profile altitude bands. Each component of the geometric term has a clear definition: connectivity is measured by the distance from the end of the candidate flight segment to the starting point of the next segment of the target; azimuth difference is measured by the angle between the direction vectors of the horizontal track; and altitude overlap is measured by the intersection-union ratio of the two altitude intervals. These three terms, along with the edge weights, constitute the original matching coefficients under a unified scale.

[0101] The flight segment feature matching model normalizes the original matching coefficients and outputs comparable flight segment scores. The normalization process includes two steps: first, quantile stretching of the original coefficients based on the distribution within the candidate set to reduce the impact of extreme values; then, linear synthesis by dimension weights, and upper limit constraints are applied to candidates with low-confidence labels to prevent data quality issues from causing score anomalies. The model also performs connectivity constraint checks, eliminating candidates that fall outside the spatial adjacency threshold and altitude continuity threshold; and downgrading candidates near the boundaries of restricted areas based on risk coefficients.

[0102] The resulting flight segment score sequences are sorted in descending order and compared with a score threshold to filter valid flight segments. The score threshold is not a fixed constant, but is automatically determined by the historical distribution corresponding to the current mission type and time phase, ensuring a stable and effective set in both dense and sparse candidate scenarios. The filtering results retain the version numbers of the scores, contribution decomposition, and source edges, serving as direct input for subsequent combination stages.

[0103] To ensure seamless integration between upstream and downstream segments, a connection cursor is attached when the selected valid segments are written to the intermediate buffer. This cursor records the alignment position of the spatial endpoints of the candidate segments with the time window. During the assembly phase, this information is used to quickly check the feasibility of connecting segments, reducing redundant calculations.

[0104] If the candidate pool is insufficient at this stage, the system only expands the nearest neighbor radius within the soft constraints, without relaxing the hard constraints, thus ensuring that subsequently generated routes still meet mission and control requirements. The entire process uses knowledge graph snapshots and matching model versions as guidelines to ensure consistent candidate and scoring outputs when the same request is replayed.

[0105] In one embodiment of the airspace planning and optimization decision-making method based on historical flight knowledge base of this application, it may further include the following: Step S701: Prioritize the candidate segments according to the association strength value, construct segment connection constraints, combine the sorted segments in order according to the connection constraints, generate multiple candidate route schemes, and calculate the comprehensive score of each candidate route. Step S702: Select the candidate route with the best score as the planning result, extract the waypoint sequence, altitude sequence, and speed sequence of the planning result, generate a standard format flight path planning document, and write the flight path planning document into the planning decision database.

[0106] In this embodiment, after obtaining the candidate segment set and its association strength values, the process proceeds to the sorting and combination stage. Candidate segments are assigned segments with segment identifiers as keys, carrying edge weights, geometric consistency terms, and version tags. Priority is based primarily on association strength, followed by geometric consistency. To reduce the impact of extreme values ​​on the sorting, the association strength is first stretched within the current set, and then linearly synthesized with fixed weights and geometric consistency terms to obtain the sorting score.

[0107] After sorting, instead of directly concatenating globally, a set of connection constraints is constructed. These constraints include four categories: spatial adjacency threshold, time window continuity threshold, profile height overlap threshold, and upper limit of turning angle. Each threshold is derived from a versioned configuration of the preceding rule base to ensure consistency with the data caliber.

[0108] The combinatorial algorithm uses the origin and destination as anchors to construct a directed layered graph. The graph's hierarchy is divided according to task phases; only mutually exclusive candidates are retained within the same layer, and edges are connected across layers only if connectivity constraints are satisfied. The cost of each edge consists of two parts: a negative ranking score and a connection cost. The connection cost is further subdivided into three components: spatial endpoint distance, heading angle, and altitude band misalignment, which are summed after being normalized to a uniform scale. To avoid combinatorial explosion, a width-limited search strategy is introduced. The width depends on the task type, prioritizing the expansion of branches with higher ranking scores and lower connection costs. During expansion, branches are pruned when the cumulative cost exceeds a dynamic upper bound, which is given by the comprehensive score of the found feasible routes.

