An intelligent decision-making method based on an unmanned aerial vehicle autonomous support platform

Abstract

This invention discloses an intelligent decision-making method based on an autonomous unmanned aerial vehicle (UAV) support platform, comprising the following steps: collecting information on the task, UAV, support resources, airspace, communication, and environment; generating a standardized platform state through spatiotemporal alignment; generating a rule threshold index table and calculating an evidence commitment vector; constructing a budget ledger and maintaining gating consistency to obtain a feasible domain proof fragment; generating a joint scheduling scheme and shadow price state based on evidence coupling with a shadow price field; using a risk budget corridor to constrain incremental replanning of output routes, support replanning, and occupancy indicators; and performing closed-loop updates. This invention employs an isomorphic decision kernel of evidence commitment vector – budget ledger – feasible domain proof fragment, coordinating shadow price joint scheduling and risk budget corridor incremental replanning to achieve interpretable and auditable closed-loop updates, possessing advantages such as stable decision-making, low jitter, and strong continuity.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) system technology, and in particular to an intelligent decision-making method based on an autonomous support platform for UAVs. Background Technology

[0002] With the large-scale application of drones in scenarios such as inspection, logistics and emergency response, existing technologies for platform-based operation typically adopt route planning and dynamic management schemes under airspace rule constraints, and make rolling adjustments in combination with task allocation, site support and communication link status. At the same time, some systems decouple scheduling optimization and path planning, and generate task plans, support arrangements and route updates based on rule base judgment, resource scheduling model and dynamic shortest path search algorithm, respectively, and trigger recalculation through monitoring data to maintain task continuity and airspace compliance.

[0003] However, the aforementioned existing technologies mainly rely on the interconnection of rule determination, scheduling solution, and replanning modules, lacking a unified evidence encapsulation and consistency maintenance mechanism. This makes it difficult to form reusable feasible domain proofs for rule conflicts, budget constraints, and evidence credibility, resulting in unauditable conflict interpretation and minimum repair. At the same time, shadow price-related constraint adjustment and incremental replanning often lack corridor-based entry points and influence domain limitations for risk budget constraints, which can easily lead to excessively large replanning scopes, insufficient linkage on the protection side, and scheme instability. Execution feedback also mostly relies on experience-based parameter tuning, making it difficult to form a controllable closed-loop feedback update.

[0004] Therefore, how to provide an intelligent decision-making method based on an autonomous support platform for unmanned aerial vehicles (UAVs) is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose an intelligent decision-making method based on an autonomous support platform for unmanned aerial vehicles (UAVs). This invention adopts an isomorphic decision-making kernel of evidence commitment vector-budget ledger-feasibility domain proof fragment, and coordinates shadow price joint scheduling and risk budget corridor incremental replanning to achieve interpretable and auditable closed-loop updates. It has the advantages of stable decision-making, low jitter and strong continuity.

[0006] An intelligent decision-making method based on an unmanned aerial vehicle (UAV) autonomous support platform according to an embodiment of the present invention includes the following steps: Acquire task queues, UAV status sets, support resource status sets, airspace constraint information sets, communication status sets, and environmental information sets; unify spatiotemporal benchmarks and generate standardized platform status sets. Based on the standardized platform state set and rule entry set, a rule threshold index table is generated, and rule hit determination, constraint violation magnitude generation and evidence credibility generation are performed, and the evidence commitment vector is encapsulated. A budget ledger is generated based on a standardized platform state set, and budget gating consistency is maintained by combining it with a rule threshold index table and an evidence commitment vector to form a feasible domain proof fragment. Based on the evidence commitment vector, budget ledger and feasible region proof fragment, construct the evidence-coupled shadow price field and perform Lagrange decomposition. Combined with the interpretation-oriented large neighborhood search repair generation task-UAV-support resource joint scheduling scheme, output the shadow price state. Based on the joint scheduling scheme of mission-UAV-support resources, shadow price status, budget ledger and feasible domain proof fragment, an incremental reachable corridor with risk budget constraints is generated. D*Lite incremental replanning is performed to obtain incrementally updated route scheme, support replanning scheme and risk budget occupancy index. Based on the shadow price status, risk budget occupancy indicators, and execution feedback, the rule threshold index table, shadow price field, and budget ledger are updated by performing a set of closed actions such as adjustment, reduction, freeze, and rollback.

[0007] Optionally, the generation of the standardized platform state set specifically includes: Acquire task queues, drone status sets, support resource status sets, airspace constraint information sets, communication status sets, and environmental information sets; extract and organize fields to generate a set of original platform elements. Perform unified time base alignment on the timestamps of the platform's original element set, and select the platform clock as the unified time base to generate a set of time mapping parameters; A unified sampling time sequence is generated based on a unified time benchmark. Time synchronization resampling is performed on the original set of elements of the platform to construct an aligned time index table and obtain a synchronized set of elements of the platform. The spatial reference is unified for the set of elements of the synchronization platform, and a set of unified coordinate system definition information and spatial mapping parameters for the platform is constructed. Under the unified coordinate system of the platform, the spatial constraint information set is rasterized and indexed to generate a spatial raster occupancy table and a spatial availability index table. The combined index is then encapsulated to obtain a spatial constraint index set, which is written into the synchronous platform element set. Under the unified coordinate system of the platform, the site-level resource aggregation and occupancy coding are performed on the set of guaranteed resource statuses to generate a set of guaranteed resource occupancy vectors and a set of guaranteed resource capacity indexes, which are then written back to the set of elements of the synchronous platform to form the basic resource status. The task queue is aggregated into task clusters and the constraint fields are fixed. A task cluster index table and a task time window index table are generated. A composite index is encapsulated to obtain a structured task index set and written into the synchronization platform element set. The consistency of the platform element set, alignment time index table, and platform unified coordinate system definition information is checked and conflicts are eliminated to generate a standardized platform state set.

[0008] Optionally, the generation of the evidence commitment vector specifically includes: Load a set of rule entries from the rule base of the UAV autonomous support platform, and construct a rule entry specification table based on the standardized platform status set; A rule threshold index table is generated for each rule based on the rule entry specification table; Based on the rule threshold index table, rule hit determination is performed on the standardized platform state set, and the rule hit determination result set is output. Constraint violation magnitude is generated based on the rule threshold index table and the rule hit judgment result set, and bound to the rule identifier, object identifier and alignment time index to form a violation magnitude index record; A set of evidence credibility entries is generated based on a standardized platform state set, and then bound to rule identifiers, object identifiers, and alignment time indexes to form credibility index records; Perform credibility aggregation and gate tag generation on the evidence credibility item set to obtain rule-level evidence credibility and evidence gate tag set, and write them into the credibility index record; The rule hit result set, violation magnitude index record, and credibility index record are indexed and encapsulated to generate an evidence commitment vector.

[0009] Optionally, the generation of the feasible region proof fragment specifically includes: Read the standardized platform status set, perform resource aggregation, window aggregation, and status aggregation, and generate a candidate set of budget ledger entries; Generate a budget ledger based on the candidate set of budget ledger entries; Map the rule threshold index table and the evidence commitment vector to the budget ledger to construct a budget gating entry mapping table; Based on the budget gating entry mapping table, perform rule enable and rule disable determination on the rule threshold index table, output the rule enable flag set and rule disable flag set and write them back to the budget gating entry mapping table; Under the constraint of the rule enable tag set, perform conflict candidate generation on the rule hit identifier set and the violation amplitude index record in the evidence commitment vector, and output the conflict candidate set; Perform conflict resolution priority determination on the conflict candidate set, obtain the conflict resolution result set and generate a conflict interpretation package; A minimum set of correction rules is generated based on the set of conflict resolution results and the budget gating entry mapping table, and together with the conflict interpretation package, a feasibility label is generated. The conflict interpretation package, the minimum correction rule set, and the feasibility label are indexed and encapsulated to generate a feasible domain proof fragment.

