An unmanned aerial vehicle real-time risk control evaluation method and system based on data efficient caching and parameter analysis

Abstract

This invention discloses a real-time risk control assessment method and system for unmanned aerial vehicles (UAVs) based on efficient data caching and parameter parsing. The method includes: analyzing risk control scenarios such as pre-flight, in-flight, and airspace route conflicts; determining assessment items and configuring trigger conditions, dependent parameters, and data reading path expressions; and completing pre-compilation of the parsing engine. It also involves receiving raw UAV data, determining risk control scenarios, acquiring external risk data, and calculating flight-related data in real time; aggregating the two types of data to generate standardized objects and composite cache keys, and writing them to the cache; triggering assessment items through semantic analysis; and batch loading standardized objects from the cache based on the composite cache keys; extracting parameter values ​​through the expression parsing engine and sending them to the rule engine to complete real-time risk control assessment, efficiently adapting to UAV risk control requirements. This invention achieves millisecond-level real-time risk control response capabilities while ensuring the system possesses high flexibility, high configurability, and high operational reliability to adapt to low-altitude operation characteristics.

Description

Technical Field

[0001] This invention relates to the field of drone risk control technology, specifically to a real-time risk control assessment method and system for drones based on efficient data caching and parameter parsing. Background Technology

[0002] In today's era of integrated digitalization and low-altitude economy, real-time risk control for drones has become a core element in ensuring flight safety and mitigating operational risks. Whether it's route compliance verification in drone logistics delivery, equipment safety monitoring in power line inspections, or airspace intrusion warnings in security surveillance, all require comprehensive analysis and decision-making from massive amounts of streaming data within a very short timeframe. This massive streaming data includes flight attitude data, airspace environment data, equipment status data, and mission payload data. However, the performance bottleneck of such systems often lies not in the calculation logic of the risk control rules themselves, but in the unique characteristics and efficiency limitations of the data storage and retrieval phase in drone scenarios.

[0003] Traditional drone risk control systems, in terms of data processing, are limited by the low-altitude operating environment, the characteristics of multi-source data, and real-time requirements, and thus have the following inherent defects and technical difficulties:

[0004] First, data access I / O bottlenecks are prominent and scenario adaptability is poor: complex drone risk control rules rely on multi-dimensional heterogeneous data (such as real-time flight parameters, airspace control data, equipment health status, historical operation records, environmental perception data, etc.). Traditional architectures typically initiate independent data query requests for each parameter. This model not only creates enormous network I / O and local storage I / O pressure, but also suffers from limited bandwidth and susceptibility to interference in low-altitude communication links. Furthermore, for rules requiring N parameters, the overall response time is roughly the sum of the delays of N queries. In scenarios with high concurrency, dispersed data sources, and unstable links, it is impossible to meet the stringent requirements of millisecond-level risk control decisions for drones, easily leading to missed flight risk assessments or delayed responses.

[0005] Second, the problem of redundant data transmission and resource waste is exacerbated: In multi-mission scenarios of UAVs, multiple risk control rules often rely on different fields of the same original data. For example, flight speed data is used for both route compliance verification and equipment overload warning. The independent query mode leads to the repeated transmission of the same basic data in the air-to-ground link. However, low-altitude communication bandwidth is usually limited. Redundant transmission not only seriously wastes valuable bandwidth resources, but also increases the computational overhead of data serialization / deserialization, resulting in the waste of limited CPU and computing power resources on the airborne end, which in turn affects the UAV's endurance and the efficiency of core mission execution.

[0006] Third, the contradiction between system flexibility and flight scenario adaptability is difficult to reconcile: While hard-coding the UAV data location and acquisition logic into the business code allows for caching optimization for specific UAV models and operational scenarios, it leads to extreme system rigidity. UAV applications are diverse, UAV models vary greatly, and data sources and structures change frequently. Hard-coding requires repeated code modifications, testing, and deployment, failing to achieve the "configurable" flexibility expected by business personnel. Conversely, while a fully configurable interpreted execution approach ensures scenario adaptability flexibility, it introduces performance overhead from interpreted execution, making it difficult to simultaneously achieve the high-performance goal of real-time UAV risk control.

[0007] Fourth, caching strategies are mismatched with the multi-dimensional data query requirements of drones: Drone risk control data is characterized by its multi-dimensional, multi-layered, and multi-structured nature. Designing cache key and value structures that can both meet the efficient access needs of airborne equipment and support flexible multi-dimensional queries is a key technical challenge. Traditional, simple caching strategies struggle to handle the complex query scenarios of drone risk control, easily leading to low cache hit rates and inefficient data access.

[0008] Fifth, the reliability and fault tolerance of the system face severe challenges in low-altitude environments: the operating environment of UAVs is complex, and problems such as cache failure, air-to-ground communication interruption, and temporary unavailability of upstream data sources are prone to occur. Traditional systems lack efficient degradation and fault tolerance mechanisms adapted to low-altitude scenarios: for example, when the communication link is interrupted, it is impossible to quickly switch to onboard local cache data, and when the data source fails, there is no backup data acquisition channel, which may lead to the spread of failure of a single data node, triggering a chain of failures in risk control services, such as the inability to verify flight route compliance and the inability to monitor equipment anomalies, thereby affecting the flight safety of UAVs and indicating insufficient system resilience.

[0009] Therefore, there is an urgent need in this field for an innovative data processing architecture and method that can specifically address the series of problems mentioned above in UAV risk control scenarios. While ensuring the system's high configurability and flexibility across multiple UAV models and operational scenarios, it should break through the data access performance bottleneck in low-altitude environments, achieve an order-of-magnitude performance improvement, and ensure the overall reliability of UAV flight and operations.