[0109] After generating candidate route options, a comprehensive score is calculated. The comprehensive score is a time-weighted sum of segment matching scores, minus the sum of connecting costs along the route. Two types of penalty constraints are introduced: one from safety intervals, which is deducted based on a risk coefficient if the route crosses a historically high-density area; the other from control boundaries, which is linearly penalized if the distance to the boundary of a restricted area is below a threshold. The scoring criteria remain consistent within a single route, and all parameters use the same version of configuration to ensure comparability.

[0110] When selecting the candidate route with the best score as the planning result, a consistency review is required. The review includes three checks: the route is connected from start to finish without self-intersections; the profile is continuous with no reverse altitude jumps; and the cumulative deviation within the time window is within acceptable limits. If the review fails, the route reverts to the suboptimal route, and the local threshold for connection costs is relaxed accordingly, but the hard boundary conditions are not relaxed to ensure the result meets regulatory requirements. If the review passes, trajectory calculation and sequence generation proceed.

[0111] Waypoint sequences are extracted from combined flight segments. For straight segments, endpoints are directly taken; for turning segments, interpolation is performed at fixed arc length intervals along the curve based on the average turning rate and turning radius to ensure uniform navigation density. Altitude sequences are extrapolated linearly or piecewise linearly within segments based on profile statistics, with short windows used to smooth out splicing peaks between segments. Velocity sequences are taken as average level velocity in horizontal segments and average linear velocity in turning segments; gradual constraints are applied at transition points between segments when necessary to avoid velocity abrupt changes. All three sequences are resampled on a unified time axis, with time steps compatible with the execution system, forming a set of paths that can be directly deployed.

[0112] When generating standard-format flight path planning documents, in addition to the waypoint, altitude, and speed sequences, the document also includes the mission identifier, generation time, map snapshot number used, and matching model version number. It also includes a candidate set summary and evidence entries for key connections, including triggered connection constraints and corresponding metrics. A declaration field for zero constraint violations is retained within the document for rapid downstream verification. To ensure traceability, each flight segment identifier and its source edge weight are appended, facilitating the reproduction of scoring and combined paths during playback.

[0113] Before the document is written to the planning decision database, a consistency check is performed. Check items include complete spatiotemporal coverage, non-intersecting restricted areas, speed and altitude variations within reasonable thresholds, and constant path sampling intervals. Upon passing these checks, a master input database is created based on the task identifier and generation time. The value range includes three sequences, a comprehensive score, detailed connection costs, and a version fingerprint. An immutable snapshot number is generated on the database side, supporting consistent output when replaying the same request. This result simultaneously triggers downstream approval processes, providing stable input for resource allocation and approval document generation.

[0114] To effectively address the shortcomings of traditional technologies in flight path processing, knowledge modeling, and decision optimization, and to provide technical support for flight path planning, this application provides an embodiment of an airspace planning and optimization decision-making device based on a historical flight knowledge base, for implementing all or part of the aforementioned airspace planning and optimization decision-making method. See [link to embodiment]. Figure 2 The airspace planning and optimization decision-making device based on historical flight knowledge base specifically includes the following components: The segment determination module 10 is used to collect historical flight track data streams, construct an airspace planning meta-model containing flight segment boundary parameters and splitting rules, perform filtering preprocessing on the historical flight track data streams according to the airspace planning meta-model, determine flight segment switching points based on preset segmenting thresholds, calculate segmenting parameters to generate a flight segment feature model, record the associated scene features of the flight segment feature model, and establish a flight segment dataset. The feature matching module 20 is used to construct a multi-dimensional association matrix, calculate the membership coefficient between the flight segment feature model and the associated scene features, generate a weight index based on the similarity distribution, combine the membership coefficient and the weight index to form an association strength value, establish a knowledge graph containing flight segment features, scene attributes and association strength, and construct a flight segment feature matching model. The route planning module 30 is used to receive flight route planning requests, extract planning constraint parameters, retrieve segment feature models with matching degrees exceeding a threshold based on the knowledge graph, calculate candidate segment scores using the segment feature matching model, sort and filter according to the association strength value, combine to generate a complete route, output flight route planning results, and write the flight route planning results into the planning decision database.

[0115] As described above, the airspace planning and optimization decision-making device based on a historical flight knowledge base provided in this application can effectively utilize data through an innovatively designed trajectory processing system, meta-modeling, and feature extraction. It constructs a knowledge graph mechanism, combining association analysis and weight calculation to establish a reliable matching model. A scoring and screening process is introduced, using candidate evaluation and path combination to ensure the rationality of the planning. This method effectively addresses the shortcomings of traditional technologies in trajectory processing, knowledge modeling, and decision optimization, providing technical support for flight path planning.