[0010] Optionally, the conflict adjudication priority determination adopts a closed priority sequence for sequential adjudication, with the airspace object rule entry set taking precedence over the UAV object rule entry set, the UAV object rule entry set taking precedence over the support resource object rule entry set, the support resource object rule entry set taking precedence over the communication object rule entry set, and the communication object rule entry set taking precedence over the task object rule entry set. Within the same priority, the adjudication is based on the violation magnitude index, from largest to smallest.

[0011] Optionally, the conflict interpretation package includes a set of conflict minimum set identifiers, a set of conflict cause categories, a conflict rule association table, a conflict budget association table, a summary of conflict violation magnitude, and conflict evidence citation records.

[0012] Optionally, the generation of the shadow price state specifically includes: Read the evidence commitment vector, budget ledger and feasible region proof fragment, and generate a scheduling index structure based on the task cluster identifier, guarantee site identifier and alignment time index table in the normalized platform state set; Construct evidence-coupled shadow price fields based on scheduling index structures and generate a set of shadow price field entries; Map the conflict budget association table and the conflict cause category set in the feasible region proof fragment to the shadow price field entry set to form a targeted update index set; Based on the budget ledger and the set of shadow price field entries, construct the input set for the Lagrange decomposition, forming a set of decomposition subproblems; Iterative solutions are performed on the set of subproblems to generate a set of candidate joint scheduling solutions; Perform a targeted update on the shadow price field entry set based on the targeted update index set and output the shadow price status, including the updated shadow price field entry set and its quad octet closure. Based on the conflict interpretation package and the minimum correction rule set, an interpretation-guided large neighborhood search is performed on the candidate joint scheduling solution set to obtain feasible repair results; The feasible repair results and candidate joint scheduling solutions are consistent and normalized, and then merged according to the task identifier and alignment time index table to generate a joint scheduling scheme for task-UAV-support resources, which is then output in conjunction with the shadow price status.

[0013] Optionally, the generation of the route plan, the support replanning plan, and the risk budget occupancy index specifically includes: Read the joint scheduling scheme of task-drone-support resources, expand the task allocation results, take-off and landing scheduling results, charging scheduling results, maintenance scheduling results and communication window scheduling results according to the alignment time index table, and generate a replanning input set; A risk budget constraint parameter set is generated based on the set of risk budget entries in the budget ledger and the set of conflict cause categories in the feasible region proof fragment; Based on the risk budget constraint parameter set and the airspace grid occupancy table, airspace availability index table and link quality observation corresponding to the communication link identifier in the normalized platform state set, an incremental reachable corridor for risk budget constraints is generated. Perform edge cost reparameterization on incremental reachable corridors to form a set of reachable corridor edge weights; Based on the incremental reachable corridor, the set of reachable corridor edge weights, and the replanning input set, D*Lite incremental replanning is performed to generate a set of influence domain objects; The route plan and support replanning plan are updated incrementally based on the set of objects in the influence domain and the set of reachable corridor edges. Based on incrementally updated route plans and support replanning plans, risk budget occupancy indicators are generated and linked to the budget ledger for output.

[0014] Optionally, the incremental reachable corridor consists of a reachable corridor geometric domain set, a reachable corridor edge set, and a reachable corridor edge weight set. The reachable corridor geometric domain set is obtained by expanding the centerline of the flight segment set to be executed in the platform's unified coordinate system and being clipped by the corridor width parameter. The reachable corridor edge set is generated by the flight segment adjacency relationship and the airspace grid connectivity relationship. The reachable corridor edge weight set is obtained by reparameterizing the edge cost. The corridor width parameter is limited by the minimum corridor width and the maximum corridor width, and is obtained by proportionally mapping the remaining available amount of the risk budget item set between the minimum corridor width and the maximum corridor width.

[0015] Optionally, the generation of the reflow update specifically includes: Read the shadow price status, risk budget occupancy indicator sequence and budget ledger, collect the execution feedback of incrementally updated route plans and support replanning plans, and generate an execution feedback set; Based on the execution feedback set and the task-UAV-support resource joint scheduling scheme, a plan-execution deviation record set is generated, which is then grouped into a set of return flow statistics entries according to the support site identifier, task cluster identifier, and alignment time index table. A risk repatriation trigger flag set is generated based on the risk budget occupancy indicator sequence, the set of repatriation statistics items, and the set of risk budget items in the budget ledger; A set of resource repatriation trigger markers is generated based on the set of repatriation statistics entries, shadow price status, and budget ledger. Under the constraints of the risk backflow trigger flag set and the resource backflow trigger flag set, execute the backflow action adjudication to generate a backflow action set; Based on the set of backflow actions, the rule threshold index table, shadow price field and budget ledger are updated and a set of updated configurations is generated and written back to the UAV autonomous support platform.

[0016] The beneficial effects of this invention are: This invention constructs an isomorphic decision kernel consisting of an evidence commitment vector, a budget ledger, and feasible domain proof fragments, using a standardized platform state set as a unified input. This kernel enables the rule threshold index table, budget gating entry mapping table, conflict interpretation package, minimum correction rule set, and feasibility label to be connected in a closed loop within the same link. Under the dynamic coupling of task, security resources, and airspace communication constraints, this invention achieves interpretable conflict resolution and auditable minimum repair, significantly reducing decision failures and repeated recalculations caused by inconsistent rules and unreliable evidence, and improving the stability and executability of joint scheduling.

[0017] Meanwhile, this invention uses evidence-coupled shadow price field quaternary index closures to achieve targeted constraint adjustment of resource-site-time-task clusters, and maps the risk tension classification results to corridor parameter index records, forming an incremental reachable corridor and influence domain object set for risk budget constraints. Then, in D*Lite incremental replanning, route generation is restricted to the reachable corridor edge set covered by the influence domain object set and aligned and bound with the guarantee replanning scheme, reducing route and guarantee scheme jitter and invalid switching. Combined with closed action feedback updates of adjustment, reduction, freeze, and rollback, the shadow price status, rule threshold index table, and budget ledger evolve controllably under execution feedback, continuously improving task continuity guarantee and risk budget controllability. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of an intelligent decision-making method based on an unmanned aerial vehicle (UAV) autonomous support platform proposed in this invention; Figure 2 This is a flowchart illustrating the generation of a feasible domain proof fragment for budget gating consistency maintenance in an intelligent decision-making method based on an unmanned aerial vehicle (UAV) autonomous support platform proposed in this invention. Figure 3 This is a flowchart illustrating the process of generating route plans, support replanning plans, and risk budget occupancy indicators through the linkage of incremental reachable corridors and D*Lite incremental replanning with a risk budget constraint in an intelligent decision-making method based on an autonomous UAV support platform proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-3 An intelligent decision-making method based on an autonomous support platform for unmanned aerial vehicles (UAVs) includes the following steps: Acquire task queues, UAV status sets, support resource status sets, airspace constraint information sets, communication status sets, and environmental information sets; unify spatiotemporal benchmarks and generate standardized platform status sets. Based on the standardized platform state set and rule entry set, a rule threshold index table is generated, and rule hit determination, constraint violation magnitude generation and evidence credibility generation are performed, and the evidence commitment vector is encapsulated. A budget ledger is generated based on a standardized platform state set, and budget gating consistency is maintained by combining it with a rule threshold index table and an evidence commitment vector to form a feasible domain proof fragment. Based on the evidence commitment vector, budget ledger and feasible region proof fragment, construct the evidence-coupled shadow price field and perform Lagrange decomposition. Combined with the interpretation-oriented large neighborhood search repair generation task-UAV-support resource joint scheduling scheme, output the shadow price state. Based on the joint scheduling scheme of mission-UAV-support resources, shadow price status, budget ledger and feasible domain proof fragment, an incremental reachable corridor with risk budget constraints is generated. D*Lite incremental replanning is performed to obtain incrementally updated route scheme, support replanning scheme and risk budget occupancy index. Based on the shadow price status, risk budget occupancy indicators, and execution feedback, the rule threshold index table, shadow price field, and budget ledger are updated by performing a set of closed actions such as adjustment, reduction, freeze, and rollback.