[0010] In the field of general risk control such as financial transactions and e-commerce risk control, existing technologies have adopted batch reading and centralized caching to alleviate the data I / O bottleneck in high-frequency scenarios. However, their technical paradigms and optimization strategies cannot be directly transferred to the real-time risk control scenario of drones. The two have essential differences in data characteristics, real-time constraints, communication environment, fault tolerance mechanism and scenario dynamics, as follows: (1) Data in the field of general risk control is mainly structured and stable business data, while drone risk control involves multi-source heterogeneous, high-frequency dynamic, and spatiotemporally coupled data such as airborne sensing, airspace management, meteorological environment, and terrain obstacles. Traditional reading technologies lack the ability to integrate and adapt to airspace-specific data types and multimodal heterogeneous data, making it difficult to support the unified acquisition of all-dimensional risk elements of drones. (2) Drone flight has characteristics such as high-speed maneuverability, complex airspace environment, and sudden risk events. Key information such as airspace conflicts and meteorological changes need to meet sub-second or even millisecond-level response delays. Delays can easily lead to safety accidents such as collisions and loss of control. The optimization goal of general risk control is mainly business availability. Its data reading latency and performance limit cannot meet the extreme safety and real-time requirements of low-altitude flight. (3) UAVs rely on air-to-ground wireless, satellite or mobile networks to complete data interaction. The operation area often has problems such as weak signal, obstruction interference and unstable link, which easily leads to data packet loss and transmission interruption. Traditional data reading technology is based on stable wired network design. The cache synchronization mechanism and weak network adaptation capability are insufficient, making it difficult to support continuous and reliable data access in long-distance, outdoor non-line-of-sight scenarios. (4) When the cache fails or the data is abnormal, general risk control can use historical backtracking, manual review, downgrade shutdown and other methods to make up for it. The safety risks and accident consequences are controllable. UAV risk control needs to take into account the bottom line of flight safety and the continuity of operation. It requires multi-level cache to prevent avalanche, provide degraded services with lossy usability when local data fails, and support dedicated handling logic such as emergency landing and route correction. The traditional single cache architecture and general downgrade strategy cannot adapt to the low-altitude safety resilience requirements, which can easily lead to the failure of risk control decisions and cause flight accidents. (5) General risk control is mostly a closed-loop scenario with fixed triggers and fixed data dimensions, and the data reading path and rules are relatively fixed. Drone risk control covers multiple scenarios throughout the entire life cycle, such as pre-flight compliance verification, in-flight dynamic monitoring, and real-time airspace conflict judgment. The evaluation items, parameter sets, and reading dimensions at different stages are significantly different, requiring rapid scenario switching and flexible configuration. Traditional fixed configuration logic is difficult to adapt to dynamic switching of multiple scenarios, and data reading redundancy or missing data is likely to occur.

[0011] In summary, batch reading and I / O optimization technologies in the general risk control field have not been adapted to the architecture and innovated in terms of mechanism for the characteristics of low-altitude operation of UAVs. They have fundamental defects in data processing, real-time response, link adaptation, fault tolerance and degradation and dynamic adaptation to scenarios, and cannot be directly applied to UAV real-time risk control systems. Summary of the Invention

[0012] The purpose of this invention is to provide a real-time risk control assessment method and system for drones based on efficient data caching and parameter parsing. By constructing a core technical paradigm of unified preprocessing, batch reading, and dynamic expression parsing, it accurately solves the problems of data access I / O bottlenecks, air-to-ground link transmission constraints, and dynamic adaptation to multiple scenarios in drone risk control scenarios, achieving millisecond-level real-time risk control decision response. This enables the system to have high flexibility, configurability, and operational reliability for low-altitude operations.

[0013] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows:

[0014] A real-time risk control assessment method for drones based on efficient data caching and parameter parsing, the method comprising:

[0015] S1 analyzes multiple risk control scenarios for drones, including pre-flight, in-flight, and airspace route conflict detection, and obtains several risk control evaluation items. For each risk control evaluation item, it configures the corresponding trigger conditions, dependent risk control parameters, and data reading path expressions for each risk control parameter, and completes the pre-compilation of the expression parsing engine.

[0016] S2 receives raw flight-related data sent by the drone, determines the risk control scenario corresponding to the data, calls external services to obtain external flight risk data for all risk assessment items under the risk control scenario based on the determination result, and obtains drone flight-related data by performing real-time calculations on the raw flight-related data.

[0017] S3 aggregates external flight risk data and drone flight-related data, generates corresponding standardized objects and globally unique composite cache keys, and writes them into the cache.

[0018] S4. Based on the evaluation items set for the fields in the original flight data in the risk control system, automatically trigger one or more risk control evaluation items; for each triggered risk control evaluation item, load the list of all risk control parameters and data reading path expressions that the risk control evaluation item depends on, and generate a composite cache key based on the semantic analysis results; based on the composite cache key, load the corresponding standardized objects from the cache in batches at once.

[0019] S5 uses an expression parsing engine to dynamically extract risk control parameter values ​​from standardized objects, and then sends the extracted parameter values ​​into the rule engine for real-time risk control assessment.

[0020] Furthermore, in step S1, the process of analyzing multiple risk control scenarios for the drone, including pre-flight, in-flight, and airspace route conflict detection, to obtain several risk control assessment items includes:

[0021] S11, based on the risk transmission characteristics of the entire drone flight lifecycle, deconstructs risk control scenarios in layers, including pre-flight preparation, dynamic in-flight management, and airspace route conflict detection. The specific deconstruction process includes:

[0022] For the pre-flight scenario, based on the airworthiness of the UAV itself and the compliance risks of the mission planning, multiple risk control assessment items are obtained, including UAV equipment status verification items, flight plan airspace permission verification items, payload adaptability assessment items, and take-off and landing point environmental safety assessment items.

[0023] For in-flight scenarios, considering the dynamic constraints and environmental interference risks of UAV dynamic flight, multiple risk control assessment items are obtained, including real-time flight attitude compliance assessment item, power system health status monitoring item, flight path deviation assessment item, real-time weather environment adaptability assessment item, and battery remaining range and mission range matching degree assessment item.

[0024] For the airspace route conflict detection scenario, the collaborative management and control requirements of the corresponding mission airspace are retrieved and decomposed into multiple risk control assessment items, including the multi-aircraft flight interval assessment item in the same airspace, the collision risk assessment item between the flight path and static obstacles, the conflict prediction item between the flight path and dynamic obstacles, and the compliance assessment item for cross-resolution airspace transition between layers.

[0025] S12, for each risk control assessment item obtained from the breakdown, clarify its applicable drone flight stage, risk level weight, and trigger priority, and extract the drone-specific risk control parameters required for the assessment item to form a parameter set:

[0026] S13 combines the dynamic constraints of UAV flight with airspace control rules to extract basic thresholds, including the dynamic parameter thresholds corresponding to the UAV model and the airspace management specification thresholds. Then, it supplements dynamic context trigger constraints by combining flight stage and mission type. It uses a composite logic of basic thresholds and dynamic context trigger constraints to configure scenario-based trigger conditions for each risk control assessment item.

[0027] S14. For each risk control assessment item's dependent parameter set, and considering the data source type of the parameters, a standardized data reading path expression is customized. Data source types include real-time acquisition from UAV onboard sensors, pre-configuration from ground stations, airspace management platform interface calls, and third-party meteorological service pushes. The data reading path expression is adapted to the parameter characteristics. For dynamic parameters that are frequently updated during UAV flight, a timestamp matching identifier is added to the path expression; for static parameters in the airspace environment, a cache validity period identifier is added; for hierarchical parameters that cross resolution airspaces, a resolution adaptation identifier is added. Finally, a syntax parser performs validity checks on all customized path expressions, eliminating invalid expressions that do not conform to the UAV data transmission format.

[0028] S15: Input the validated path expression and the trigger condition logic of each evaluation item into the expression parsing engine for pre-compilation processing.

