A low-altitude logistics distribution decision-making method and system based on multi-source data fusion

Abstract

The application relates to the technical field of multi-source data processing, and discloses a low-altitude logistics distribution decision-making method and system based on multi-source data fusion, which is used for solving the problem that a cost-benefit evaluation framework for large-scale deployment of a low-altitude logistics system is lacked in a traditional method; according to the application, unified data is formed by collecting multi-source data; the data is subjected to time-space alignment and weight integration to generate a logistics state data set; a cost model is established based on the data set, equipment procurement, maintenance, network expansion and manpower input are parameterized, and are associated with deployment scale operation strategies; the model is used to generate benefit indexes such as return on investment, operation efficiency and resource utilization rate; the cost model and the benefit indexes are integrated into a closed-loop evaluation framework which is dynamically adjusted according to scene parameters; the framework output is applied to distribution decision-making, path optimization and resource allocation are adjusted through iteration, and management optimization of large-scale deployment of low-altitude logistics is realized.

Description

Technical Field

[0001] This invention relates to the field of multi-source data processing technology, specifically to a low-altitude logistics distribution decision-making method and system based on multi-source data fusion. Background Technology

[0002] With the rapid development of the low-altitude economy, low-altitude logistics and distribution, as an efficient and flexible transportation mode, has been widely used in urban e-commerce, medical supplies, and emergency rescue. The use of multi-source data fusion technologies, such as satellite navigation, meteorological monitoring, and sensor data integration, to achieve route planning and real-time decision-making has become a hot research topic in the industry. However, existing low-altitude logistics and distribution decision-making methods and systems still face significant challenges in practical applications, particularly the lack of a cost-benefit evaluation framework for large-scale deployment. This makes it impossible to systematically quantify return on investment, operating costs, and resource allocation efficiency, thus hindering the industrialization and promotion of low-altitude logistics.

[0003] In the prior art, CN117669993A discloses a low-altitude logistics delivery route planning method based on multi-source data fusion. This patent integrates satellite, meteorological, and ground sensor data and uses the Bayes inference algorithm to optimize route planning and safety early warning, making it suitable for delivery decisions in complex environments. While this method improves route accuracy and real-time response capabilities, its focus is limited to algorithm optimization for single or small-scale delivery scenarios. It does not address the overall cost assessment framework for large-scale deployment and cannot analyze comprehensive economic factors such as equipment procurement, maintenance, network expansion, and manpower investment, making it difficult to predict the long-term return on investment during actual promotion. And resource utilization efficiency; similarly, CN116976597A discloses a key technology and logistics application demonstration of low-altitude communication, navigation and monitoring fusion service based on Beidou-3 RDSS. This patent integrates Beidou satellite data and AI prediction models to build a centimeter-level positioning and spatiotemporal fusion system to support safety early warning for urban air traffic and drone delivery; although this technology enhances the accuracy of data fusion and emergency response, it also ignores the economic dimension of large-scale deployment, such as infrastructure investment, energy consumption and scale effect assessment, and cannot provide quantitative tools to optimize resource allocation, which can easily lead to cost overruns and inefficiencies in multi-fleet or cross-regional operations;

[0004] While the aforementioned existing technologies have made progress in data fusion and route decision-making, they have not addressed the core pain point of large-scale deployment of low-altitude logistics systems: the lack of a comprehensive cost-benefit assessment framework. This makes it difficult to integrate economic indicators in the decision-making process, and to effectively quantify the balance between initial investment, operation and maintenance, and potential benefits, resulting in resource waste and limited industry scalability. Therefore, there is an urgent need for a new method and system to construct an integrated cost-benefit assessment framework for low-altitude logistics delivery decisions based on multi-source data fusion, in order to achieve efficient and sustainable large-scale deployment. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a low-altitude logistics delivery decision-making method and system based on multi-source data fusion, which solves the problem of the lack of a cost-benefit evaluation framework for large-scale deployment of low-altitude logistics systems in traditional methods.

[0006] To achieve the goal of efficient and sustainable large-scale deployment mentioned in the background section, this invention provides the following technical solution:

[0007] A low-altitude logistics delivery decision-making method based on multi-source data fusion includes:

[0008] S1: Collect multi-source data from the low-altitude logistics system, use the collected multi-source data as input sources, and access, initially classify and store it through a unified interface to form a unified data foundation;

[0009] S2: The collected multi-source data is fused and processed, and the data from different sources are aligned according to timestamps and spatial coordinates, and integrated into a unified logistics status dataset through a preset weight allocation mechanism;

[0010] S3: Based on the fused logistics status dataset, establish a cost model for large-scale deployment scenarios, parameterize various cost variables, and establish a correlation mapping relationship with deployment scale and operation strategy;

[0011] S4: Construct net cash flow based on the period cost and period revenue series, discount it to obtain net present value and iteratively calculate internal rate of return, and use task logs to statistically analyze efficiency and resource utilization indicators under threshold constraints;

[0012] S5: Integrate cost models and benefit indicators to form an evaluation framework, link all elements into a closed-loop structure within the evaluation framework, and adjust the framework parameters according to the parameters of different delivery scenarios.

[0013] S6: Apply the evaluation framework to large-scale delivery decisions in low-altitude logistics systems, use the framework output as decision input, and optimize delivery routes and resource allocation under large-scale deployment conditions through iterative adjustments.

[0014] In a preferred embodiment, multi-source data from the low-altitude logistics system is collected. This collected multi-source data is used as input, accessed through a unified interface, preliminarily classified, and stored to form a unified data foundation, including:

[0015] Configure a low-altitude logistics data acquisition scheme, using satellite navigation position, meteorological elements, airborne sensor status, and historical efficiency and fault data as multiple input sources;

[0016] Standardized access is performed through a unified API interface. Input data is sequentially checked for format, numerical range and integrity. A queue buffering mechanism is configured to queue and cache data arriving during peak periods.

[0017] The data is initially classified, and metadata tags containing data source identifiers, collection timestamps, and priority markers are generated for each group of data.

[0018] In the cloud storage platform, location groups, environment groups, real-time groups, and historical groups are stored in geographically indexed partitions, time-series engines, low-latency storage areas, and relational databases, respectively. Sensitive fields are encrypted and access controlled, and data management is carried out in conjunction with backup strategies and audit logs.