[0021] In this embodiment, the generation of the standardized platform state set specifically includes: Acquire task queues, drone status sets, support resource status sets, airspace constraint information sets, communication status sets, and environmental information sets; extract and organize fields to generate a set of original platform elements. The platform's original element set includes task identifier, task location, task time window, task cluster identifier, UAV identifier, UAV current location, UAV remaining range, UAV health status, support station identifier, support station location, take-off and landing position occupancy status, charging position occupancy status, maintenance position occupancy status, airspace area identifier, airspace availability status, communication link identifier, link quality observation, and environmental observation items. Perform unified time base alignment on the timestamps of the platform's original element set, and select the platform clock as the unified time base to generate a set of time mapping parameters; A unified sampling time sequence is generated based on a unified time benchmark. Time synchronization resampling is performed on the original set of elements of the platform to construct an aligned time index table and obtain a synchronized set of elements of the platform. The time synchronization resampling includes missing frame interpolation and duplicate frame deduplication, and is arranged according to a unified sampling time. The set of elements of the synchronization platform includes a set of synchronization task queue segments, a set of synchronization UAV status segments, a set of synchronization support resource status segments, a set of synchronization airspace constraint segments, a set of synchronization communication status segments, and a set of synchronization environmental information segments. The spatial reference is unified for the set of elements of the synchronization platform, and a set of unified coordinate system definition information and spatial mapping parameters for the platform is constructed. The platform's unified coordinate system definition information includes the coordinate origin, coordinate axis directions, distance scale reference, and height reference. The spatial mapping parameter set maps the mission location, the current location of the UAV, the location of the support station, and the airspace boundary to the platform's unified coordinate system. The position in the platform's unified coordinate system is obtained by rotating and then superimposing translation transformations on the position in the original coordinate system. Under the unified coordinate system of the platform, the spatial constraint information set is rasterized and indexed to generate a spatial raster occupancy table and a spatial availability index table. The combined index is then encapsulated to obtain a spatial constraint index set, which is written into the synchronous platform element set. The spatial raster occupancy table contains the mapping relationship between spatial region identifiers and raster cell sets and the availability status of raster cells, while the spatial availability index table contains the binding relationship between alignment time indexes and spatial availability status sequences. Under the unified coordinate system of the platform, the site-level resource aggregation and occupancy coding are performed on the set of guaranteed resource statuses to generate a set of guaranteed resource occupancy vectors and a set of guaranteed resource capacity indexes, which are then written back to the set of elements of the synchronous platform to form the basic resource status. The guaranteed resource occupancy vector set includes the unified encoding results of take-off and landing position occupancy status, charging position occupancy status, and maintenance position occupancy status, while the guaranteed resource capacity index set includes the binding results of site identifier and alignment time index. The task queue is aggregated into task clusters and the constraint fields are fixed. A task cluster index table and a task time window index table are generated. A composite index is encapsulated to obtain a structured task index set and written into the synchronization platform element set. The task cluster index table maps task identifiers to task cluster identifiers and binds task positions to alignment time indexes, while the task time window index table maps task identifiers to task time windows and binds alignment time indexes. The consistency of the platform element set, alignment time index table, and platform unified coordinate system definition information is checked and conflicts are eliminated to generate a standardized platform state set.

[0022] In this embodiment, the generation of the evidence commitment vector specifically includes: Load a set of rule entries from the rule base of the UAV autonomous support platform, and construct a rule entry specification table based on the standardized platform status set; The rule base consists of a rule identifier set, a rule trigger condition description set, a rule threshold configuration set, and a rule output action tag set. The rule entry set is classified according to the type of object to which the rule applies: task object rule entry set, UAV object rule entry set, support resource object rule entry set, airspace object rule entry set, and communication object rule entry set. The rule entry specification table includes rule identifier, rule applicable object type, rule applicable scope, rule triggering condition description, rule threshold type marker and rule output action marker, and limits the rule applicable object type to at least one of the following: task object, UAV object, support resource object, airspace object and communication object; A rule threshold index table is generated for each rule based on the rule entry specification table; The rule threshold index table contains the mapping relationship between rule identifiers and threshold entry sets, the binding relationship between threshold entries and normalized platform state set fields, the binding relationship between threshold entries and alignment time indexes, and the binding relationship between threshold entries and object identifiers. The threshold entry set includes airspace availability threshold entries, mission time window threshold entries, UAV remaining range threshold entries, UAV health status threshold entries, take-off and landing slot occupancy threshold entries, charging slot occupancy threshold entries, maintenance slot occupancy threshold entries, and link quality threshold entries. Based on the rule threshold index table, rule hit determination is performed on the standardized platform state set, and the rule hit determination result set is output. The rule hit determination includes extracting the state fragment of the corresponding object according to the binding relationship of the object identifier for each rule, extracting the corresponding time fragment according to the binding relationship of the alignment time index, and comparing the state fragment with the threshold entry set one by one; The rule hit determination result set includes a rule hit identifier set, a rule miss identifier set, and a rule applicability missing identifier set. The rule applicability missing identifier set is generated from the status flags in the standardized platform status set that indicate that the corresponding field is missing or the corresponding object is unavailable. Constraint violation magnitude is generated based on the rule threshold index table and the rule hit judgment result set, and bound to the rule identifier, object identifier and alignment time index to form a violation magnitude index record; The constraint violation magnitude is generated one by one according to the threshold entries, and the upper bound threshold entries, lower bound threshold entries, and interval threshold entries are distinguished. Among them, the constraint violation magnitude of the upper bound threshold item is determined by the amount by which the state value exceeds the upper bound threshold. The constraint violation magnitude is zero when the state value does not exceed the upper bound threshold. The constraint violation magnitude of the lower bound threshold item is determined by the amount by which the lower bound threshold exceeds the state value. The constraint violation magnitude is zero when the state value is not lower than the lower bound threshold. The constraint violation magnitude of the interval threshold item is determined by the sum of the amount by which the lower bound threshold exceeds the state value and the amount by which the state value exceeds the upper bound threshold. The constraint violation magnitude is zero when the state value is between the lower bound threshold and the upper bound threshold. A set of evidence credibility entries is generated based on a standardized platform state set, and then bound to rule identifiers, object identifiers, and alignment time indexes to form credibility index records; The credibility of the evidence is jointly determined by sensor availability, communication availability, airspace data availability and support resource data availability. Sensor availability is generated by the state update integrity mark and anomaly removal mark of the UAV state set. Communication availability is generated by the link quality observation and link continuity mark of the communication state set. Airspace data availability is generated by the integrity mark of the airspace availability state and airspace constraint index set. Support resource data availability is generated by the write-back integrity mark of the support resource occupancy vector set. Perform credibility aggregation and gate tag generation on the evidence credibility item set to obtain rule-level evidence credibility and evidence gate tag set, and write them into the credibility index record; The credibility aggregation is aligned with the object type in the rule entry specification table. The sensor availability, communication availability, airspace data availability and security resource data availability bound to the same rule identifier are multiplied and aggregated to generate rule-level evidence credibility. Evidence gating tags are generated based on the rule-level evidence credibility and the rule threshold type tags in the rule entry specification table. The rule hit result set, violation magnitude index record, and credibility index record are indexed and encapsulated to generate an evidence commitment vector.