[0116] From a hardware perspective, in order to effectively address the shortcomings of traditional technologies in areas such as trajectory processing, knowledge modeling, and decision optimization, and to provide technical support for flight path planning, this application provides an embodiment of an electronic device for implementing all or part of the aforementioned airspace planning and optimization decision-making method based on a historical flight knowledge base. The electronic device specifically includes the following components: The system comprises a processor, memory, a communication interface, and a bus; wherein the processor, memory, and communication interface communicate with each other via the bus; the communication interface is used to realize information transmission between the airspace planning and optimization decision-making device based on historical flight knowledge base and core business systems, user terminals, and related databases and other related equipment; the logic controller can be a desktop computer, tablet computer, or mobile terminal, etc., and this embodiment is not limited to these. In this embodiment, the logic controller can be implemented with reference to the embodiments of the airspace planning and optimization decision-making method based on historical flight knowledge base and the embodiments of the airspace planning and optimization decision-making device based on historical flight knowledge base in the embodiments, the contents of which are incorporated herein, and repeated details will not be described again.

[0117] It is understood that the user terminal may include smartphones, tablet computers, network set-top boxes, portable computers, desktop computers, personal digital assistants (PDAs), in-vehicle devices, smart wearable devices, etc. Among these, the smart wearable devices may include smart glasses, smartwatches, smart bracelets, etc.

[0118] In practical applications, the airspace planning and optimization decision-making method based on historical flight knowledge base can be partially executed on the electronic device side as described above, or all operations can be completed in the client device. The choice can be made based on the processing power of the client device and the limitations of the user's usage scenario. This application does not impose any limitations on this. If all operations are completed in the client device, the client device may further include a processor.

[0119] The aforementioned client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side; in other implementation scenarios, it may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, a server cluster consisting of multiple servers, or a distributed server structure.

[0120] Figure 3 This is a schematic block diagram illustrating the system configuration of the electronic device 9600 according to an embodiment of this application. Figure 3 As shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It is worth noting that... Figure 3 This is an example; other types of structures can also be used to supplement or replace this structure to achieve telecommunications functions or other functions.

[0121] In one embodiment, the airspace planning and optimization decision-making method based on a historical flight knowledge base can be integrated into the central processing unit 9100. The central processing unit 9100 can be configured to perform the following control: Step S101: Collect historical flight track data streams, construct an airspace planning meta-model containing flight segment boundary parameters and splitting rules, perform filtering preprocessing on the historical flight track data streams according to the airspace planning meta-model, determine flight segment switching points based on preset segmenting thresholds, calculate segmenting parameters to generate flight segment feature models, record the associated scene features of the flight segment feature models, and establish a flight segment dataset. Step S102: Construct a multidimensional association matrix, calculate the membership coefficient between the flight segment feature model and the associated scene features, generate a weight index based on the similarity distribution, combine the membership coefficient and the weight index to form an association strength value, establish a knowledge graph containing flight segment features, scene attributes, and association strength, and construct a flight segment feature matching model. Step S103: Receive flight path planning request, extract planning constraint parameters, retrieve segment feature models with matching degree exceeding the threshold according to the knowledge graph, calculate candidate segment scores using the segment feature matching model, sort and filter according to the association strength value, combine to generate a complete route, output flight path planning results, and write the flight path planning results into the planning decision database.

[0122] As described above, the electronic device provided in this application, through an innovative design of a trajectory processing system, achieves effective data utilization through meta-modeling and feature extraction. It constructs a knowledge graph mechanism, combining association analysis and weight calculation to establish a reliable matching model. A scoring and screening process is introduced, ensuring the rationality of the planning through candidate evaluation and path combination. This method effectively addresses the shortcomings of traditional technologies in trajectory processing, knowledge modeling, and decision optimization, providing technical support for flight path planning.

[0123] In another embodiment, the airspace planning and optimization decision-making device based on historical flight knowledge base can be configured separately from the central processing unit 9100. For example, the airspace planning and optimization decision-making device based on historical flight knowledge base can be configured as a chip connected to the central processing unit 9100, and the airspace planning and optimization decision-making method function based on historical flight knowledge base can be realized through the control of the central processing unit.