[0019] In a preferred embodiment, the collected multi-source data is fused, aligning the data from different sources according to timestamps and spatial coordinates, including:

[0020] Align multi-source data in the time dimension, using the timestamp of satellite navigation data as a benchmark, perform synchronous matching on meteorological monitoring data, real-time sensor data and historical operational data, and retrieve the most recent valid record of each and associate it with the corresponding navigation time point;

[0021] After time alignment, spatial coordinate mapping is performed on the multi-source data. A unified geographic reference system is adopted to convert the location of meteorological monitoring stations, the relative orientation information of sensors, and historical path records into unified global latitude and longitude coordinates. Altitude is introduced as a third dimension to interpolate the height data.

[0022] A weight allocation mechanism is introduced to assess the reliability of each data source to determine its relative weight, and to dynamically adjust the ratio between the weights according to the application scenario.

[0023] In a preferred embodiment, the data is integrated into a unified logistics status dataset through a preset weight allocation mechanism, including:

[0024] By using weights to hierarchically integrate aligned data, fusion of navigation and sensor data is used to generate flight status data, and meteorological data is overlaid and linked with historical records to generate risk markers.

[0025] The integrated dataset output is used as input to the cost model, and historical feature fields of path meteorological equipment are provided.

[0026] The fused data is used as a unified input for cost model building and benefit index calculation, and is processed continuously between the data layer and the model layer in a predetermined order.

[0027] In a preferred embodiment, a cost model for large-scale deployment scenarios is established based on the fused logistics status dataset, including:

[0028] The equipment procurement cost variable is defined by extracting hardware specification requirements from the fused dataset and combining them with configuration quantity and unit price;

[0029] Define maintenance cost variables and extract maintenance frequency and maintenance expenditure types based on environmental data, real-time operating data and historical fault records;

[0030] Based on location data, environmental data, and historical data, statistical communication coverage density and infrastructure configuration parameters are used to define network expansion cost variables.

[0031] In a preferred embodiment, the various cost variables are parametrically modeled and correlated with deployment scale and operational strategy, including:

[0032] Define the human resource input cost variable, and determine the monitoring positions, training cycles, and personnel configuration parameters based on flight status records and operational records;

[0033] Each cost variable is converted into period cost according to a unified aggregation granularity. The number of occurrences in a period is calculated based on the quantity, unit price, cycle, and frequency of occurrence in that period. The total period cost is calculated based on equipment procurement, maintenance, network, and manpower.

[0034] Establish correlation mapping relationships to associate each cost variable with the deployment scale and operation strategy, and construct a cost framework structure with total cost and sub-cost views.

[0035] In a preferred embodiment, net cash flow is constructed based on the periodic cost and periodic revenue sequence, discounted to calculate net present value, and iteratively calculated internal rate of return. Efficiency and resource utilization indicators are statistically analyzed from the task log under threshold constraints, including:

[0036] Construct an investment return rate indicator, establish a cash flow series from cost variables and historical revenue data, and calculate net present value and internal rate of return;

[0037] Construct operational efficiency metrics, extract relevant variables from the dataset output by the model, and calculate unit delivery time and task success rate;

[0038] Construct a resource utilization index, combine scheduling records with the human resource input and equipment configuration variables in the cost model, calculate the idle time ratio of each drone, and calculate the load balance degree based on the dispersion of task allocation.

[0039] The net cash flow sequence is composed of the period cost sequence and the period revenue sequence. The net present value is discounted, the internal rate of return is iteratively calculated, and the unit delivery time and task completion rate are statistically analyzed under the threshold set constraints. The idle time ratio and load balance are calculated. The above indicators are combined into a multi-dimensional benefit indicator set as the framework input.

[0040] In a preferred embodiment, a cost model and benefit indicators are integrated to form an evaluation framework. Within this framework, various elements are linked into a closed-loop structure, and the framework parameters are adjusted according to parameters for different delivery scenarios, including:

[0041] Cost variables and benefit indicators are aggregated to form an evaluation element set, and the corresponding correlation between equipment procurement costs, maintenance costs, network expansion costs, human resource input and return on investment, operational efficiency and resource utilization rate are established in the element set;

[0042] The evaluation framework constructs a cost-benefit linkage structure, sets preset weights for the mapping paths between cost variables and benefit indicators, and adjusts the influence coefficients of the weights according to scenario parameters.

[0043] A closed-loop feedback mechanism is introduced to set target ranges for each benefit indicator and adjust cost variables and path configuration parameters based on the deviation between the evaluation results and the target ranges.

[0044] Scenario parameters are introduced into the evaluation framework to dynamically adjust the weights of each evaluation indicator, and the corresponding framework configuration is selected or updated based on city density, flight distance, and fleet size.

[0045] In a preferred embodiment, the evaluation framework is applied to large-scale delivery decisions in a low-altitude logistics system. The framework output is used as the decision input, and the delivery routes and resource allocation under large-scale deployment conditions are optimized through iterative adjustments, including:

[0046] The quantitative evaluation results are extracted from the evaluation framework as decision inputs and combined with cost variables and scenario parameters to generate a multi-dimensional set of decision inputs.

[0047] An initial delivery plan is generated based on order demand and the low-altitude transportation network. The data corresponding to the delivery plan is then input into an evaluation framework to calculate various benefit indicators and compare them with preset target ranges.

[0048] An iterative adjustment mechanism is introduced in the decision-making process. Based on the evaluation deviation, the delivery route and resource allocation are cyclically optimized. The delivery route is reduced, merged and its layout is adjusted, and the order aggregation method is adjusted. The resource allocation is subject to task migration and fleet rotation operations.

[0049] After each round of path and resource adjustments, the solution is re-entered into the evaluation framework to obtain metrics. The iteration is terminated when the deviation is within the allowable range or the number of iterations reaches the upper limit.

[0050] When order distribution, environmental parameters, or cost parameters change during system operation, cost and benefit indicators are recalculated based on updated multi-source data. The recalculated indicators are then input into the evaluation framework, and the optimization steps of path planning and resource allocation are re-executed.

[0051] On the other hand, the present invention provides a low-altitude logistics delivery decision-making system based on multi-source data fusion, comprising:

[0052] Data acquisition module: Collects multi-source data from the low-altitude logistics system and performs standardized access, classification, and cloud storage through a unified API interface;

[0053] Data fusion module: It integrates the collected data by merging them, aligning them with timestamps and spatial coordinates, and applying a weighting mechanism to create a unified logistics status dataset.

[0054] Cost model building module: Based on the fused dataset, a cost model for large-scale deployment is established. Equipment procurement, maintenance, network expansion and human resources are used as cost variables. Parametric modeling is performed and correlation mapping is established to form a scalable cost framework, providing a quantitative basis for the generation of benefit indicators.