[0023] In this embodiment, the generation of the feasible region proof fragment specifically includes: Read the standardized platform status set, perform resource aggregation, window aggregation, and status aggregation, and generate a candidate set of budget ledger entries; Among them, the resource aggregation of take-off and landing position occupancy status, charging position occupancy status and maintenance position occupancy status is carried out according to the support station identifier and alignment time index table; the link quality observation is aggregated according to the communication link identifier and alignment time index table; and the airspace availability status is aggregated according to the airspace area identifier and alignment time index table. Generate a budget ledger based on the candidate set of budget ledger entries; The budget ledger includes a set of resource budget items, a set of risk budget items, a set of time budget items, and a set of update density budget items. The set of resource budget items includes take-off and landing capacity items, charging capacity items, maintenance capacity items, and communication window capacity items. The set of risk budget items includes airspace unavailability risk budget items, link discontinuity risk budget items, and UAV health status risk budget items. The set of time budget items includes task time window relaxation budget items and queuing delay budget items. The set of update density budget items includes scheduling refresh interval budget items and incremental replanning trigger frequency budget items. Map the rule threshold index table and the evidence commitment vector to the budget ledger to construct a budget gating entry mapping table; The budget gating entry mapping table maps rule identifiers and threshold item sets to the target budget item set in the budget ledger, and binds the rule hit identifier set, violation amplitude index record, credibility index record and evidence gating mark set in the evidence commitment vector to the target budget item set, serving as the sole gating input for budget gating consistency maintenance. Based on the budget gating entry mapping table, perform rule enable and rule disable determination on the rule threshold index table, output the rule enable flag set and rule disable flag set and write them back to the budget gating entry mapping table; The rule activation determination is jointly determined by the availability status of the target budget item set and the gating status of the evidence gating mark set, while the rule disabling determination is triggered by the unavailability status of the target budget item set or the disabling status of the evidence gating mark set. Under the constraint of the rule enable tag set, perform conflict candidate generation on the rule hit identifier set and the violation amplitude index record in the evidence commitment vector, and output the conflict candidate set; The conflict candidate generation combines mutually exclusive rule pairs, resource capacity-related rule pairs, and risk budget-related rule pairs within the same alignment time index table into conflict candidate sets, and binds each conflict candidate set to the corresponding target budget item set and violation magnitude index record. Perform conflict resolution priority determination on the conflict candidate set, obtain the conflict resolution result set and generate a conflict interpretation package; A minimum set of correction rules is generated based on the set of conflict resolution results and the budget gating entry mapping table, and together with the conflict interpretation package, a feasibility label is generated. The minimum correction rule set is generated using the closed minimum repair decision criterion, and the decision order is based on the minimum budget ledger disturbance priority and the minimum time budget occupation second-best. The minimum budget ledger disturbance is determined by the minimum number of adjustments to the target budget item set, and the minimum time budget occupation is determined by the minimum number of task time window relaxation budget items and queuing delay budget item occupation items. The feasibility label is limited to three categories: feasible, infeasible, and downgraded feasible. When the set of conflict minimum set identifiers in the conflict interpretation package is empty or the set of conflict cause categories is empty, the feasibility label is feasible. When the set of conflict minimum set identifiers in the conflict interpretation package is not empty and the set of minimum correction rules is empty, the feasibility label is infeasible. When the set of conflict minimum set identifiers in the conflict interpretation package is not empty and the set of minimum correction rules is not empty, the feasibility label is downgraded feasible if the target budget entries referenced in the conflict budget association table satisfy the gating constraints of the budget gating entry mapping table and do not trigger the disabled state of the rule disabled flag set after the repair is performed according to the rule identifier corresponding to each repair rule in the minimum correction rule set in the threshold entry set of the rule threshold index table, and the disabled state is not triggered. Otherwise, the feasibility label is infeasible. The conflict interpretation package, the minimum correction rule set, and the feasibility label are indexed and encapsulated to generate a feasible domain proof fragment.

[0024] In this embodiment, the conflict adjudication priority is determined by a closed priority sequence. The set of rules for airspace objects takes precedence over the set of rules for UAV objects, the set of rules for UAV objects takes precedence over the set of rules for support resources objects, the set of rules for support resources objects takes precedence over the set of rules for communication objects, and the set of rules for communication objects takes precedence over the set of rules for task objects. Within the same priority, the adjudication is based on the violation severity index, from largest to smallest.

[0025] In this embodiment, the conflict interpretation package includes a minimum set of conflict identifiers, a set of conflict cause categories, a conflict rule association table, a conflict budget association table, a summary of conflict violation magnitude, and a record of conflict evidence citations. The conflict minimum set identifier set is formed by removing redundancy from the rule identifier set corresponding to each adjudicated conflict in the conflict adjudication result set. The conflict cause category set is generated by the conflict candidate set type tag corresponding to the conflict and limited to three categories: mutual exclusion conflict, resource capacity conflict, and risk budget conflict. The conflict rule association table is generated by extracting the binding relationship between rule identifiers and object identifiers and the binding relationship between alignment time index in the rule threshold index table. The conflict budget association table is generated by extracting the mapping from rule identifiers to target budget item set in the budget gating entry mapping table. The conflict violation magnitude summary is obtained by aggregating the violation magnitudes bound to the conflict minimum set identifier set in the violation magnitude index record according to the alignment time index table and binding it to the target budget item set. The conflict evidence citation record is generated by extracting the evidence credibility item set and evidence gating tag set bound to the conflict minimum set identifier set in the credibility index record and binding it to the conflict cause category set.