[0029] Furthermore, the pre-compilation process of the expression parsing engine is achieved by converting the string-form expression into intermediate code or a pre-initialized context that can be efficiently executed within the expression parsing engine. The pre-compilation result is cached in the local memory of the application server. During real-time parsing, the expression parsing engine directly calls the pre-compiled result. The specific pre-compilation process includes:

[0030] The validated path expressions, including data source identifiers and drone parameter characteristic adaptation identifiers, along with the scenario-based triggering condition logic for each risk control assessment item, are input into the expression parsing engine for pre-compilation processing to generate a directly executable syntax tree.

[0031] For the path expressions corresponding to high-frequency dynamic parameters during drone flight, including real-time attitude, position, and remaining battery power, the lexical analysis rules of the parsing engine are optimized, and a mapping cache between parameter names and data transmission fields is established; common path expressions of multiple evaluation items under the same risk control scenario are merged and compiled to generate a shared syntax tree fragment.

[0032] During the pre-compilation process, multiple UAV parameter type verification rules are embedded, including coordinate format verification, flight speed unit verification, BeiDou grid code validity verification, and battery endurance parameter range verification. This results in a pre-compiled parsing engine instance for subsequent dynamic parameter extraction.

[0033] Furthermore, the process of merging and compiling common path expressions of multiple evaluation items under the same risk control scenario to generate a shared syntax tree fragment includes the following steps:

[0034] For each risk control scenario, each risk control assessment item is further subdivided according to the type of assessment purpose, and divided into multiple assessment-specific clusters;

[0035] Select each evaluation cluster one by one, extract its parameter set and construct a parameter-evaluation item mapping table, and clarify the type and data source type of each parameter; use the following formula to calculate the parameter sharing rate between each evaluation item in the evaluation cluster: Parameter sharing rate = Number of shared parameters / Total number of parameters between the two evaluation items;

[0036] Considering the dense nature of drone risk control parameters, a sharing rate threshold is set, and evaluation item combinations with a sharing rate greater than the sharing rate threshold and shared parameters being core parameters are retained; core parameters refer to key parameters that directly affect risk control decisions.

[0037] Extract the path expressions corresponding to each evaluation item combination, and verify the compatibility of syntax structure, shared parameter adaptation identifier, and data type respectively. If there are completely identical expressions, they are directly marked as mergeable. If the expressions are not completely identical but compatible, they are merged through shared syntax tree fragments and branch parsing, and finally multiple mergeable compilation evaluation item groups are formed. The risk control evaluation items in each mergeable compilation evaluation item group do not exceed the preset evaluation item number threshold.

[0038] For each mergeable compilation evaluation item group, perform pre-compilation simulation to verify whether the merged syntax tree fragment can be parsed normally and without logical conflicts; if the merged expression of any evaluation item causes the new branch to exceed the preset branch threshold, remove it from the mergeable compilation evaluation item group and compile it separately.

[0039] Establish a mapping between mergeable evaluation item groups and shared syntax trees, and store it in the pre-compiled cache manager.

[0040] Step S2 further includes:

[0041] S21, parse the raw flight-related data sent by the drone, determine the risk control scenario corresponding to the data, and based on the determined risk control scenario, identify and extract the object type and unique identifier of the data packet;

[0042] S22. Generate a globally unique composite cache key based on the identified object type and unique identifier. The format of the composite cache key is {object type: unique identifier}.

[0043] S23, based on the pre-configured data source rules corresponding to the risk control scenario, calls external services in parallel to obtain external flight risk data, and performs real-time calculations on the flight-related raw data to obtain drone flight-related data.

[0044] Furthermore, in step S23, when the flight-related raw data contains a flight plan, the risk control scenario is determined to be before flight, the object type of the data packet is identified as FlightPlan, the plan number is extracted as a unique identifier, and a globally unique composite cache key {FlightPlan: plan number} is generated; then, multiple external services, including airspace compliance service, meteorological data service, signal coverage service, and air traffic conflict detection service, are called in parallel to obtain external flight risk data; at the same time, the inherent parameters of the flight plan are extracted from the flight-related raw data to obtain UAV flight-related data.

[0045] Further, in step S23, when the original flight-related data includes real-time flight status data, the risk control scenario is determined to be in flight, the object type of the data packet is identified as UAVRealtime, the drone number is extracted as a unique identifier, and a globally unique composite cache key {UAVRealtime: drone number} is generated; the built-in spatial computing engine is called to determine whether the drone has entered a temporary restricted area based on the current latitude, longitude and altitude, to estimate the remaining flight time based on the current remaining battery power and average power consumption, and to convert the received signal strength indication into an enumerated type of signal status to obtain drone flight-related data.

[0046] Furthermore, a local cache is established on the application server where the expression parsing engine resides. When data is successfully loaded from the central cache, it is written to the local cache while returning the data to the upper layer. Subsequently, for data requests with the same composite cache key, the local cache is read first.

[0047] Furthermore, a static degradation default value is preset for each data source parameter in the configuration; when batch data loading fails or a field is found to be empty during dynamic parsing, the preset static degradation default value is automatically used to participate in rule calculation and an alarm log is recorded.

[0048] Secondly, this invention discloses a real-time risk control assessment system for unmanned aerial vehicles based on efficient data caching and parameter parsing. The system includes a data preprocessing and caching module, a unified data access layer, an expression parsing engine module, a pre-compiled cache manager, and a configuration management module.

[0049] The data preprocessing and caching module receives flight plan data, real-time flight status data, and raw airspace environment data sent by the UAV. Based on UAV risk control scenario identification rules, it determines the risk control scenario corresponding to the data and extracts the object type and unique identifier of the data packet. According to pre-configured data source rules, it calls external services in parallel, including airspace compliance service, meteorological data service, and air traffic conflict detection service, to obtain external flight risk data. At the same time, it performs real-time calculations on the raw data through the built-in spatial computing engine to obtain UAV flight-related data. Then, by aggregating the external flight risk data and UAV flight-related data, it generates a standardized object containing UAV-specific parameters and generates a globally unique composite cache key based on the object type and unique identifier, and writes the standardized object into the central cache.

[0050] The unified data access layer is used to encapsulate the compound cache key generation logic and multi-level cache query strategy, and provide a unified data access interface for the expression parsing engine module. When a risk control assessment item is triggered, the target compound cache key is generated according to the risk control scenario and semantic analysis results corresponding to the assessment item, and the corresponding standardized objects are loaded in batches from the local memory cache or central cache of the pre-compiled cache manager.

[0051] The expression parsing engine module receives the pre-compiled results passed by the pre-compiled cache manager, directly calls the pre-compiled results during the real-time parsing phase, dynamically extracts the drone risk control parameter values ​​from the standardized objects loaded by the unified data access layer, and automatically performs drone parameter exclusive type verification during the parsing process.