[0055] Benefit indicator generation module: Generates benefit indicators using the cost model, and quantifies each indicator based on the relationship between the above variables to form a multi-dimensional benefit indicator set;

[0056] Evaluation framework integration module: Integrates cost models and benefit indicators into a closed-loop evaluation framework, links various evaluation elements through feedback loops, and dynamically adjusts framework parameters according to scenario parameters to ensure that the evaluation framework adapts to different deployment needs;

[0057] Decision Application Module: The output of the evaluation framework is used as the input for delivery decisions. By iteratively adjusting and refining route selection and resource allocation, a business closed loop from data collection to decision optimization is formed.

[0058] Compared with existing technologies, this invention provides a low-altitude logistics distribution decision-making method and system based on multi-source data fusion, which has the following beneficial effects:

[0059] 1. This invention, through the systematic collection and fusion of data from satellite navigation, meteorological monitoring, airborne sensors, and historical operational data, establishes a unified and standardized data foundation, providing reliable support for the accurate calculation of cost models and benefit indicators. This method, through parametric modeling and dynamic adjustment, can accurately estimate costs such as equipment procurement, maintenance, network expansion, and manpower input, and generate multi-dimensional benefit indicators including return on investment, operational efficiency, and resource utilization. This helps decision-makers optimize path planning and resource allocation in large-scale deployment environments. Furthermore, based on an adaptive adjustment mechanism within the evaluation framework, the system can dynamically adjust according to changes in the environment, tasks, and equipment during actual operation, ensuring optimal cost control and resource allocation efficiency for the low-altitude logistics system under various deployment conditions. This provides feasible technical support for the widespread application of low-altitude logistics and addresses the problem of traditional methods lacking a cost-benefit evaluation framework for large-scale deployment of low-altitude logistics systems.

[0060] 2. This invention integrates satellite navigation, meteorological monitoring, airborne sensor data, and historical operational records to form a comprehensive real-time logistics status data foundation, supporting dynamic optimization. Based on this data, this invention employs a combination of cost models and benefit indicators to conduct real-time evaluations of delivery routes and resource allocation, automatically adjusting the decision-making process to cope with different geographical environments, meteorological conditions, and task requirements. Furthermore, the system can continuously learn from real-time equipment status and historical data, gradually improving the accuracy of route planning and resource allocation, reducing manual intervention. Through this method, the low-altitude logistics system can effectively avoid resource waste and misscheduling, ensuring the efficient execution of delivery tasks, and optimizing costs and benefits based on actual operating conditions, providing an innovative solution for the large-scale application of low-altitude logistics systems. Attached Figure Description

[0061] Figure 1 This is a schematic diagram of the low-altitude logistics distribution decision-making method based on multi-source data fusion according to the present invention.

[0062] Figure 2 This is a schematic diagram of the structure of a low-altitude logistics distribution decision system based on multi-source data fusion according to the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Example 1: Figure 1A decision-making method for low-altitude logistics delivery based on multi-source data fusion is presented, including:

[0065] S1: Collect multi-source data from the low-altitude logistics system, use the collected multi-source data as input sources, and access, initially classify and store it through a unified interface to form a unified data foundation;

[0066] S2: The collected multi-source data is fused and processed, and the data from different sources are aligned according to timestamps and spatial coordinates, and integrated into a unified logistics status dataset through a preset weight allocation mechanism;

[0067] S3: Based on the fused logistics status dataset, establish a cost model for large-scale deployment scenarios, parameterize various cost variables, and establish a correlation mapping relationship with deployment scale and operation strategy;

[0068] S4: Use the cost model to generate benefit indicators, taking return on investment, operational efficiency and resource utilization as the basis for calculating benefit indicators, and quantify the indicators through the relationship between variables;

[0069] S5: Integrate cost models and benefit indicators to form an evaluation framework, link all elements into a closed-loop structure within the evaluation framework, and adjust the framework parameters according to the parameters of different delivery scenarios.

[0070] S6: Apply the evaluation framework to large-scale delivery decisions in low-altitude logistics systems, use the framework output as decision input, and optimize delivery routes and resource allocation under large-scale deployment conditions through iterative adjustments.

[0071] S1: Collect multi-source data from the low-altitude logistics system, use the collected multi-source data as input, access, preliminarily classify and store it through a unified interface to form a unified data foundation. The specific implementation is as follows:

[0072] First, a data acquisition scheme is configured based on the low-altitude logistics scenario. Real-time position coordinate data provided by the satellite navigation system is used as the core positioning input, including latitude, longitude, altitude, and velocity vectors, to depict the UAV's trajectory in low-altitude space. Simultaneously, meteorological data such as wind speed, wind direction, rainfall, temperature, humidity, and atmospheric pressure collected by meteorological monitoring stations or remote sensing platforms are used as environmental input to reflect the impact of low-altitude meteorological changes on flight safety and delivery timeliness. Furthermore, flight status and environmental perception data generated by the sensors onboard the UAV are used as real-time feedback input, including at least acceleration, attitude angle, battery level, vibration level, and obstacle distance, to monitor real-time flight conditions and the surrounding environment. Finally, historical operational records such as average delivery time, mission success rate, and load utilization, as well as fault logs including equipment failure types, frequency of occurrence, and repair time, are extracted from the operation management system or log database as historical input. This ensures comprehensive coverage of the collected multi-source data across four dimensions: positioning, environment, real-time status, and historical operation.

[0073] Based on multi-source data acquisition, standardized data access is achieved through a unified API interface. This interface employs a modular architecture, supporting import, parsing, real-time push, or batch retrieval of data from different data sources according to preset communication protocols and data formats. Internally, a data verification layer is set up to sequentially perform format checks, numerical range verification, and integrity checks on the input data. This verifies whether the location coordinates are within a valid latitude and longitude range, whether meteorological parameters are within a reasonable physical range, whether sensor data is missing or exhibits abnormal fluctuations, and whether historical records are continuous on the timeline. It also filters invalid or abnormal inputs. The numerical range verification is performed based on preset range thresholds for different data types, determined by combining sensor range, physical limit parameters, and historical statistical data. Simultaneously, a buffering mechanism is configured on the interface side, using a queue structure to temporarily cache and queue data arriving during peak periods, thereby avoiding congestion caused by a large amount of instantaneous data writing and ensuring the stable operation of the acquisition link.