[0026] In this embodiment, the generation of the shadow price state specifically includes: Read the evidence commitment vector, budget ledger and feasible region proof fragment, and generate a scheduling index structure based on the task cluster identifier, guarantee site identifier and alignment time index table in the normalized platform state set; The scheduling index structure includes a task-UAV candidate matching set, a support site candidate set, a time period grid set, and a resource type set, limiting the resource type to take-off and landing positions, charging positions, maintenance positions, and communication windows. Construct evidence-coupled shadow price fields based on scheduling index structures and generate a set of shadow price field entries; The set of entries in the shadow price field is generated using a four-element index closure of resource type-site-time period-task cluster. The resource type is taken from the resource type set, the site is taken from the guaranteed site identifier, the time period is taken from the time period grid of the aligned time index table, and the task cluster is taken from the task cluster identifier. Map the conflict budget association table and the conflict cause category set in the feasible region proof fragment to the shadow price field entry set to form a targeted update index set; The targeted update index set is limited to performing targeted upward or downward adjustments on the shadow price field entries corresponding to the target budget entry set referenced in the conflict budget association table; Based on the budget ledger and the set of shadow price field entries, construct the input set for the Lagrange decomposition, forming a set of decomposition subproblems; The set of decomposed subproblems includes the task-UAV matching subproblem, the resource scheduling subproblem, and the communication window scheduling subproblem. The set of resource budget entries and the set of time budget entries in the budget ledger are converted into the set of local constraints of the set of decomposed subproblems. The set of conflict minimum set identifiers and violation magnitude index records in the conflict interpretation package are converted into the set of penalty terms inputs of the set of decomposed subproblems, and a Lagrange target quantity for unified summarization is generated. The Lagrange target quantity consists of the sum of the basic cost quantity and the coupling constraint penalty term. The coupling constraint penalty term is obtained by multiplying the values ​​of the shadow price field entries and the corresponding constraint violation magnitudes one by one, and then summing the results over all referenced entries. Iterative solutions are performed on the set of subproblems to generate a set of candidate joint scheduling solutions; The iterative solution includes solving the task-UAV matching subproblem, the guaranteed resource scheduling subproblem, and the communication window scheduling subproblem respectively under a fixed shadow price field item set. The consistency of the task time window index table, the guaranteed resource occupancy vector set, and the spatial constraint index set is checked according to the alignment time index table to eliminate infeasible candidate solutions. The candidate solutions that pass the check are written into the candidate joint scheduling solution set. Perform a targeted update on the shadow price field entry set based on the targeted update index set and output the shadow price status, including the updated shadow price field entry set and its quad octet closure. The targeted update limits the conflict cause category set to resource capacity conflict and performs a targeted upward adjustment on the shadow price field entry corresponding to the resource budget entry set; limits the conflict cause category set to risk budget conflict and performs a targeted upward adjustment on the shadow price field entry corresponding to the risk budget entry set; and limits the conflict cause category set to mutual exclusion conflict and performs a targeted downward adjustment on the shadow price field entry mapped by the low-priority rule in the conflict rule association table. The value of the shadow price entry in the targeted update is obtained by superimposing the value of the shadow price entry before the update, the product of the update step size coefficient, the constraint violation magnitude, and the targeted update symbol quantity; Based on the conflict interpretation package and the minimum correction rule set, an interpretation-guided large neighborhood search is performed on the candidate joint scheduling solution set to obtain feasible repair results; The explanation-guided large neighborhood search repair generates a set of damaged objects and performs repair operations. The set of damaged objects is limited to the associated entries in the task-UAV candidate matching set, the site selection entries in the support site candidate set, and the take-off and landing scheduling segments, charging scheduling segments, maintenance scheduling segments, and communication window scheduling segments in the time period grid set. It is obtained by reverse indexing the rule identifiers hit by the conflict minimum set identifier set in the threshold entry set in the rule threshold index table, and the damage order is selected from large to small according to the violation magnitude recorded in the violation magnitude index. Under the constraints of the budget ledger, the repair operation sequentially performs task reassignment, site switching, time window reordering, and communication window reordering. After each repair, the feasibility screening is performed based on the gating constraints of the budget ledger and the feasible domain proof fragment to retain the feasible repair results. The feasible repair results and the candidate joint scheduling solution set are uniformly adjusted, and the task-UAV-support resource joint scheduling scheme is generated by merging the task identifier and the alignment time index table, and then output by binding the shadow price status. The joint scheduling scheme for tasks, drones, and support resources includes task allocation results, drone rotation results, support site selection results, take-off and landing scheduling results, charging scheduling results, maintenance scheduling results, and communication window scheduling results.

[0027] In this implementation method, the generation of route plans, support replanning plans, and risk budget occupancy indicators specifically includes: Read the joint scheduling scheme of task-drone-support resources, expand the task allocation results, take-off and landing scheduling results, charging scheduling results, maintenance scheduling results and communication window scheduling results according to the alignment time index table, and generate a replanning input set; The replanning input set includes the set of flight segments to be executed, the set of candidate support station switching, the set of candidate window adjustments, and the reference relationship between the corresponding target budget item set; A risk budget constraint parameter set is generated based on the set of risk budget entries in the budget ledger and the set of conflict cause categories in the feasible region proof fragment; The risk budget constraint parameter set includes corridor width parameters, corridor edge cost amplification parameters, upper limit parameters of corridor topology change intensity, and influence domain expansion threshold parameters. The risk tension classification result is generated by the remaining available quantity of the risk budget item set and the risk budget occupancy index sequence. Discrete level mapping and truncation correction are performed according to the classification result to obtain the corridor width parameters, corridor edge cost amplification parameters, upper limit parameters of corridor topology change intensity, and influence domain expansion threshold parameters, respectively. These parameters are then bound to the alignment time index table to form corridor parameter index records. Based on the risk budget constraint parameter set and the airspace grid occupancy table, airspace availability index table and link quality observation corresponding to the communication link identifier in the normalized platform state set, an incremental reachable corridor for risk budget constraints is generated. Perform edge cost reparameterization on incremental reachable corridors to form a set of reachable corridor edge weights; The edge cost reparameterization uses the risk budget constraint parameter set as the sole entry point. For each edge in the reachable corridor edge set, the basic cost is calculated and the risk amplification term and price amplification term are superimposed to obtain the edge weight. The risk amplification term is jointly determined by the unavailability mark corresponding to the time period in the airspace availability index table, the low reliability mark corresponding to the time period in the link quality observation, and the abnormal mark corresponding to the time period in the UAV health status. The price amplification term is obtained by mapping the values ​​of the resource type-site-time period-task cluster quadruple index closure entries corresponding to the edge in the shadow price status. The reparameterized edge weight is obtained by summing the products of the basic cost, the corridor edge cost amplification parameter, and the risk amplification term, and superimposing the product of the price amplification coefficient and the value of the shadow price entry. Based on the incremental reachable corridor, the set of reachable corridor edge weights, and the replanning input set, D*Lite incremental replanning is performed to generate a set of influence domain objects; D*Lite incremental replanning uses the set of reachable corridor edge weights as input to maintain a search state with consistent path costs. It uses the set of conflict cause categories in the feasible domain proof fragment to adjudicate the expansion of the influence domain. When the set of conflict cause categories includes path inaccessibility, it performs hierarchical expansion of the grid cells adjacent to inaccessible segments in the airspace grid occupancy table according to the influence domain expansion threshold parameter to generate an influence domain object set and incrementally updates the edge weights within the influence domain object set. When the set of conflict cause categories includes risk budget conflict or resource capacity conflict, it keeps the influence domain object set from expanding and prioritizes triggering the candidate support site switching set and the candidate window adjustment set to enter the same round of incremental replanning. The route plan and support replanning plan are updated incrementally based on the set of objects in the influence domain and the set of reachable corridor edges. The incrementally updated route plan is generated within the set of reachable corridor edges covered by the set of influence domain objects. It is generated by the path with the minimum cumulative edge weight on the set of reachable corridor edges and expanded according to the alignment time index table. The support replanning plan is obtained by filtering the candidate support site switching set and the candidate window adjustment set under the budget ledger constraint and is aligned and bound to the route plan. Risk budget occupancy indicators are generated based on incrementally updated route plans and support replanning plans and are linked to the budget ledger for output. The risk budget occupancy index is generated item by item according to the risk budget item set, including airspace unavailability risk budget occupancy, link discontinuity risk budget occupancy, and UAV health status risk budget occupancy. It is collected into a risk budget occupancy index sequence according to the alignment time index table. It is obtained by multiplying the trigger flag in the risk amplification item by the corresponding occupancy weight and then summing them. The number of trigger flags is determined by the number of flags that are triggered among the airspace unavailability flag, low confidence flag, and abnormal flag.