[0052] The pre-compiled cache manager, associated with the configuration management module and the expression parsing engine module, receives the validated path expressions and scenario-based trigger condition logic issued by the configuration management module. It groups these expressions according to UAV risk control scenarios, optimizes lexical analysis rules for expressions corresponding to high-frequency dynamic parameters, performs merged compilation on common expressions, generates a shared syntax tree fragment, and embeds UAV parameter type validation rules to form a pre-compiled result. The pre-compiled result is cached in the application server's local memory, and a mapping relationship is established between mergeable evaluation item groups and the shared syntax tree for real-time access by the expression parsing engine module.

[0053] The configuration management module provides a visual configuration interface and API interface to support the configuration of scenario-based triggering conditions for risk control assessment items, drone-specific risk control parameter sets, multi-source data source rules, path expressions, and static degradation default values ​​of parameters; and synchronizes configuration information to the data preprocessing and caching module and the pre-compiled cache manager in real time.

[0054] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0055] First, this invention simplifies traditional multiple distributed I / O operations into a single cache read by aggregating multi-source heterogeneous data into standardized objects and employing a composite cache key and a one-time batch loading method, effectively reducing low-altitude communication and data access latency. Combined with expression pre-compilation to reduce real-time parsing overhead, this enables the system to meet the high-speed response requirements of real-time UAV risk control.

[0056] Secondly, this invention enables the visual configuration of risk control parameters, triggering conditions, and data source access logic through data reading path expression configuration and scenario-based evaluation item configuration. It can adapt to different drone models, different flight stages, and different airspace control rules without modifying the code, thereby improving the system's adaptability to low-altitude multi-scenario risk control.

[0057] Third, this invention pre-compiles string expressions using an expression parsing engine, converting them into directly executable syntax trees and merging common expressions to reduce redundant parsing overhead. Simultaneously, it embeds UAV-specific parameter validation rules during the parsing process, improving parameter extraction efficiency and data validity verification accuracy.

[0058] Fourth, this invention employs unified data aggregation and multi-level caching to avoid the repeated transmission of the same data in the air-to-ground link, thereby reducing communication bandwidth usage and the consumption of computing resources on the airborne and ground ends, and improving data access and system operation efficiency.

[0059] Fifth, this invention employs a multi-level caching structure combining local caching and central caching, and configures statically degraded default values ​​for parameters. Even in the event of cache failure, data loading failure, or communication anomalies, the risk control assessment can still be continuously executed, improving the system's fault tolerance and operational stability in complex low-altitude environments such as weak networks and interference. Attached Figure Description

[0060] Figure 1 This is a flowchart of the real-time risk control assessment method for drones based on efficient data caching and parameter parsing of the present invention.

[0061] Figure 2 This is a structural diagram of the real-time risk control assessment system for drones based on efficient data caching and parameter parsing of the present invention.

[0062] Figure 3 A detailed business process diagram for data preprocessing and standardized caching;

[0063] Figure 4 A detailed business process diagram for batch data loading and dynamic parameter parsing;

[0064] Figure 5 A diagram illustrating the pre-compilation process of an expression parsing engine;

[0065] Figure 6 An architecture diagram for multi-level caching and degradation fault tolerance strategies;

[0066] Figure 7 This diagram illustrates the performance comparison between traditional data access methods and the method of this invention. Detailed Implementation

[0067] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0068] This invention discloses a real-time risk control assessment method for drones based on efficient data caching and parameter parsing, the method comprising:

[0069] S1 analyzes multiple risk control scenarios for drones, including pre-flight, in-flight, and airspace route conflict detection, and obtains several risk control evaluation items. For each risk control evaluation item, it configures the corresponding trigger conditions, dependent risk control parameters, and data reading path expressions for each risk control parameter, and completes the pre-compilation of the expression parsing engine.

[0070] S2 receives raw flight-related data sent by the drone, determines the risk control scenario corresponding to the data, calls external services to obtain external flight risk data for all risk assessment items under the risk control scenario based on the determination result, and obtains drone flight-related data by performing real-time calculations on the raw flight-related data.

[0071] S3 aggregates external flight risk data and drone flight-related data, generates corresponding standardized objects and globally unique composite cache keys, and writes them into the cache.

[0072] S4 performs contextual semantic analysis on the raw flight-related data sent by the drone, automatically triggering one or more risk control assessment items; for each triggered risk control assessment item, it loads the list of all risk control parameters and data reading path expressions that the risk control assessment item depends on, and generates a composite cache key based on the semantic analysis results; based on the composite cache key, it loads the corresponding standardized objects from the cache in batches at once.

[0073] S5 uses an expression parsing engine to dynamically extract risk control parameter values ​​from standardized objects, and then sends the extracted parameter values ​​into the rule engine for real-time risk control assessment.

[0074] See Figure 1 The real-time risk control assessment method for drones of the present invention includes the following core steps:

[0075] (1) Data preprocessing and standardized caching.

[0076] Step S1 involves analyzing multiple risk control scenarios for drones, including pre-flight, in-flight, and airspace route conflict detection, to obtain several risk control assessment items.

[0077] Step S1 involves analyzing multiple risk control scenarios for drones, including pre-flight, in-flight, and airspace route conflict detection, to obtain several risk control assessment items.

[0078] S11, based on the risk transmission characteristics of the entire drone flight lifecycle, deconstructs risk control scenarios in layers, including pre-flight preparation, dynamic in-flight management, and airspace route conflict detection. The specific deconstruction process includes:

[0079] For the pre-flight scenario, based on the airworthiness of the UAV itself and the compliance risks of the mission planning, multiple risk control assessment items are obtained, including UAV equipment status verification items, flight plan airspace permission verification items, payload adaptability assessment items, and take-off and landing point environmental safety assessment items.

[0080] For in-flight scenarios, considering the dynamic constraints and environmental interference risks of UAV dynamic flight, multiple risk control assessment items are obtained, including real-time flight attitude compliance assessment item, power system health status monitoring item, flight path deviation assessment item, real-time weather environment adaptability assessment item, and battery remaining range and mission range matching degree assessment item.

[0081] For the airspace route conflict detection scenario, the collaborative management and control requirements of the corresponding mission airspace are retrieved and decomposed into multiple risk control assessment items, including the multi-aircraft flight interval assessment item in the same airspace, the collision risk assessment item between the flight path and static obstacles, the conflict prediction item between the flight path and dynamic obstacles, and the compliance assessment item for cross-resolution airspace transition between layers.

[0082] S12, for each risk control assessment item obtained from the breakdown, clarify its applicable drone flight stage, risk level weight, and trigger priority, and extract the drone-specific risk control parameters required for the assessment item to form a parameter set:

[0083] S13 combines the dynamic constraints of UAV flight with airspace control rules to extract basic thresholds, including the dynamic parameter thresholds corresponding to the UAV model and the airspace management specification thresholds. Then, it supplements dynamic context trigger constraints by combining flight stage and mission type. It uses a composite logic of basic thresholds and dynamic context trigger constraints to configure scenario-based trigger conditions for each risk control assessment item.