[0074] After standardized access is completed, the multi-source data entering the system is initially classified: spatial location-related data is classified into a location group for constructing time-stamped trajectory sequences; meteorological parameters are classified into an environment group for characterizing the impact of the external environment on mission execution; flight status data and environmental perception data generated by airborne sensors are classified into a real-time group for supporting online monitoring and anomaly detection; and operational efficiency indicators and fault logs are classified into a historical group for forming a basis for long-term performance and reliability assessment. On the other hand, metadata tags are uniformly generated for each group of data, including data source identifiers, collection timestamps, and priority markers, to support subsequent data retrieval, filtering, and fusion operations, and to avoid confusion between data from different sources during use.

[0075] Subsequently, the data from each group is structured and stored in the cloud storage platform. The cloud storage platform employs a distributed storage architecture, allocating corresponding storage areas for different data types: location data is stored in a high-availability partition supporting geographic index queries, used for retrieving trajectory information by spatial range and time interval; environmental data is stored in a storage engine adapted for time-series data processing, used for analyzing meteorological change patterns along the time axis; real-time data is stored in a low-latency storage area combining memory caching and persistent disks to meet online access requirements; historical data is stored in a relational database, recording efficiency indicators and fault records through table structures, and establishing relationships between different tables using primary and foreign keys to support multidimensional statistical analysis. Furthermore, during data writing and access, sensitive fields, including location coordinates, are encrypted and access-controlled, allowing only authorized business modules and users to decrypt and read them. Regular backups and off-site redundant storage strategies are configured to improve data integrity and disaster recovery capabilities. Simultaneously, an audit log function is enabled to record operational information and timestamps at each stage of data collection, access, classification, and storage, achieving traceable management of the entire data process.

[0076] By collecting, accessing, grouping, managing, and storing the aforementioned multi-source data, a reliable data foundation is built for large-scale deployment scenarios of low-altitude logistics systems. This organizes the scattered raw data into an orderly and searchable resource library, thereby eliminating the problems of data silos and inconsistent formats in the traditional decentralized collection model.

[0077] S2: The collected multi-source data is fused, aligning the data from different sources according to timestamps and spatial coordinates, and integrating them into a unified logistics status dataset through a preset weight allocation mechanism. The specific implementation is as follows:

[0078] First, various data types are aligned along the time dimension. Using the real-time timestamp of satellite navigation data as the baseline time axis, meteorological monitoring data, real-time sensor data, and historical operational data are synchronized and matched. For meteorological monitoring data, it is checked whether its update time falls within a preset time window centered on the navigation timestamp. If it is not within the time window, the most recent valid meteorological record is retrieved forward or backward and associated with the corresponding navigation point, so that each location point has the corresponding weather conditions. For real-time sensor data and historical operational data, the same time window matching strategy is adopted, pairing the sensor acquisition time with the operational log time, and associating the current flight status with the efficiency indicators and fault records of the corresponding time period to form a multi-source sequence on a unified time axis, thereby avoiding time misalignment and decision bias caused by different sampling periods and update frequencies.

[0079] Secondly, based on time alignment, spatial coordinate mapping is performed on various types of data to unify them under the same geographic reference system. Firstly, the WGS84 coordinate system is adopted as the unified reference system, with the latitude and longitude provided by satellite navigation directly used as spatial anchor points. For meteorological monitoring station locations, when expressed in local coordinates or administrative grids, a coordinate transformation step is performed to convert them to the corresponding global latitude and longitude, based on preset coordinate system transformation parameters or a transformation model provided by the geographic information system. For the relative azimuth or distance information of the sensor real-time data relative to the UAV body, it is superimposed on the navigation coordinates to obtain the corresponding absolute position. For path records in historical operational data, backtracking is used to map historical trajectories to the current reference frame, enabling comparative analysis of historical paths and current operating trajectories in the same three-dimensional space. During spatial mapping, altitude or flight altitude is used as the third dimension to compare altitude information from different data sources. When the altitude difference exceeds a preset altitude difference threshold, intermediate points are supplemented by interpolation or smoothing. The preset altitude difference threshold can be set by combining the drone's altitude control accuracy, mission safety margin, and historical altitude fluctuation statistics to eliminate local gaps and abrupt changes, thereby constructing a spatial view that is consistent in both planar position and altitude dimensions.

[0080] After time and space alignment is achieved, a weighting mechanism is introduced. The reliability of each data source is evaluated based on indicators such as positioning accuracy, data freshness, data integrity, and historical sample size, and a relative weight is determined accordingly. Specifically, satellite navigation data is considered the data source with the highest positioning accuracy and is given a higher priority weight. For meteorological data, the weight is adjusted based on whether the interval between its update time and the current time is within a preset freshness threshold; data updates that are timely are given a higher weight, while the weight is gradually reduced for longer update intervals to reflect differences in timeliness. For real-time sensor data, the weight is adjusted based on whether the data packet loss rate is below a preset loss rate threshold; the weight is increased when the loss rate is low and data continuity is good, and decreased when the loss rate is high. To mitigate the impact of noise, historical operational data is weighted based on whether the number of historical records exceeds a preset sample size threshold. The weight is increased when the sample size is sufficient and appropriately decreased when it is insufficient to avoid statistical bias in small samples. Furthermore, the relative weight relationships between navigation data, meteorological data, sensor data, and historical data are determined, and the weights of meteorological data or other data sources can be adjusted appropriately according to different operational scenarios, forming a dynamically configurable weighted integration basis. The preset time window, preset altitude difference threshold, preset freshness threshold, preset loss rate threshold, and preset sample size threshold can be configured based on the sampling period of various data types, sensor range, operational security redundancy requirements, and historical operational statistical characteristics, and are not limited to specific values.

[0081] Subsequently, the aligned data is hierarchically integrated using predetermined weights, gradually merging data from different sources into a unified logistics status dataset. First, navigation data, aligned in both time and space, is fused with real-time sensor data to construct a core flight status subset. Navigation data provides the flight path and spatial position framework, while sensor data supplements dynamic details such as attitude changes, vibration levels, battery status, and obstacle distances. Based on this, meteorological data is integrated into the core subset, mapping environmental parameters such as wind speed, wind direction, rainfall, temperature, and humidity to various time and space points, forming an enhanced environment subset describing the coupling relationship between location and environment. Finally, historical operational data is introduced into the above data structure, associating historical efficiency indicators and fault records with the current path and status. When the current flight position or path segment overlaps with historically high-fault or low-efficiency areas, risk or performance labels are marked on the corresponding records. Through this hierarchical integration method, a comprehensive logistics status dataset is constructed that is integrated across four dimensions: location, environment, real-time status, and historical experience. Each data source participates in the integration according to preset weights, avoiding bias caused by a single dominant data source while maintaining the dominant role of high-quality information in the overall dataset.