[0124] like Figure 3 As shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily need to include these components. Figure 3All components shown; in addition, the electronic device 9600 may also include Figure 3 For components not shown, please refer to existing technologies.

[0125] like Figure 3 As shown, the central processing unit 9100, sometimes also referred to as a controller or operating control, may include a microprocessor or other processor device and / or logic device, which receives inputs and controls the operation of various components of the electronic device 9600.

[0126] The memory 9140 may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned failure-related information, and also store a program for executing that information. The central processing unit 9100 may execute the program stored in the memory 9140 to perform information storage or processing, etc.

[0127] Input unit 9120 provides input to central processing unit 9100. Input unit 9120 may be, for example, a keypad or touch input device. Power supply 9170 provides power to electronic device 9600. Display 9160 displays images and text. Display may be, for example, an LCD display, but is not limited thereto.

[0128] The memory 9140 can be a solid-state memory, such as a read-only memory (ROM), random access memory (RAM), a SIM card, etc. It can also be a memory that retains information even when power is off, can be selectively erased, and contains more data; examples of this type of memory are sometimes referred to as EPROMs. The memory 9140 can also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142 for storing application programs and function programs or processes for executing the operation of the electronic device 9600 via the central processing unit 9100.

[0129] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for the electronic device for communication functions and / or for performing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0130] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processing unit 9100 to provide input signals and receive output signals, which is the same as in a conventional mobile communication terminal.

[0131] Based on different communication technologies, multiple communication modules 9110 can be configured in the same electronic device, such as cellular network modules, Bluetooth modules, and / or wireless LAN modules. The communication module 9110 (transmitter / receiver) is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide audio output via the speaker 9131 and receive audio input from the microphone 9132, thereby realizing typical telecommunications functions. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. Additionally, the audio processor 9130 is coupled to a central processing unit 9100, enabling on-device recording via the microphone 9132 and on-device playback of stored audio via the speaker 9131.

[0132] Embodiments of this application also provide a computer-readable storage medium capable of implementing all steps of the airspace planning and optimization decision-making method based on a historical flight knowledge base, where the execution subject is a server or client, as described in the above embodiments. The computer-readable storage medium stores a computer program that, when executed by a processor, implements all steps of the airspace planning and optimization decision-making method based on a historical flight knowledge base, where the execution subject is a server or client, as described in the above embodiments. For example, when the processor executes the computer program, it implements the following steps: Step S101: Collect historical flight track data streams, construct an airspace planning meta-model containing flight segment boundary parameters and splitting rules, perform filtering preprocessing on the historical flight track data streams according to the airspace planning meta-model, determine flight segment switching points based on preset segmenting thresholds, calculate segmenting parameters to generate flight segment feature models, record the associated scene features of the flight segment feature models, and establish a flight segment dataset. Step S102: Construct a multidimensional association matrix, calculate the membership coefficient between the flight segment feature model and the associated scene features, generate a weight index based on the similarity distribution, combine the membership coefficient and the weight index to form an association strength value, establish a knowledge graph containing flight segment features, scene attributes, and association strength, and construct a flight segment feature matching model. Step S103: Receive flight path planning request, extract planning constraint parameters, retrieve segment feature models with matching degree exceeding the threshold according to the knowledge graph, calculate candidate segment scores using the segment feature matching model, sort and filter according to the association strength value, combine to generate a complete route, output flight path planning results, and write the flight path planning results into the planning decision database.

[0133] As described above, the computer-readable storage medium provided in this application, through an innovative design of a trajectory processing system, achieves effective data utilization through meta-modeling and feature extraction. It constructs a knowledge graph mechanism, combining association analysis and weight calculation to establish a reliable matching model. A scoring and screening process is introduced, ensuring the rationality of the planning through candidate evaluation and path combination. This method effectively addresses the shortcomings of traditional technologies in trajectory processing, knowledge modeling, and decision optimization, providing technical support for flight path planning.