[0084] S14. For each risk control assessment item's dependent parameter set, and considering the data source type of the parameters, a standardized data reading path expression is customized. Data source types include real-time acquisition from UAV onboard sensors, pre-configuration from ground stations, airspace management platform interface calls, and third-party meteorological service pushes. The data reading path expression is adapted to the parameter characteristics. For dynamic parameters that are frequently updated during UAV flight, a timestamp matching identifier is added to the path expression; for static parameters in the airspace environment, a cache validity period identifier is added; for hierarchical parameters that cross resolution airspaces, a resolution adaptation identifier is added. Finally, a syntax parser performs validity checks on all customized path expressions, eliminating invalid expressions that do not conform to the UAV data transmission format.

[0085] S15: Input the validated path expression and the trigger condition logic of each evaluation item into the expression parsing engine for pre-compilation processing.

[0086] Based on this, the raw flight-related data sent by the drone is parsed to determine the corresponding risk control scenario. Based on the determined risk control scenario, the object type and unique identifier of the data packet are identified and extracted. According to the identified object type and unique identifier, a globally unique composite cache key is generated. The format of the composite cache key is {object type: unique identifier}. According to the data source rules corresponding to the pre-configured risk control scenario, external services are called in parallel to obtain external flight risk data, and drone flight-related data is obtained by real-time calculation of the raw flight-related data. Figure 3 It is a detailed business process diagram of the data preprocessing and standardized caching process.

[0087] For example, during the flight plan submission phase of the UAV, the system receives flight plan data from the ground station or operating platform and performs the following detailed operations:

[0088] Step 1: Data Reception and Recognition

[0089] First, the system receives flight plan JSON data packets via multiple methods, including HTTP, RESTful API, and MQTT.

[0090] Secondly, parse the data packet, identify the objectType as "FlightPlan", and extract the unique identifier dataId (such as "FP_20241027_001").

[0091] Step 2: Generate composite cache key

[0092] Based on the identified object type and ID, a globally unique composite cache key is generated in the format {ObjectType}:{DataId}, for example, FlightPlan:FP_20241027_001.

[0093] Step 3: Parallel External Data Acquisition and Processing

[0094] Based on pre-configured data source rules, the system calls multiple external services in parallel to enrich the data dimensions and ensure processing efficiency.

[0095] Airspace compliance service: Input the flight path (plannedPath) and obtain information such as whether it crosses a no-fly zone (is_cross_restricted) and the nearest distance (nearest_distance).

[0096] Meteorological data service: Input flight route and time to obtain weather forecast (weather_condition), wind speed (wind_speed), etc.

[0097] Signal coverage service: Input flight route and obtain the estimated signal coverage score (coverage_score).

[0098] Air traffic conflict detection service: Takes in planning information and determines whether there is a conflict with other plans (has_traffic_conflict).

[0099] At the same time, inherent parameters such as maxAltitude (maximum altitude) and maxSpeed ​​(maximum speed) are extracted directly from the raw data.

[0100] Step 4: Construct standardized data objects

[0101] All acquired and extracted data is aggregated into a unified, structured JSON object. This object integrates the raw data and external service results, with a well-formatted structure for easy subsequent parsing.

[0102] {

[0103] "plan_id":"FP_20241027_001",

[0104] "max_altitude":500,

[0105] "max_speed":50,

[0106] "is_cross_restricted":false,

[0107] "nearest_distance_to_zone":1200,

[0108] "weather_condition":"clear",

[0109] "wind_speed":5.2,

[0110] "coverage_score":85,

[0111] "has_traffic_conflict":false

[0112] }

[0113] During the drone's flight, the system continuously receives real-time status data reported by the drone via the MQTT protocol and performs the following detailed operations:

[0114] Step 1: Real-time data reception and recognition

[0115] The system subscribes to MQTT topics and receives status data packets reported by drones at high frequency (usually several times per second).

[0116] Parse the data packet, identify objectType as "UAVRealtime", and extract the drone's unique identifier dataId (e.g., "UAV_456").

[0117] Step 2: Generate composite cache key

[0118] Generate a composite cache key, such as UAVRealtime:UAV_456.

[0119] Step 3: Ultra-fast real-time computing and rich data

[0120] To meet millisecond-level response requirements, the system performs lightweight, high-performance real-time computation:

[0121] Real-time airspace intrusion detection: The built-in high-performance spatial computing engine is invoked to instantly determine whether the current latitude and longitude (is_in_temp_zone) has been entered.

[0122] Battery status estimation: Based on the current remaining battery power (batteryRemaining) and average power consumption, quickly estimate the remaining flight time (estimated_flight_time).

[0123] Signal status transition: Converts the received signal strength indicator (signalStrength) into an enumerated type signal status (signal_status), such as "excellent", "good", "medium", "poor".

[0124] Step 4: Build the real-time state object and update the cache

[0125] The calculation results and key raw data are aggregated to form a real-time status JSON object.

[0126] {

[0127] "current_lon":116.401,

[0128] "current_lat":39.92,

[0129] "current_alt":450,

[0130] "current_speed": 12.5,

[0131] "battery_remaining":45,

[0132] "estimated_flight_time":15,

[0133] "signal_status":"medium",

[0134] "is_in_temp_zone":true,

[0135] "last_update_time":1730012345

[0136] }

[0137] Write the object to the cache with a composite cache key and set a short TTL (e.g., 15 seconds) to ensure that the data expires quickly and reflects the latest state.

[0138] (2) Batch data loading and dynamic parameter parsing process

[0139] See Figure 4 When the risk control assessment engine is triggered (whether it's a pre-flight plan assessment or real-time in-flight monitoring), the following data loading and parsing process is uniformly executed:

[0140] Step 1: Trigger evaluation and dependency analysis

[0141] The risk control engine loads a list of all risk control parameters that the assessment item depends on from the configuration center, based on the assessment item (such as "Comprehensive Risk of Flight Plan" or "Intrusion into No-Fly Zone").

[0142] Each parameter is associated with a predefined data read path expression (e.g., JSONPath format: $.is_in_temp_zone, $.battery_remaining).

[0143] Step 2: Generate cache keys and batch load

[0144] Determine the composite cache key based on the current evaluation context (e.g., whether the evaluation object is FlightPlan:FP_20241027_001 or UAVRealtime:UAV_456).

[0145] The unified data access layer uses this key to initiate a one-time GET request to the in-memory cache database to retrieve the complete standardized data object. This step is crucial for performance improvement, compressing N potential queries into a single one.

[0146] Step 3: Dynamic Parameter Analysis

[0147] Load the complete data object into memory.

[0148] An expression parsing engine (such as one that supports JSONPath or SPEL) iterates through each data reading path expression that requires parameters.