[0082] Finally, the output of the comprehensive logistics status dataset is used as the input data source for cost model and benefit index calculation. The feature fields such as path characteristics, weather risk, equipment status and historical efficiency are provided to the subsequent modeling steps, so as to form a smooth connection between multi-source data fusion and economic evaluation, and ensure the continuity and consistency of low-altitude logistics deployment decisions between the data layer and the model layer.

[0083] S3: Based on the fused logistics status dataset, establish a cost model for large-scale deployment scenarios, parameterize various cost variables, and establish a correlation mapping relationship with deployment scale and operation strategy. The specific implementation is as follows:

[0084] First, based on the comprehensive logistics status dataset, the equipment procurement cost variable is defined as a core component of the initial capital expenditure. By analyzing real-time sensor data, the specification requirements of the UAV body, navigation module, and auxiliary sensors are derived. Combined with equipment usage records in historical operational data, the configuration quantity and reference unit price of different types of equipment are determined. Accordingly, the equipment procurement cost is expressed as a weighted sum of the quantity and unit price of various types of equipment, so that the cost can directly reflect the current deployment scale and configuration structure, avoiding static estimates that are detached from the actual operational scale.

[0085] Secondly, maintenance costs are introduced into the cost variable system to characterize the inspection and maintenance expenditures during the long-term operation of the system. Based on the environmental and real-time groups of the comprehensive logistics status dataset, factors that affect maintenance needs are extracted. For example, the frequency and duration of severe weather are statistically analyzed from meteorological data, and the number of occurrences of events such as abnormal vibration levels and abnormal temperatures are statistically analyzed from sensor data. These are then correlated with the fault logs in the historical group to distinguish between regular maintenance and fault repair activities. On this basis, maintenance costs are modeled as a function that varies with operating time, with cumulative operating time or years of operation as independent variables. This allows the expected values ​​of maintenance frequency and single maintenance cost to increase moderately over time, reflecting the impact of equipment aging and fatigue accumulation on maintenance costs under large-scale deployment conditions.

[0086] This model incorporates network expansion costs as an infrastructure cost variable. Based on the location group of the comprehensive logistics status dataset, it statistically analyzes the spatial coverage and task density of low-altitude logistics tasks to deduce the deployment density requirements for communication base stations, edge nodes, or relay equipment. Combined with channel quality-related parameters in the environment group, it assesses the redundancy required for data transmission links. Furthermore, it utilizes network failure rates and link interruption records from historical operational data to adjust the configuration requirements for network reliability and backup links. Consequently, network expansion costs are decomposed into sub-items such as base station construction costs, link leasing or bandwidth usage costs, and network equipment upgrade costs. Each sub-item is modeled based on variables such as coverage area, node density, node unit cost, and service traffic. For example, base station construction costs can be expressed as the number of nodes determined by the coverage area and node density multiplied by the single-node construction cost; link leasing or bandwidth usage costs can be expressed as a function of service traffic and bandwidth unit price. This demonstrates that network expansion costs show a significant upward trend with the expansion of the deployment area and the increase in task volume, thus reflecting changes in network investment under cross-regional and large-scale operations.

[0087] Furthermore, human resource input costs are incorporated as operational variables into the cost model. Based on the flight status, alarm events, and scheduling records centrally recorded in the comprehensive logistics status dataset, the workload requirements for monitoring and scheduling are assessed. Based on historical efficiency indicators, task completion quality and error rates are analyzed to determine the training needs of maintenance and operation personnel. Regarding human resource costs, operator salaries, training expenses, and scheduling personnel configuration are broken down into several sub-variables and their dependencies are established with variables such as maintenance costs and task complexity. For example, when the failure frequency statistics exceed a preset failure frequency threshold, the training frequency is increased or the number of on-duty personnel is increased. The preset failure frequency threshold can be set in conjunction with historical failure frequency distribution and safety management requirements, thereby enabling human resource input costs to be dynamically adjusted according to the system's health status and risk level.

[0088] After defining various cost variables, they are parameterized to transform the cost structure into a computable form. When parameterizing cost variables, the same evaluation period and unified aggregation granularity are used, decomposing the evaluation period into continuous periods t, and outputting the corresponding period t for each cost output. A parameter set is constructed for each cost sub-item. Fixed in the settings: Q is the scale parameter in period T, representing the countable scale such as the number of devices, personnel, network nodes, and business volume. The statistical caliber of the scale parameter is based on the system metering caliber and the contract billing caliber; P is the unit cost parameter, representing the unit purchase cost, unit maintenance cost, unit rental cost, unit bandwidth cost, unit hourly wage, etc. The unit cost parameter is provided by the contract, quotation, or budget sheet; T is the occurrence or settlement cycle parameter, representing the time such as the maintenance cycle, settlement cycle, or amortization cycle; F is the number of occurrences per cycle with reference to period T. The frequency value is derived from the maintenance plan, training plan, settlement contract, or operation rule table; To implement the cycle and frequency into the actual number of occurrences within the period t, the event count formula is defined as:

[0089]

[0090] Where Δt is the duration of time segment t and its unit is consistent with T. This indicates rounding up. Rounding up is used to avoid underestimating the number of occurrences when aggregating across periods. The rounding rules and the sources of T and F are written into the configuration so that the number of events in any period can be directly determined by Δt, T, and F.

[0091] Under a unified standard, each cost variable is expressed as quantity Q, unit price P, period T, and frequency F, and uniformly aggregated according to period t. The cost of purchasing equipment is a one-time investment, and is calculated according to the equipment scale parameters during the period in which the purchase occurs. Unit price parameters The equipment purchase cost is obtained by multiplying and summing the results; when the equipment purchase cost is depreciated or amortized, the one-time purchase cost is calculated according to the amortization period. Linear amortization is performed to obtain the periodic procurement amortization costs for each period; maintenance costs within period t are obtained by multiplying the cost per maintenance by the number of maintenance operations in that period, with the cost per maintenance determined by the unit price parameter. Confirmed, the number of maintenance sessions is based on the maintenance cycle. and frequency The conversion formula is:

[0092]

[0093] in, The number of maintenance events within period t is the number of maintenance occurrences that should be included in the cost calculation during this period. The periodic maintenance cost is obtained by summing the costs of scheduled maintenance and fault repair separately. Network costs within period t consist of expansion construction costs and operation leasing costs. Expansion construction costs are determined by the parameter of the scale of newly added nodes. With node unit price parameter The product of bandwidth or traffic volume parameters, the operating lease cost is the cost of operating the lease. With bandwidth unit price parameter The product; network fees are calculated on a per-settlement basis. During billing, the billing cost is allocated to each period according to the proportion of the covered time period and then aggregated; among which, the cost of human resources input is calculated according to the personnel size parameter within period t. With average annual salary or average hourly rate parameter Multiply to obtain; training costs are included as an independent sub-item in human resource input costs, based on the parameter of the number of trainees. Unit price parameters for a single training session The formula for calculating the number of initial training sessions is: (Multiply by the number of training sessions in the current period).