[0134] Embodiments of this application also provide a computer program product capable of implementing all steps in the airspace planning and optimization decision-making method based on a historical flight knowledge base, where the execution subject is a server or client, as described in the above embodiments. When executed by a processor, this computer program / instruction implements the steps of the airspace planning and optimization decision-making method based on a historical flight knowledge base. For example, the computer program / instruction implements the following steps: Step S101: Collect historical flight track data streams, construct an airspace planning meta-model containing flight segment boundary parameters and splitting rules, perform filtering preprocessing on the historical flight track data streams according to the airspace planning meta-model, determine flight segment switching points based on preset segmenting thresholds, calculate segmenting parameters to generate flight segment feature models, record the associated scene features of the flight segment feature models, and establish a flight segment dataset. Step S102: Construct a multidimensional association matrix, calculate the membership coefficient between the flight segment feature model and the associated scene features, generate a weight index based on the similarity distribution, combine the membership coefficient and the weight index to form an association strength value, establish a knowledge graph containing flight segment features, scene attributes, and association strength, and construct a flight segment feature matching model. Step S103: Receive flight path planning request, extract planning constraint parameters, retrieve segment feature models with matching degree exceeding the threshold according to the knowledge graph, calculate candidate segment scores using the segment feature matching model, sort and filter according to the association strength value, combine to generate a complete route, output flight path planning results, and write the flight path planning results into the planning decision database.

[0135] As described above, the computer program product provided in this application, through an innovative design of a trajectory processing system, achieves effective data utilization through meta-modeling and feature extraction. It constructs a knowledge graph mechanism, combining association analysis and weight calculation to establish a reliable matching model. A scoring and screening process is introduced, ensuring the rationality of the planning through candidate evaluation and path combination. This method effectively addresses the shortcomings of traditional technologies in trajectory processing, knowledge modeling, and decision optimization, providing technical support for flight path planning.

[0136] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0137] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0138] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0139] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0140] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for airspace planning and optimization decision-making based on a historical flight knowledge base, characterized in that, The method includes: Collect historical flight track data streams, construct an airspace planning meta-model that includes flight segment boundary parameters and splitting rules, perform filtering preprocessing on the historical flight track data streams based on the airspace planning meta-model, determine flight segment switching points based on preset segmenting thresholds, calculate segmenting parameters to generate flight segment feature models, record the associated scene features of the flight segment feature models, and establish a flight segment dataset. Construct a multidimensional association matrix, calculate the membership coefficient between the flight segment feature model and the associated scene features, generate a weight index based on the similarity distribution, combine the membership coefficient and the weight index to form an association strength value, establish a knowledge graph containing flight segment features, scene attributes, and association strength, and construct a flight segment feature matching model. The system receives flight path planning requests, extracts planning constraint parameters, retrieves segment feature models with matching degrees exceeding a threshold based on the knowledge graph, calculates candidate segment scores using the segment feature matching models, sorts and filters according to the association strength values, combines them to generate complete routes, outputs flight path planning results, and writes the flight path planning results into the planning decision database.

2. The airspace planning and optimization decision-making method based on historical flight knowledge base according to claim 1, characterized in that, The process involves collecting historical flight track data streams, constructing an airspace planning meta-model that includes segment boundary parameters and splitting rules, and performing filtering preprocessing on the historical flight track data streams based on the airspace planning meta-model, including: By accessing multi-source historical flight track data, extracting the position coordinates, heading angle, speed, and altitude parameters of the flight track points, performing time-series alignment on the data, constructing a flight track parameter sequence, generating data filtering rules based on a preset noise threshold, and forming a standardized flight track dataset; Based on the flight track dataset, establish flight segment classification boundary conditions, construct an airspace planning meta-model that includes location distribution, speed variation, and altitude variation, write the data filtering rules into the airspace planning meta-model, and generate a flight segment classification and preprocessing rule library.

3. The airspace planning and optimization decision-making method based on historical flight knowledge base according to claim 1, characterized in that, The process of determining segment switching points based on preset segmentation thresholds, calculating segmentation parameters to generate a segment feature model, recording the associated scene features of the segment feature model, and establishing a segment dataset includes: Based on the airspace planning meta-model, a segment switching judgment threshold is set, a sliding window analysis is performed on the trajectory data, the parameter change rate between adjacent windows is calculated, and the segment switching point is marked according to the comparison result of the parameter change rate and the judgment threshold, thereby generating a segment segmentation mark sequence. The flight track data is divided into multiple sub-segments using the segment segmentation marker sequence. The feature parameters of each sub-segment are calculated, a segment feature vector is constructed, the scene attributes associated with the segments are extracted, and a segment feature model data structure is established.