[0149] The engine applies expressions to data objects in memory, dynamically locating and retrieving the specific values ​​needed. For example, for the expression $.is_in_temp_zone, the engine retrieves true from the real-time state object.

[0150] All extracted parameter values ​​are assembled and fed into the rule engine for logical calculation.

[0151] (3) Expression parsing engine pre-compilation and multi-level caching optimization

[0152] When the system starts up or when the configuration management module detects changes in the data source configuration, the pre-compiled cache manager proactively pre-compiles all data read path expressions. Pre-compilation converts string-based expressions (such as `$.weather_condition`) into intermediate code or a pre-initialized context that can be efficiently executed within the expression parsing engine. These pre-compiled results are cached in the application server's local memory (such as a HashMap). During real-time parsing, the engine directly calls the pre-compiled results, avoiding the overhead of repeated lexical and syntax analysis. Figure 5 This demonstrates an optimization mechanism where data read path expressions are pre-compiled and cached during system startup or configuration changes, thus avoiding repeated parsing in real-time requests. The specific pre-compilation process includes:

[0153] The validated path expressions, including data source identifiers and drone parameter characteristic adaptation identifiers, along with the scenario-based triggering condition logic for each risk control assessment item, are input into the expression parsing engine for pre-compilation processing to generate a directly executable syntax tree.

[0154] To optimize the path expressions corresponding to high-frequency dynamic parameters of UAVs during flight, including real-time attitude, position, and remaining battery power, the lexical analysis rules of the parsing engine are improved. A mapping cache between parameter names and data transmission fields is established to reduce field matching overhead during real-time parsing and adapt to the high-frequency data update characteristics of UAV dynamic flight. Common path expressions for multiple evaluation items in the same risk control scenario are merged and compiled to generate shared syntax tree fragments. For example, by merging and compiling expressions corresponding to shared parameters such as airspace coordinates and no-fly zone boundaries, a shared syntax tree fragment can be generated, effectively reducing the overhead of repeated parsing under low-altitude communication links and saving computing resources for both the airborne and ground systems. During the pre-compilation process, multiple UAV parameter type verification rules are embedded, including coordinate format verification, flight speed unit verification, BeiDou grid code validity verification, and battery endurance parameter range verification. This results in a pre-compiled parsing engine instance for subsequent dynamic parameter extraction.

[0155] Specifically, the process of merging and compiling common path expressions of multiple evaluation items under the same risk control scenario to generate a shared syntax tree fragment includes the following steps:

[0156] For each risk control scenario, each risk control assessment item is further subdivided according to the type of assessment purpose, and divided into multiple assessment-specific clusters. For example, during flight, the risk control attitude management cluster, power monitoring cluster, and endurance matching cluster can be divided. Taking the risk control attitude management cluster and collision risk cluster as examples, the real-time flight attitude compliance assessment item and the trajectory deviation assessment item can be divided into the risk control attitude management cluster, while the trajectory collision risk assessment item and the inter-layer cross-resolution airspace transition compliance assessment item can be divided into the collision risk cluster.

[0157] Select each evaluation item cluster one by one, extract its parameter set and construct a parameter-evaluation item mapping table, and clarify the type and data source type of each parameter; use the following formula to calculate the parameter sharing rate among the evaluation items in the evaluation item cluster: Parameter sharing rate = Number of shared parameters / Total number of parameters between the two evaluation items.

[0158] Considering the dense nature of UAV risk control parameters, a sharing rate threshold is set, and evaluation item combinations with a parameter sharing rate greater than the sharing rate threshold and whose shared parameters are core parameters are retained. Core parameters refer to key parameters that directly affect risk control decisions, such as real-time latitude and longitude, pitch angle, and roll angle of the attitude control cluster during flight, and Beidou grid code and airspace coordinates in the airspace conflict cluster. The real-time flight attitude compliance evaluation item and the track offset evaluation item share core parameters such as real-time latitude and longitude, pitch angle, and planned track coordinates.

[0159] Extract the path expressions corresponding to each evaluation item combination, and verify the compatibility of syntax structure (avoiding syntax format conflicts), compatibility of shared parameter adaptation identifiers, and data type compatibility (meaning that the data types of the parameters pointed to by the expressions are consistent, avoiding type conversion conflicts). If there are completely identical expressions, they are directly marked as mergeable. If the expressions are not completely identical but compatible, they are merged through shared syntax tree fragments and branch parsing to avoid parsing conflicts. Finally, multiple mergeable compilation evaluation item groups are formed. The risk control evaluation items in each mergeable compilation evaluation item group do not exceed the preset evaluation item number threshold to avoid too many parameters in the group, which would reduce the merging efficiency.

[0160] For each mergeable compilation evaluation item group, a pre-compilation simulation is performed to verify whether the merged syntax tree fragment can be parsed normally and without logical conflicts. If the merged expression of any evaluation item causes the number of new branches to exceed the preset branch threshold, it is removed from the mergeable compilation evaluation item group and compiled separately to avoid the situation where the merging benefit is less than the parsing cost.

[0161] Establish a mapping between mergeable evaluation item groups and shared syntax trees, and store it in the pre-compiled cache manager.

[0162] The aforementioned common path expression merging and compilation process, based on scenario homogeneity, can effectively ensure that the merged evaluation items are still adapted to the specific risk control scenario of UAVs and do not deviate from actual needs. The merged and compiled expressions can effectively reduce the overhead of repeated parsing and adapt to the constraints of low-altitude communication links and computing resources.

[0163] On the application server where the risk control assessment engine resides, a local cache is established (using Caffeine or GuavaCache). When data is successfully loaded from the central cache (such as Redis), it is written to the local cache simultaneously with the data being returned to the upper layer. Subsequent requests for data with the same cache key will preferentially read from the local cache. Because local memory access is extremely fast (nanoseconds to microseconds), this further significantly reduces data access latency.

[0164] In the configuration, a static degradation default value is preset for each data source parameter (e.g., the default value for weather_condition is "unknown"). When batch data loading fails (e.g., cache breakdown, network anomaly), or when a field is found to be empty during dynamic parsing, the system will automatically use the preset default value in rule calculation and record an alarm log. This mechanism ensures that the risk control service can still provide a lossy but usable service when some data is temporarily unavailable, guaranteeing the overall resilience of the system. Figure 6 It explains the relationship between local cache and central cache, and how a degradation strategy is triggered when data cannot be retrieved, thus ensuring the working mechanism of system availability. Figure 7 This visually demonstrates the significant performance improvement that the present invention brings in terms of reducing the number of requests and lowering latency.