[0094]

[0095] in, The number of maintenance events within period t, i.e., the number of maintenance events that should be included in the cost calculation within this period; t is the period number or period identifier, used to refer to a certain period of time after the evaluation period has been divided; The duration of period t, the time unit must be consistent with... Maintain consistency; The maintenance cycle indicates the length of the period on which maintenance activities are conducted or on which settlements are based. In order to complete a training cycle Training frequency, based on the standard, represents the number of training sessions that occur within each training cycle; This indicates rounding up; thus, labor-related costs can be determined and calculated based on changes in job configuration, working hours, training cycle, and training frequency; thereby obtaining the equipment, maintenance, network, and labor itemized period costs and total period costs aggregated by period t, which are used for subsequent net cash flow series construction;

[0096] Subsequently, based on parameterization, a correlation mapping relationship between variables is established, linking each cost variable with deployment scale and operational strategy in a linked model. By introducing scale factors such as the number of drone clusters, average daily workload, and service area, equipment procurement costs, network expansion costs, maintenance costs, and manpower input costs are expressed as functions of scale factors. Simultaneously, based on operational strategy parameters such as communication redundancy levels, inspection frequency, and task scheduling strategies, the corresponding cost variables are adjusted. For example, equipment procurement costs are linked to network expansion costs; when the number of newly added drones exceeds a preset scale threshold, the number of base station nodes or bandwidth resources are increased accordingly. Maintenance costs are linked to manpower input costs; when the failure frequency or environmental risk indicators exceed a preset safety threshold, expected maintenance expenditures and manpower allocation levels are increased. The scale threshold and preset safety threshold can be set based on system design capacity, resource upper limit constraints, industry safety standards, and historical operational statistics, and are not limited to specific values. Through the above correlation mapping, a network structure reflecting the interaction between cost variables and their interaction with deployment scale and operational strategy is constructed, ensuring that the total cost is not merely a simple summation of individual costs but also reflects the coupling effects across variables.

[0097] Based on the establishment of variable definitions, parameterized expressions, and correlation mappings, a cost model framework for large-scale deployment scenarios is constructed. The bottom layer consists of cost variables and their parameters, the middle layer describes the linkage between variables and deployment scale and operation strategy through mapping relationships, and the upper layer forms a total cost function and sub-cost views divided by dimensions such as equipment, network, maintenance, and manpower. The deployment scale parameter is used as an adjustment factor to support the smooth expansion of the model from small-scale pilot to large-scale normalized operation. The output of this cost model serves as the input basis for subsequent benefit analysis steps to further calculate various benefit indicators, so that the cost modeling results can directly serve the large-scale deployment decision of low-altitude logistics systems.

[0098] S4: Construct net cash flow based on the period cost and period revenue series, discount it to obtain net present value, and iteratively calculate the internal rate of return. Under threshold constraints, use task logs to statistically analyze efficiency and resource utilization indicators. The specific implementation is as follows:

[0099] First, based on the cost model established above, the evaluation period and unified statistical granularity are determined, and cost collection, revenue recognition, task log statistics and indicator calculation are limited to the same period and the same granularity. The discount rate ρ, threshold set θ, risk weight table and load balancing mapping rules are fixed as versioned configurations and remain unchanged within the evaluation period. Subsequently, each indicator can be directly recalculated from the period cost series, period revenue series and task log.

[0100] From cash flow construction to cash flow aggregation, the cash outflows within the maturity cost sequence output by the cost model must include at least initial equipment investment, periodic maintenance investment, network expansion and operation investment, and human resource investment. The timing and aggregation method are clearly defined: one-time investments are included in cash outflows from the start of the assessment period, and periodic investments are included in cash outflows from the period in which they occur. If there are inter-period amortization and depreciation, the amortization caliber is reflected in the cash outflows of each period according to the amortization caliber of the cost model. The period revenue sequence is constructed by extracting revenue-related operational volumes from the comprehensive logistics status dataset. Period cash inflows are determined by the number of tasks completed in the current period and the billing rules. The billing rules include task type, billing unit, billing unit price, and settlement caliber. Revenue can be verified by retrospectively reviewing each task invoice. The period net cash flow sequence is constructed from the period revenue sequence and the period cost sequence. The period net cash flow is the current period cash inflow minus the current period cash outflow, with a unified input for the return on investment indicator.

[0101] To ensure that judgments such as "effective completion, success, availability, and rejection" have executable boundaries, a threshold set θ, along with its source and calibration, is predefined and used as a judgment and statistical constraint throughout the entire process. The threshold set θ should contain at least a subset of communication thresholds θc, a subset of energy thresholds θe, a subset of airspace and compliance thresholds θr, a subset of safety margin thresholds θs, and a subset of log integrity thresholds θd. The communication thresholds define link quality and disconnection criteria, and should contain at least one or a combination of an upper limit for continuous disconnection duration, an upper limit for packet loss rate, or a lower limit for link quality. The energy thresholds define takeoff and execution safety margins, and should contain at least a minimum takeoff battery threshold, a mission safety margin threshold, and an upper limit for energy prediction error. The airspace and compliance thresholds define boundary crossing and restricted flight constraints, and should contain at least an upper limit for boundary crossing distance, an upper limit for boundary crossing duration, and a restricted flight buffer zone. Thresholds; security margin thresholds are defined to reserve redundancy on the basis of compliance baselines to cover environmental disturbances and ensure security; log integrity thresholds are defined to define rules for missing key fields and their exclusion, and must contain at least the criteria for determining missing key fields, the allowable proportion of missing fields, or the criteria for the duration of missing fields; the generation of the threshold set θ is based on the determination process of rule baseline plus data calibration. First, the threshold θreg is set by the compliance rule base and enterprise security policy. Then, θcal is obtained by combining the successful and failed task samples marked in the historical operation data. The failure omission threshold is selected with an acceptable omission rate constraint as the threshold. Finally, the security margin is superimposed on the threshold to obtain the final threshold set θ; the threshold set θ records the calibration data window, calibration criteria, and effective window with version number, so that the evaluation results can be traced back to the source and judgment boundary of the threshold;