4. The airspace planning and optimization decision-making method based on historical flight knowledge base according to claim 1, characterized in that, The process of constructing a multidimensional association matrix, calculating the membership coefficients between the flight segment feature model and the associated scene features, and generating weight indicators based on similarity distribution includes: The flight segment feature model is constructed into an association matrix according to spatial attributes, temporal attributes, and scene attributes. The data distribution difference value between each flight segment feature and the associated scene feature is calculated. Based on the difference value, a multi-dimensional feature vector is generated, and a feature mapping space is constructed. The feature mapping space is normalized, the spatial distribution density of the multidimensional feature vector is calculated, a membership calculation function is generated based on the density distribution, and the correlation coefficient matrix between the flight segment features and the scene features is generated using the membership calculation function.

5. The airspace planning and optimization decision-making method based on historical flight knowledge base according to claim 1, characterized in that, The step of combining the membership coefficient with the weight index to form a correlation strength value, establishing a knowledge graph containing flight segment features, scene attributes, and correlation strength, and constructing a flight segment feature matching model includes: Construct a combination function of membership coefficient and weight index, calculate the correlation strength value for each pair of segment features and scene features, generate a correlation strength matrix, and establish a mapping relationship table between segment features and scene features; Based on the mapping relationship table, a knowledge graph node set is constructed, and the association strength matrix is ​​written into the connection weights between nodes to generate a segment feature matching model that includes node attributes, edge attributes, and weight parameters.

6. The airspace planning and optimization decision-making method based on historical flight knowledge base according to claim 1, characterized in that, The process of receiving a flight path planning request, extracting planning constraint parameters, retrieving segment feature models with matching degrees exceeding a threshold based on the knowledge graph, and calculating candidate segment scores using the segment feature matching model includes: The system receives flight path planning request messages, parses the coordinates of the planning start and end points, flight time window, and scenario constraints, converts the parsing results into planning parameter vectors, and retrieves flight segment nodes in the knowledge graph based on the planning parameter vectors to generate a set of candidate flight segments. The matching coefficients between the segment feature model and the planning parameter vector in the candidate segment set are calculated. The matching coefficients are then normalized using the segment feature matching model to generate a segment score sequence. Valid segments are then selected according to the score threshold.

7. The airspace planning and optimization decision-making method based on historical flight knowledge base according to claim 1, characterized in that, The process of sorting and filtering according to the correlation strength value, combining them to generate a complete flight route, outputting the flight path planning result, and writing the flight path planning result into the planning decision database includes: Candidate flight segments are prioritized based on their correlation strength values, flight segment connection constraints are constructed, and the sorted flight segments are combined sequentially according to the connection constraints to generate multiple candidate flight route schemes. The comprehensive score of each candidate flight route is calculated. The candidate route with the best score is selected as the planning result. The waypoint sequence, altitude sequence, and speed sequence of the planning result are extracted to generate a standard format flight path planning document, which is then written into the planning decision database.

8. An airspace planning and optimization decision-making device based on a historical flight knowledge base, characterized in that, The device includes: The segment determination module is used to collect historical flight track data streams, construct an airspace planning meta-model containing segment boundary parameters and splitting rules, perform filtering preprocessing on the historical flight track data streams based on the airspace planning meta-model, determine segment switching points based on preset segmenting thresholds, calculate segmenting parameters to generate a segment feature model, record the associated scene features of the segment feature model, and establish a segment dataset. The feature matching module is used to construct a multi-dimensional association matrix, calculate the membership coefficient between the flight segment feature model and the associated scene features, generate a weight index based on the similarity distribution, combine the membership coefficient and the weight index to form an association strength value, establish a knowledge graph containing flight segment features, scene attributes, and association strength, and construct a flight segment feature matching model. The route planning module is used to receive flight route planning requests, extract planning constraint parameters, retrieve segment feature models with matching degrees exceeding a threshold based on the knowledge graph, calculate candidate segment scores using the segment feature matching models, sort and filter according to the association strength values, combine to generate a complete route, output flight route planning results, and write the flight route planning results into the planning decision database.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the airspace planning and optimization decision-making method based on historical flight knowledge base as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the airspace planning and optimization decision-making method based on a historical flight knowledge base as described in any one of claims 1 to 7.

Images