[0165] To achieve the above method, this invention discloses a real-time risk control assessment system for drones based on efficient data caching and parameter parsing. (See [link to relevant documentation]). Figure 2 The system includes the following functional modules, and is specifically designed and coordinated for drone risk control scenarios. Its architecture is as follows:

[0166] Data preprocessing and caching module: As the system's data entry point, it is responsible for receiving the UAV's flight plan and real-time status data. It has built-in data cleaning and format validation logic, and encapsulates the ability to call external services (airspace, weather, etc.) and perform high-speed calculations on real-time data. Ultimately, it is responsible for building standardized objects and writing them to the central cache.

[0167] Unified Data Access Layer (DataLoader): This module provides a transparent and unified data access interface to the upper-layer business (risk control rule engine). It encapsulates the logic for generating composite cache keys, batch loading logic, and multi-level caching (local first, then central) query strategies, and is the core of reducing I / O bottlenecks.

[0168] Expression parsing engine module: This module is the core of dynamic parameter parsing, with built-in high-performance parsers (such as those supporting JSONPath and SpringSPEL). It utilizes pre-compiled results to efficiently extract values ​​from data objects in memory, balancing flexibility and performance.

[0169] Precompiled Cache Manager: This module is responsible for managing the caching of the precompiled lifecycle of data read path expressions, and also physically manages local data caching (such as Caffeine cache instances), making it a key component for system performance optimization.

[0170] Configuration Management Module: This module provides a user interface (UI) or application programming interface (API) for business personnel or administrators to dynamically configure data source rules, data read path expressions, and downgraded default values, thus realizing the system's "configuration-driven" flexibility.

[0171] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0172] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A real-time risk control assessment method for drones based on efficient data caching and parameter parsing, characterized in that, The method includes: S1 analyzes multiple risk control scenarios for drones, including pre-flight, in-flight, and airspace route conflict detection, and obtains several risk control evaluation items. For each risk control evaluation item, it configures the corresponding trigger conditions, dependent risk control parameters, and data reading path expressions for each risk control parameter, and completes the pre-compilation of the expression parsing engine. S2 receives raw flight-related data sent by the drone, determines the risk control scenario corresponding to the data, calls external services to obtain external flight risk data for all risk assessment items under the risk control scenario based on the determination result, and obtains drone flight-related data by performing real-time calculations on the raw flight-related data. S3 aggregates external flight risk data and drone flight-related data, generates corresponding standardized objects and globally unique composite cache keys, and writes them into the cache. S4. Based on the evaluation items set for the fields in the original flight data in the risk control system, automatically trigger one or more risk control evaluation items; for each triggered risk control evaluation item, load the list of all risk control parameters and data reading path expressions that the risk control evaluation item depends on, and generate a composite cache key based on the semantic analysis results; based on the composite cache key, load the corresponding standardized objects from the cache in batches at once. S5 uses an expression parsing engine to dynamically extract risk control parameter values ​​from standardized objects, and then sends the extracted parameter values ​​into the rule engine for real-time risk control assessment.

2. The real-time risk control assessment method for UAVs based on efficient data caching and parameter parsing according to claim 1, characterized in that, Step S1 involves analyzing multiple risk control scenarios for drones, including pre-flight, in-flight, and airspace route conflict detection, to obtain several risk control assessment items. S11, based on the risk transmission characteristics of the entire drone flight lifecycle, deconstructs risk control scenarios in layers, including pre-flight preparation, dynamic in-flight management, and airspace route conflict detection. The specific deconstruction process includes: For the pre-flight scenario, based on the airworthiness of the UAV itself and the compliance risks of the mission planning, multiple risk control assessment items are obtained, including UAV equipment status verification items, flight plan airspace permission verification items, payload adaptability assessment items, and take-off and landing point environmental safety assessment items. For in-flight scenarios, considering the dynamic constraints and environmental interference risks of UAV dynamic flight, multiple risk control assessment items are obtained, including real-time flight attitude compliance assessment item, power system health status monitoring item, flight path deviation assessment item, real-time weather environment adaptability assessment item, and battery remaining range and mission range matching degree assessment item. For the airspace route conflict detection scenario, the collaborative management and control requirements of the corresponding mission airspace are retrieved and decomposed into multiple risk control assessment items, including the multi-aircraft flight interval assessment item in the same airspace, the collision risk assessment item between the flight path and static obstacles, the conflict prediction item between the flight path and dynamic obstacles, and the compliance assessment item for cross-resolution airspace transition between layers. S12, for each risk control assessment item obtained from the breakdown, clarify its applicable drone flight stage, risk level weight, and trigger priority, and extract the drone-specific risk control parameters required for the assessment item to form a parameter set: S13 combines the dynamic constraints of UAV flight with airspace control rules to extract basic thresholds, including the dynamic parameter thresholds corresponding to the UAV model and the airspace management specification thresholds. Then, it supplements dynamic context trigger constraints by combining flight stage and mission type. It uses a composite logic of basic thresholds and dynamic context trigger constraints to configure scenario-based trigger conditions for each risk control assessment item. S14. For each risk control assessment item's dependent parameter set, and considering the data source type of the parameters, a standardized data reading path expression is customized. Data source types include real-time acquisition from UAV onboard sensors, pre-configuration from ground stations, airspace management platform interface calls, and third-party meteorological service pushes. The data reading path expression is adapted to the parameter characteristics. For dynamic parameters that are frequently updated during UAV flight, a timestamp matching identifier is added to the path expression; for static parameters in the airspace environment, a cache validity period identifier is added; for hierarchical parameters that cross resolution airspaces, a resolution adaptation identifier is added. Finally, a syntax parser performs validity checks on all customized path expressions, eliminating invalid expressions that do not conform to the UAV data transmission format. S15: Input the validated path expression and the trigger condition logic of each risk control assessment item into the expression parsing engine for pre-compilation processing.

3. The real-time risk control assessment method for UAVs based on efficient data caching and parameter parsing according to claim 2, characterized in that, The pre-compilation process of the expression parsing engine is achieved by converting the string-form expression into intermediate code or pre-initialized context that can be efficiently executed within the expression parsing engine. The pre-compilation result is cached in the local memory of the application server. During real-time parsing, the expression parsing engine directly calls the pre-compiled results; The specific pre-compilation process includes: The validated path expressions, including data source identifiers and drone parameter characteristic adaptation identifiers, along with the scenario-based triggering condition logic for each risk control assessment item, are input into the expression parsing engine for pre-compilation processing to generate a directly executable syntax tree. For the path expressions corresponding to high-frequency dynamic parameters during drone flight, including real-time attitude, position, and remaining battery power, the lexical analysis rules of the parsing engine are optimized, and a mapping cache between parameter names and data transmission fields is established; common path expressions of multiple evaluation items under the same risk control scenario are merged and compiled to generate a shared syntax tree fragment. During the pre-compilation process, multiple drone parameter type verification rules are embedded, including coordinate format verification, flight speed unit verification, BeiDou grid code validity verification, and battery endurance parameter range verification, ultimately forming a pre-compiled parsing engine instance.