[0102] For the return on investment (ROI) metric, the net present value (NPV) and internal rate of return (IRR) are calculated using the net cash flow series over the investment period and the discount rate parameter ρ, respectively. The discount rate ρ is written into a versioned configuration according to pre-defined rules. These pre-defined rules can be the enterprise's cost of capital, the project's target return, or a unified discount benchmark combining the risk-free rate and a risk premium. The source of the discount rate is recalcible. The NPV is calculated by discounting and summing the net cash flows of each period in the evaluation period according to the discount rate ρ, resulting in a discounted net income, which is then matched with the initial investment calculation to obtain a single NPV. The IRR is calculated by obtaining the discount rate at which the discount is zero, using a reproducible numerical solution method: first, a return search interval is set... The sign of the net present value function changes at the endpoints of the search interval to ensure the existence of a root. Within the interval, the interval is iteratively narrowed to approximate the root, yielding a return estimate. The calculation ends when the absolute value of the net present value is less than a preset error threshold or the difference between two consecutive return updates is less than a preset error threshold, and a maximum number of iterations is set. If the sign change does not satisfy the condition within the search interval, resulting in no root or multiple roots causing a non-unique solution, an anomaly marker is output according to pre-defined rules, and the available solution strategy result is returned. This could be returning the smallest positive root or "no available internal rate of return," while retaining the corresponding net present value curve diagnostic information. This ensures that the internal rate of return result has a verifiable solution path and anomaly handling boundaries.

[0103] The key metrics are unit delivery time and task success rate. Unit delivery time is recalculated based on statistical objects and judgment rules. A valid task set is defined as follows: a task enters the execution state and is checked for critical log fields based on a log integrity threshold subset θd. At least the following fields are required: scheduling issuance timestamp, execution state entry timestamp, execution state exit timestamp, track or execution segment identifier, and delivery confirmation receipt identifier. Tasks lacking critical fields are filled in or removed according to θd, and the reason for removal is recorded, ensuring the valid task set can be reconstructed by others using the same rules. Unit delivery time is measured by the task loop duration. The starting point of the loop duration is the timestamp when the scheduling system generates executable instructions and issues them to the cluster, and the ending point is the timestamp when the task reaches the target point and completes delivery confirmation. For tasks involving relays or transfers... Tasks involving aircraft, charging queue jumping, or multiple flight segments are not recalculated in terms of duration; instead, they are timed uniformly at the end of a closed loop, and direct comparisons are made between tasks. The unit delivery time cycle statistics use the statistical mean of the closed-loop duration of the effective task set as the main output, which can output the median and high percentile. The main indicator is a single value, and comparisons are made across different scenarios. The task success rate is measured by the percentage of tasks that complete closed-loop delivery under the constraints of a threshold set θ. The task success rule is: any task termination at any time corresponds to any threshold among θc, θe, θr, and θs. Triggering the threshold results in a failed or terminated task; tasks that do not trigger the threshold but have a valid delivery confirmation receipt are considered successful. The task success rate is the percentage of successful tasks out of the total number of tasks entering the execution state, outputting a probability from zero to one, and corresponding to the threshold version for verification.

[0104] To reflect environmental differences without changing the main indicator's caliber, a risk-weighted supplementary indicator can be created and a determination mechanism provided. The risk score R can be obtained from multi-source data, and can include at least meteorological level, airspace restriction level, terrain complexity, communication coverage quality, and task density. After unification and standardization of dimensions, the risk score is obtained by weighting and summing the data according to the configured weights. The risk tier threshold is determined by the quantile of the historical risk score distribution or backtesting results. The tier threshold and tiering rules are set using versioning. The weight coefficient W for each risk tier is configured to solidify its source. The weight coefficient can be determined from the initial value of the business strategy and adjusted through backtesting with the goal of minimizing weighted delays and failures to finally obtain the weight table. The weighted success rate or weighted unit delivery time is obtained from the weight table as a supplementary output. The weighted success rate is the ratio of the sum of successful tasks and total tasks weighted by risk weights. The supplementary output is still a probabilistic output and can be directly calculated from the risk score and weight table.

[0105] The idle time ratio of drones is determined primarily based on idle time proportion and load balancing, defining its statistical boundaries, deduction rules, and mapping function. The drone idle time ratio is described as the percentage of available time spent not participating in tasks. Available time is the remaining time after deducting planned maintenance periods, downtime due to malfunctions, and non-flying periods specified by regulations or policies from the total evaluation period. Planned maintenance periods are determined by maintenance plans and records; downtime due to malfunctions is determined by fault logs and state machine records; and non-flying periods are determined by airspace and operational rules. Deductions are all constrained by corresponding criteria in the threshold set θ, and abnormal labeling should not be expanded. Task execution time is the cumulative duration of a drone in execution mode. The idle time ratio of a single drone is the sum of its non-execution time and available time. The idle time ratio of the drone fleet is the weighted average of the idle time ratios of each drone based on available time, eliminating the influence of unused time. The statistical bias is caused by the difference in available time of the same drone; the load balance is measured by whether the load distribution in the fleet is uniform, and a uniform load scale is selected and written into the configuration. The load scale is selected as the main scale of the number of tasks, flight time or payload; during the evaluation period, the main scale load of each drone is counted, and the relative dispersion of the load sequence is calculated. The relative dispersion is expressed as the ratio of the load standard deviation to the average load; the degree of imbalance is mapped to the balance score, and a monotonically decreasing mapping function is determined and made public. For example, a monotonically decreasing function with the balance greater than the relative dispersion is output, so that the larger the dispersion, the smaller the balance and the mapping result is within the preset interval; when it is necessary to normalize the balance to a fixed interval, the normalization boundary is determined by the upper and lower bounds or quantile cutoff points of the dispersion statistically obtained from the historical window and written into the configuration, because outliers will cause interval stretching and the mapping can be reproduced.

[0106] Finally, the system outputs a set of multi-dimensional benefit indicators, including net present value, internal rate of return, unit delivery time, task success rate, drone idle time ratio, and load balancing degree, and outputs versioned configurations, including evaluation period and granularity, threshold set θ version, risk score and weight table version, and load balancing mapping rules, which are recalculated from period cost, period revenue sequence, and task logs.