[0028] In this embodiment, the incremental reachable corridor consists of a set of reachable corridor geometric domains, a set of reachable corridor edges, and a set of reachable corridor edge weights. The set of reachable corridor geometric domains is obtained by expanding the centerline of the flight segments to be executed in the platform's unified coordinate system and clipping it with the corridor width parameter. The set of reachable corridor edges is generated by the adjacency relationship of flight segments and the connectivity relationship of the airspace grid. The set of reachable corridor edge weights is obtained by reparameterizing the edge cost. The corridor width parameter is limited by the minimum corridor width and the maximum corridor width, and is obtained by proportionally mapping the remaining available amount of the risk budget item set between the minimum corridor width and the maximum corridor width.

[0029] In this embodiment, the generation of the reflow update specifically includes: Read the shadow price status, risk budget occupancy indicator sequence and budget ledger, collect the execution feedback of incrementally updated route plans and support replanning plans, and generate an execution feedback set; The execution feedback set includes records of mission execution status, takeoff and landing execution status, charging execution status, maintenance execution status, communication window execution status, route tracking status, and corresponding alignment time index table bindings. Based on the execution feedback set and the task-UAV-support resource joint scheduling scheme, a plan-execution deviation record set is generated, which is then grouped into a set of return flow statistics entries according to the support site identifier, task cluster identifier, and alignment time index table. The plan-execution deviation record set includes task delay deviation records, resource usage deviation records, route deviation deviation records, and communication interruption deviation records. A risk repatriation trigger flag set is generated based on the risk budget occupancy indicator sequence, the set of repatriation statistics items, and the set of risk budget items in the budget ledger; The risk backflow trigger mark set is limited to three categories: risk over-occupancy trigger mark, risk sustained tension trigger mark, and risk sudden increase trigger mark. The risk over-occupancy trigger mark is determined by the state that the occupancy of the corresponding risk budget item in the risk budget occupancy index sequence exceeds the budget limit. The risk sustained tension trigger mark is determined by the continuous state of tension or extreme tension in the risk tension level corresponding to the corridor parameter index record. The risk sudden increase trigger mark is determined by the occupancy growth of the risk budget occupancy index sequence in the adjacent alignment time index table exceeding the preset growth threshold. A set of resource repatriation trigger markers is generated based on the set of repatriation statistics entries, shadow price status, and budget ledger. The resource backflow trigger flag set is limited to three categories: resource overload trigger flag, queuing delay trigger flag, and replanning jitter trigger flag. The resource overload trigger flag is determined when the occupancy of the takeoff and landing slot capacity item, charging slot capacity item, maintenance slot capacity item, and communication window capacity item corresponding to the resource budget item set exceeds the capacity limit. The queuing delay trigger flag is determined when the occupancy of the queuing delay budget item in the time budget item set exceeds the budget limit. The replanning jitter trigger flag is determined when the number of change items of the incremental updated route plan and the support replanning plan output by the adjacent rolling cycle in the alignment time index table exceeds the threshold corresponding to the incremental replanning trigger frequency budget item in the update density budget item set. Under the constraints of the risk backflow trigger flag set and the resource backflow trigger flag set, execute the backflow action adjudication to generate a backflow action set; The action types of the backflow action set are limited to four categories: adjustment, adjustment, freeze, and rollback. Adjustment is used to increase the constraint strength of the risk budget item set, or to increase the value of the item in the shadow price field item set that is hit by the targeted update index set, or to increase the sensitivity of the threshold item in the rule threshold index table that is associated with the conflict interpretation package. Adjustment is used to reduce the value of the low priority item in the shadow price field item set that corresponds to the mutual exclusion conflict, or to relax the constraint strength of the risk budget item set in the budget ledger that is not hit by the risk backflow trigger mark set. Freezing is used to keep the item values ​​of the rule threshold index table, shadow price field, and budget ledger unchanged within a preset freeze window and only accumulate the backflow statistical item set. Rollback is used to write back to the rule threshold index table snapshot, shadow price status snapshot, and budget ledger snapshot retained in the previous rolling cycle and trigger the downgrade feasibility strategy of the guarantee replanning scheme. Based on the set of backflow actions, the rule threshold index table, shadow price field and budget ledger are updated and a set of updated configurations is generated and written back to the UAV autonomous support platform.

[0030] Example 1: To verify the feasibility of this invention in practice, it was applied to a low-altitude emergency support scenario in a coastal city. In this scenario, drones need to continuously perform tasks such as dike inspection, temporary material delivery, and communication relay under the influence of residual severe convective weather on the periphery of a typhoon. Simultaneously, they are constrained by temporary airspace control, temporary take-off and landing point capacity, limited charging and maintenance resources, and fluctuating link quality. Existing practices typically handle rule determination, scheduling, and route updates separately. When encountering temporary airspace closures or station queuing, frequent plan overturning, inconsistencies between routes and support arrangements, and increased route jitter often occur, ultimately leading to difficulties in ensuring mission continuity and distorted risk control. This is precisely the problem that this invention aims to solve.

[0031] In this scenario, the system first aggregates information on task queues, UAV status, support resource status, airspace constraints, communication status, and environmental conditions, and unifies spatiotemporal benchmarks to form a standardized platform state set that can be directly used for decision-making. Then, rule entries in the rule base are bound to the standardized platform state set to generate a rule threshold index table. Rule hit determination, constraint violation magnitude generation, and evidence credibility generation are then performed, and an evidence commitment vector is encapsulated, ensuring a clear and traceable record of each rule's hit, deviation, and credibility. Next, a budget ledger is generated based on the standardized platform state set, uniformly registering resource budgets (takeoff and landing, charging, maintenance, and communication windows), risk budgets (airspace, link continuity, and health status), and time and update density budgets. During budget gating consistency maintenance, the budget gating entry mapping table maps the rule threshold index table and evidence commitment vector into the budget ledger, completing rule activation and disabling determinations, and generating a conflict candidate set and a conflict adjudication result set. This further forms a conflict explanation package, a minimum correction rule set, and a feasibility label, ultimately encapsulating a feasible domain proof fragment. In this way, the conflict caused by airspace blockade and resource shortages is no longer just a conclusion that is "infeasible", but is interpreted as a traceable cause of conflict and an actionable minimum repair suggestion, providing a unified constraint basis for subsequent joint scheduling and replanning.