4. The real-time risk control assessment method for UAVs based on efficient data caching and parameter parsing according to claim 3, characterized in that, The process of merging and compiling common path expressions of multiple evaluation items under the same risk control scenario to generate a shared syntax tree fragment includes the following steps: For each risk control scenario, each risk control assessment item is further subdivided according to the type of assessment purpose, and divided into multiple assessment-specific clusters; Select each evaluation cluster one by one, extract its parameter set and construct a parameter-evaluation item mapping table, and clarify the type and data source type of each parameter; use the following formula to calculate the parameter sharing rate between each evaluation item in the evaluation cluster: Parameter sharing rate = Number of shared parameters / Total number of parameters between the two evaluation items; Considering the dense nature of drone risk control parameters, a sharing rate threshold is set, and evaluation item combinations with a sharing rate greater than the sharing rate threshold and shared parameters being core parameters are retained; core parameters refer to key parameters that directly affect risk control decisions. Extract the path expressions corresponding to each evaluation item combination, and verify the compatibility of syntax structure, shared parameter adaptation identifier, and data type respectively. If there are completely identical expressions, they are directly marked as mergeable. If the expressions are not completely identical but compatible, they are merged through shared syntax tree fragments and branch parsing, and finally multiple mergeable compilation evaluation item groups are formed. The risk control evaluation items in each mergeable compilation evaluation item group do not exceed the preset evaluation item number threshold. For each mergeable compilation evaluation item group, perform pre-compilation simulation to verify whether the merged syntax tree fragment can be parsed normally and without logical conflicts; if the merged expression of any evaluation item causes the new branch to exceed the preset branch threshold, remove it from the mergeable compilation evaluation item group and compile it separately. Establish a mapping between mergeable evaluation item groups and shared syntax trees, and store it in the pre-compiled cache manager.

5. The real-time risk control assessment method for drones based on efficient data caching and parameter parsing according to claim 1, characterized in that, Step S2 further includes: S21, parse the raw flight-related data sent by the drone, determine the risk control scenario corresponding to the data, and based on the determined risk control scenario, identify and extract the object type and unique identifier of the data packet; S22. Generate a globally unique composite cache key based on the identified object type and unique identifier. The format of the composite cache key is {object type: unique identifier}. S23, based on the pre-configured data source rules corresponding to the risk control scenario, calls external services in parallel to obtain external flight risk data, and performs real-time calculations on the flight-related raw data to obtain drone flight-related data.

6. The real-time risk control assessment method for UAVs based on efficient data caching and parameter parsing according to claim 5, characterized in that, In step S23, when the flight-related raw data contains a flight plan, the risk control scenario is determined to be before flight, the object type of the data packet is identified as FlightPlan, the plan number is extracted as a unique identifier, and a globally unique composite cache key {FlightPlan: plan number} is generated; then, multiple external services, including airspace compliance service, meteorological data service, signal coverage service, and air traffic conflict detection service, are called in parallel to obtain external flight risk data; at the same time, the inherent parameters of the flight plan are extracted from the flight-related raw data to obtain the UAV flight-related data.

7. The real-time risk control assessment method for UAVs based on efficient data caching and parameter parsing according to claim 5, characterized in that, In step S23, when the original flight-related data includes real-time flight status data, the risk control scenario is determined to be in flight, the object type of the data packet is identified as UAVRealtime, the drone number is extracted as a unique identifier, and a globally unique composite cache key {UAVRealtime: drone number} is generated; the built-in spatial computing engine is called to determine whether the drone has entered a temporary restricted area based on the current latitude, longitude and altitude, to estimate the remaining flight time based on the current remaining battery power and average power consumption, and to convert the received signal strength indication into an enumerated type of signal status to obtain the drone flight-related data.

8. The real-time risk control assessment method for UAVs based on efficient data caching and parameter parsing according to claim 1, characterized in that, On the application server where the expression parsing engine resides, a local cache is established. When data is successfully loaded from the central cache, it is written to the local cache while returning the data to the upper layer. Subsequent data requests for the same composite cache key are preferentially read from the local cache.

9. The real-time risk control assessment method for UAVs based on efficient data caching and parameter parsing according to claim 1, characterized in that, In the configuration, a static degradation default value is preset for each data source parameter; when batch data loading fails or a field is found to be empty during dynamic parsing, the preset static degradation default value is automatically used to participate in the rule calculation, and an alarm log is recorded.

10. A real-time risk control assessment system for unmanned aerial vehicles (UAVs) based on efficient data caching and parameter parsing, characterized in that, The system includes a data preprocessing and caching module, a unified data access layer, an expression parsing engine module, a pre-compiled cache manager, and a configuration management module. The data preprocessing and caching module receives flight plan data, real-time flight status data, and raw airspace environment data sent by the UAV. Based on UAV risk control scenario identification rules, it determines the risk control scenario corresponding to the data and extracts the object type and unique identifier of the data packet. According to pre-configured data source rules, it calls external services in parallel, including airspace compliance service, meteorological data service, and air traffic conflict detection service, to obtain external flight risk data. At the same time, it performs real-time calculations on the raw data through the built-in spatial computing engine to obtain UAV flight-related data. Then, by aggregating the external flight risk data and UAV flight-related data, it generates a standardized object containing UAV-specific parameters and generates a globally unique composite cache key based on the object type and unique identifier, and writes the standardized object into the central cache. The unified data access layer is used to encapsulate the compound cache key generation logic and multi-level cache query strategy, and provide a unified data access interface for the expression parsing engine module. When a risk control assessment item is triggered, the target compound cache key is generated according to the risk control scenario and semantic analysis results corresponding to the assessment item, and the corresponding standardized objects are loaded in batches from the local memory cache or central cache of the pre-compiled cache manager. The expression parsing engine module receives the pre-compiled results passed by the pre-compiled cache manager, directly calls the pre-compiled results during the real-time parsing phase, dynamically extracts the drone risk control parameter values ​​from the standardized objects loaded by the unified data access layer, and automatically performs drone parameter exclusive type verification during the parsing process. The pre-compiled cache manager, associated with the configuration management module and the expression parsing engine module, is used to receive the verified path expressions and scenario-based trigger condition logic issued by the configuration management module, group them according to the drone risk control scenario, optimize the lexical analysis rules for expressions corresponding to high-frequency dynamic parameters, perform merged compilation on common expressions, generate shared syntax tree fragments, and embed drone parameter type verification rules to form pre-compiled results. The pre-compiled results are cached in the local memory of the application server, and a mapping relationship is established between mergeable evaluation items and shared syntax trees for the expression parsing engine module to call in real time. The configuration management module provides a visual configuration interface and API interface to support the configuration of scenario-based triggering conditions for risk control assessment items, drone-specific risk control parameter sets, multi-source data source rules, path expressions, and static degradation default values ​​of parameters; and synchronizes configuration information to the data preprocessing and caching module and the pre-compiled cache manager in real time.

Images