[0032] During the joint scheduling phase, an evidence-coupled shadow price field is constructed using evidence commitment vectors, budget ledgers, and feasible region proof fragments. A set of shadow price entries is generated using resource type, site, time period, and task cluster as index closures. Conflict budget association tables and conflict cause categories trigger targeted updates to shadow price entries, enabling scheduling optimization to target bottleneck resources and high-risk segments. Subsequently, Lagrange decomposition is performed to solve task matching, resource scheduling, and communication window scheduling. In the interpretation-guided large neighborhood search and repair process, the rules hitting the minimum conflict set are back-indexed to the set of disruptive objects. The repair priority is determined according to the severity of the violation, sequentially completing task reallocation, site switching, time window rescheduling, and communication window rescheduling. Simultaneously, feasibility screening is performed under budget gating constraints, outputting the joint scheduling scheme for tasks, drones, and support resources, along with the shadow price status. Entering the incremental replanning phase, risk tension classification results are generated based on the remaining available quantity of the risk budget item set and the risk budget occupancy index sequence, forming corridor parameter index records, and generating incremental reachable corridors with risk budget constraints accordingly, completing edge cost reparameterization; during the D*Lite incremental replanning process, an impact domain object set is generated, restricting route updates to the reachable corridor edge set covered by the impact domain object set, while candidate support station switching and candidate window adjustment are filtered under budget constraints and aligned with the route plan, resulting in incrementally updated route plans, support replanning plans, and risk budget occupancy indicators.

[0033] During operation, the platform continuously collects execution feedback and generates a plan-execution deviation record. Combining this with the risk budget occupancy indicator sequence and the budget ledger, it generates risk backflow trigger flag sets and resource backflow trigger flag sets. It then adjudicates four types of backflow actions: upward adjustment, downward adjustment, freezing, and rollback. The updated configuration is written back to the rule threshold index table, shadow price field, and budget ledger, ensuring consistent operation in the next rolling cycle. Through this closed-loop process, temporary airspace changes, link quality fluctuations, and site congestion in the scenario are transformed into interpretable conflict explanation packages and executable minimum repair suggestions. Under the constraints of shadow prices and risk budget corridors, it achieves coordinated adjustments to flight routes and support arrangements, avoiding the jitter caused by frequent plan overturning and disorderly route changes, while keeping risk budget occupancy within a controllable range. This verifies the feasibility and stability of the invention under real-world constraint coupling environments.

[0034] Table 1. Overall Comparison of Intelligent Decision-Making Effects of Unmanned Aerial Vehicle Autonomous Support Platforms

[0035] From the task results, the on-time completion rate of the proposed solution increased from 86.70% to 92.40%, the task interruption rate decreased from 7.60% to 4.10%, and the average task delay decreased from 6.80 minutes to 4.90 minutes. Compared with existing solutions, the improvements mainly come from the feasible domain proof fragments formed by budget gating consistency maintenance: the conflict interpretation package and the minimum correction rule set structure the coupling relationship between rule conflicts, resource budgets, and risk budgets in advance, reducing infeasibility situations exposed only during the execution phase, and reducing the proportion of infeasibility recalculations from 12.30% to 5.70%, thereby reducing the "recalculation-failure-recalculation" cycle from the source.

[0036] In terms of stability and interoperability, the average number of replanning triggers in this invention decreased from 5.20 times / hour to 3.40 times / hour, and the route jitter index decreased from 18.60 to 11.20. Simultaneously, the consistency rate of interoperability improved from 81.50% to 90.80%. This is directly related to the incremental reachability of the corridor and the set of objects in the influence domain within the risk budget constraint: the corridor parameter index records map the risk tension level into constraint entry points such as corridor width and edge cost amplification. D*Lite performs incremental updates within the coverage area of ​​the set of objects in the influence domain, avoiding large-scale reassurance rearrangement caused by global route changes; and ensures that the replanning scheme and the route scheme are aligned and screened within the same influence domain, reducing inconsistencies such as "the route has changed but the station window has not kept up."

[0037] From the perspective of interpretable and controllable evolution, the median conflict resolution time of the proposed solution decreased from 1.70 seconds to 0.90 seconds, the minimum correction rule set size decreased from 3.60 rules / time to 2.10 rules / time, the risk budget over-occupancy event decreased from 2.40 times / day to 1.10 times / day, and the median number of stable cycles after backflow update decreased from 5.00 to 3.00. This is because the shadow price status uses a four-element index closure of resource type-site-time period-task cluster and is triggered by a conflict budget association table for targeted updates, making scheduling and patching more focused on bottleneck items. Simultaneously, the closed four-action backflow update transforms execution feedback into controllable adjustments to the rule threshold index table, shadow price field, and budget ledger, reducing frequent "over-corrections" and thus enabling faster entry into a stable operating range.

[0038] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An intelligent decision-making method based on an unmanned aerial vehicle autonomous support platform, characterized in that, Includes the following steps: Acquire task queues, UAV status sets, support resource status sets, airspace constraint information sets, communication status sets, and environmental information sets; unify spatiotemporal benchmarks and generate standardized platform status sets. Based on the standardized platform state set and rule entry set, a rule threshold index table is generated, and rule hit determination, constraint violation magnitude generation and evidence credibility generation are performed, and the evidence commitment vector is encapsulated. A budget ledger is generated based on a standardized platform state set, and budget gating consistency is maintained by combining it with a rule threshold index table and an evidence commitment vector to form a feasible domain proof fragment. Based on the evidence commitment vector, budget ledger and feasible region proof fragment, construct the evidence-coupled shadow price field and perform Lagrange decomposition. Combined with the interpretation-oriented large neighborhood search repair generation task-UAV-support resource joint scheduling scheme, output the shadow price state. Based on the joint scheduling scheme of mission-UAV-support resources, shadow price status, budget ledger and feasible domain proof fragment, an incremental reachable corridor with risk budget constraints is generated. D*Lite incremental replanning is performed to obtain incrementally updated route scheme, support replanning scheme and risk budget occupancy index. Based on the shadow price status, risk budget occupancy indicators, and execution feedback, the rule threshold index table, shadow price field, and budget ledger are updated by performing a set of closed actions such as adjustment, reduction, freeze, and rollback. 2.The intelligent decision-making method based on the unmanned aerial vehicle autonomous support platform of claim 1, wherein, The generation of the standardized platform state set specifically includes: Acquire task queues, drone status sets, support resource status sets, airspace constraint information sets, communication status sets, and environmental information sets; extract and organize fields to generate a set of original platform elements. Perform unified time base alignment on the timestamps of the platform's original element set, and select the platform clock as the unified time base to generate a set of time mapping parameters; A unified sampling time sequence is generated based on a unified time benchmark. Time synchronization resampling is performed on the original set of elements of the platform to construct an aligned time index table and obtain a synchronized set of elements of the platform. The spatial reference is unified for the set of elements of the synchronization platform, and a set of unified coordinate system definition information and spatial mapping parameters for the platform is constructed. Under the unified coordinate system of the platform, the spatial constraint information set is rasterized and indexed to generate a spatial raster occupancy table and a spatial availability index table. The combined index is then encapsulated to obtain a spatial constraint index set, which is written into the synchronous platform element set. Under the unified coordinate system of the platform, the site-level resource aggregation and occupancy coding are performed on the set of guaranteed resource statuses to generate a set of guaranteed resource occupancy vectors and a set of guaranteed resource capacity indexes, which are then written back to the set of elements of the synchronous platform to form the basic resource status. The task queue is aggregated into task clusters and the constraint fields are fixed. A task cluster index table and a task time window index table are generated. A composite index is encapsulated to obtain a structured task index set and written into the synchronization platform element set. The consistency of the platform element set, alignment time index table, and platform unified coordinate system definition information is checked and conflicts are eliminated to generate a standardized platform state set.

3. The intelligent decision-making method based on an autonomous support platform for unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The generation of the evidence commitment vector specifically includes: Load a set of rule entries from the rule base of the UAV autonomous support platform, and construct a rule entry specification table based on the standardized platform status set; A rule threshold index table is generated for each rule based on the rule entry specification table; Based on the rule threshold index table, rule hit determination is performed on the standardized platform state set, and the rule hit determination result set is output. Constraint violation magnitude is generated based on the rule threshold index table and the rule hit judgment result set, and bound to the rule identifier, object identifier and alignment time index to form a violation magnitude index record; A set of evidence credibility entries is generated based on a standardized platform state set, and then bound to rule identifiers, object identifiers, and alignment time indexes to form credibility index records; Perform credibility aggregation and gate tag generation on the evidence credibility item set to obtain rule-level evidence credibility and evidence gate tag set, and write them into the credibility index record; The rule hit result set, violation magnitude index record, and credibility index record are indexed and encapsulated to generate an evidence commitment vector.

4. The intelligent decision-making method based on an autonomous support platform for unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The generation of the feasible region proof fragment specifically includes: Read the standardized platform status set, perform resource aggregation, window aggregation, and status aggregation, and generate a candidate set of budget ledger entries; Generate a budget ledger based on the candidate set of budget ledger entries; Map the rule threshold index table and the evidence commitment vector to the budget ledger to construct a budget gating entry mapping table; Based on the budget gating entry mapping table, perform rule enable and rule disable determination on the rule threshold index table, output the rule enable flag set and rule disable flag set and write them back to the budget gating entry mapping table; Under the constraint of the rule enable tag set, perform conflict candidate generation on the rule hit identifier set and the violation amplitude index record in the evidence commitment vector, and output the conflict candidate set; Perform conflict resolution priority determination on the conflict candidate set, obtain the conflict resolution result set and generate a conflict interpretation package; A minimum set of correction rules is generated based on the set of conflict resolution results and the budget gating entry mapping table, and together with the conflict interpretation package, a feasibility label is generated. The conflict interpretation package, the minimum correction rule set, and the feasibility label are indexed and encapsulated to generate a feasible domain proof fragment.

5. The intelligent decision-making method based on an autonomous support platform for unmanned aerial vehicles (UAVs) according to claim 4, characterized in that, The conflict adjudication priority determination adopts a closed priority sequence for sequential adjudication. The set of rules for airspace objects takes precedence over the set of rules for UAV objects, the set of rules for UAV objects takes precedence over the set of rules for support resources objects, the set of rules for support resources objects takes precedence over the set of rules for communication objects, and the set of rules for communication objects takes precedence over the set of rules for task objects. Within the same priority, the adjudication is based on the violation magnitude index, from largest to smallest.

6. The intelligent decision-making method based on an unmanned aerial vehicle (UAV) autonomous support platform according to claim 4, characterized in that, The conflict interpretation package includes a set of conflict minimum set identifiers, a set of conflict cause categories, a conflict rule association table, a conflict budget association table, a summary of conflict violation magnitudes, and conflict evidence citation records.

7. The intelligent decision-making method based on an unmanned aerial vehicle (UAV) autonomous support platform according to claim 1, characterized in that, The generation of the shadow price state specifically includes: Read the evidence commitment vector, budget ledger and feasible region proof fragment, and generate a scheduling index structure based on the task cluster identifier, guarantee site identifier and alignment time index table in the normalized platform state set; Construct evidence-coupled shadow price fields based on scheduling index structures and generate a set of shadow price field entries; Map the conflict budget association table and the conflict cause category set in the feasible region proof fragment to the shadow price field entry set to form a targeted update index set; Based on the budget ledger and the set of shadow price field entries, construct the input set for the Lagrange decomposition, forming a set of decomposition subproblems; Iterative solutions are performed on the set of subproblems to generate a set of candidate joint scheduling solutions; Perform a targeted update on the shadow price field entry set based on the targeted update index set and output the shadow price status, including the updated shadow price field entry set and its quad octet closure. Based on the conflict interpretation package and the minimum correction rule set, an interpretation-guided large neighborhood search is performed on the candidate joint scheduling solution set to obtain feasible repair results; The feasible repair results and candidate joint scheduling solutions are consistent and normalized, and then merged according to the task identifier and alignment time index table to generate a joint scheduling scheme for task-UAV-support resources, which is then output in conjunction with the shadow price status.

8. The intelligent decision-making method based on an unmanned aerial vehicle (UAV) autonomous support platform according to claim 1, characterized in that, The generation of the route plan, support replanning plan, and risk budget occupancy indicators specifically includes: Read the joint scheduling scheme of task-drone-support resources, expand the task allocation results, take-off and landing scheduling results, charging scheduling results, maintenance scheduling results and communication window scheduling results according to the alignment time index table, and generate a replanning input set; A risk budget constraint parameter set is generated based on the set of risk budget entries in the budget ledger and the set of conflict cause categories in the feasible region proof fragment; Based on the risk budget constraint parameter set and the airspace grid occupancy table, airspace availability index table and link quality observation corresponding to the communication link identifier in the normalized platform state set, an incremental reachable corridor for risk budget constraints is generated. Perform edge cost reparameterization on incremental reachable corridors to form a set of reachable corridor edge weights; Based on the incremental reachable corridor, the set of reachable corridor edge weights, and the replanning input set, D*Lite incremental replanning is performed to generate a set of influence domain objects; The route plan and support replanning plan are updated incrementally based on the set of objects in the influence domain and the set of reachable corridor edges. Based on incrementally updated route plans and support replanning plans, risk budget occupancy indicators are generated and linked to the budget ledger for output.

9. The intelligent decision-making method based on an unmanned aerial vehicle (UAV) autonomous support platform according to claim 8, characterized in that, The incremental reachable corridor consists of a set of reachable corridor geometric domains, a set of reachable corridor edges, and a set of reachable corridor edge weights. The set of reachable corridor geometric domains is obtained by expanding the centerline of the flight segments to be executed in the platform's unified coordinate system and clipping it with the corridor width parameter. The set of reachable corridor edges is generated by the adjacency relationship of flight segments and the connectivity relationship of the airspace grid. The set of reachable corridor edge weights is obtained by reparameterizing the edge cost. The corridor width parameter is limited by the minimum corridor width and the maximum corridor width, and is obtained by proportionally mapping the remaining available amount of the risk budget item set between the minimum corridor width and the maximum corridor width.

10. The intelligent decision-making method based on an unmanned aerial vehicle (UAV) autonomous support platform according to claim 1, characterized in that, The generation of the reflow update specifically includes: Read the shadow price status, risk budget occupancy indicator sequence and budget ledger, collect the execution feedback of incrementally updated route plans and support replanning plans, and generate an execution feedback set; Based on the execution feedback set and the task-UAV-support resource joint scheduling scheme, a plan-execution deviation record set is generated, which is then grouped into a set of return flow statistics entries according to the support site identifier, task cluster identifier, and alignment time index table. A risk repatriation trigger flag set is generated based on the risk budget occupancy indicator sequence, the set of repatriation statistics items, and the set of risk budget items in the budget ledger; A set of resource repatriation trigger markers is generated based on the set of repatriation statistics entries, shadow price status, and budget ledger. Under the constraints of the risk backflow trigger flag set and the resource backflow trigger flag set, execute the backflow action adjudication to generate a backflow action set; Based on the set of backflow actions, the rule threshold index table, shadow price field and budget ledger are updated and a set of updated configurations is generated and written back to the UAV autonomous support platform.

Images