[0107] S5: Integrate cost models and benefit indicators to form an evaluation framework. Within this framework, link all elements into a closed-loop structure. Adjust the framework parameters according to different delivery scenarios. Specific implementation details are as follows:

[0108] First, based on the cost model and calculated benefit indicators established above, the evaluation elements are aggregated. The cost model's equipment procurement cost, maintenance cost, network expansion cost, and human resource input cost are correlated with benefit indicators such as return on investment, unit delivery time, task success rate, idle time ratio, and load balancing, forming a set of elements covering multiple dimensions of finance, efficiency, and resource utilization. Equipment procurement cost is linked to net present value and internal rate of return to measure the impact of different equipment investment levels on long-term returns. Maintenance cost is linked to unit delivery time and task success rate to reflect the constraint of maintenance investment on operational efficiency and reliability. Network expansion cost is linked to unit delivery time and task success rate to demonstrate the supporting role of network coverage and link quality in timeliness and success rate. Human resource input cost is linked to idle time ratio and load balancing to characterize the impact of personnel allocation on resource utilization and task allocation balance.

[0109] Secondly, based on the aggregation of elements, a linked structure for the evaluation framework is constructed, with each cost variable as an input node and each benefit indicator as an output node, establishing multiple impact paths from cost to benefit. Each path is assigned a pre-set weight or influence coefficient within the evaluation framework to describe the importance of a particular cost variable to the relevant benefit indicator. For example, in scenarios emphasizing safety and stability, the weight of the path between maintenance cost and task success rate, and load balancing, can be increased; in scenarios emphasizing rapid investment recovery, the weight of the path between equipment procurement cost and net present value, and internal rate of return, can be increased. These weights or influence coefficients can be determined by combining enterprise operational strategies, regulatory requirements, and historical evaluation results, through expert scoring, simulation analysis, or data fitting, without being limited to a single setting method. Through the above organization of input nodes, output nodes, and path weights, the originally scattered cost and benefit elements are integrated into a structured network-based evaluation framework, providing a foundation for subsequent quantitative scoring and comprehensive ranking.

[0110] A closed-loop feedback mechanism is introduced on top of the linked structure to give the evaluation framework self-correcting capabilities. Target ranges or reference ranges are set for each benefit indicator; for example, minimum target ranges are set for net present value (NPV) and internal rate of return (IRR), maximum acceptable time limits are set for unit delivery time, minimum safety requirements are set for task success rate, maximum acceptable upper limits are set for idle time ratio, and minimum balance levels are set for load balancing. When the evaluation results show that an indicator deviates from its target range, the evaluation framework generates corresponding adjustment suggestions based on the direction and magnitude of the deviation and feeds them back to relevant cost variables or path weights. For example, when NPV is lower than the target range, it is recommended to appropriately reduce equipment procurement in the next round of the plan. Scale up or optimize equipment selection; when unit delivery time exceeds the acceptable time limit, it is recommended to increase maintenance budget or optimize network expansion investment; when the task success rate is lower than the security requirements, it is recommended to prioritize increasing network expansion and maintenance investment; when the idle time ratio is higher than the upper limit or the load balancing is lower than the target level, it is recommended to adjust manpower allocation or task scheduling strategy to improve resource utilization; the target range, time limit, requirements and upper limit can be determined in combination with project planning goals, service level agreements and historical statistical data, and are not limited to specific values; through the above closed-loop feedback, the evaluation framework can not only provide the evaluation results of the current solution, but also provide a clear direction for adjustment for the next round of configuration optimization;

[0111] Subsequently, delivery scenario parameters are integrated into the evaluation framework as dynamic adjustment factors, enabling it to automatically adjust weights and target settings based on scenario differences. Scenario parameters include, but are not limited to, city density, typical flight distance, and fleet size. City density can be derived from geographical information and order distribution in the comprehensive logistics status dataset, typical flight distance can be statistically analyzed from historical trajectories and planned routes, and fleet size can be determined based on the number and type of currently deployed drones. In high-density urban scenarios, the evaluation framework can appropriately increase the weight of paths related to maintenance costs and task success rates, and tighten the target range for task success rate and unit delivery time based on risk levels to strengthen the focus on equipment reliability and operational safety. In long-distance delivery scenarios, the weight of paths related to network expansion costs, unit delivery time, and task success rates can be increased to emphasize long-distance communication and continuous coverage capabilities. In large-scale fleet scenarios, the weight of paths related to manpower input and idle time ratio, load balancing, etc., can be increased to strengthen the consideration of resource utilization and task allocation balance. Through dynamic adjustment of path weights and target ranges, the evaluation framework can smoothly switch between small-scale pilot projects and large-scale routine operations.

[0112] Furthermore, to adapt to the deployment needs of different construction phases, a tiered configuration model is introduced into the evaluation framework, corresponding to different weight combinations and evaluation focuses for the testing phase, the medium-scale operation phase, and the large-scale deployment phase. In the testing phase, the weight of return on investment indicators is appropriately reduced, while the weight of operational efficiency and task success rate is increased, focusing on verifying technical feasibility and operational stability. In the medium-scale operation phase, the weights of cost and benefit paths are balanced, allowing the framework to simultaneously consider cost control, efficiency performance, and resource utilization. In the large-scale deployment phase, the weights of financial indicators such as net present value and internal rate of return are increased, and the adjustment strength of related feedback paths is enhanced to further highlight economic efficiency and economies of scale. Through tiered configuration, the evaluation framework can cover the needs of different stages from pilot verification to full-scale rollout, avoiding the problem of insufficient applicability of a single configuration throughout the entire lifecycle.

[0113] By aggregating the aforementioned elements, constructing a linking structure, introducing closed-loop feedback, and comprehensively applying scenario parameters and stage configurations, a cost-benefit evaluation framework for large-scale deployment of low-altitude logistics systems is formed. This framework integrates cost model outputs and multi-dimensional benefit indicators within a unified structure. It can provide a comprehensive cost-benefit evaluation result for the current deployment plan, generate targeted adjustment suggestions based on evaluation deviations and scenario changes, and adaptively adjust with changes in the operational stage. This provides a systematic and quantitative decision-making basis for deployment plan selection, delivery route planning, and resource allocation optimization.

[0114] S6: Apply the evaluation framework to large-scale delivery decisions in low-altitude logistics systems, using the framework output as decision input, and iteratively adjust to optimize delivery routes and resource allocation under large-scale deployment conditions. Specifically, the implementation is as follows:

[0115] Based on the cost-benefit assessment framework constructed above, key quantitative results for decision-making are extracted from the assessment framework. Financial indicators such as net present value and internal rate of return are used as inputs for economic feasibility and expected returns. Operational indicators such as unit delivery time and task success rate are used as inputs for timeliness and operational reliability. Resource indicators such as idle time ratio and load balancing are used as inputs for resource utilization and fleet scheduling balance. At the same time, the dominant variables in the cost model, such as equipment procurement cost, maintenance cost, network expansion cost, and human resource input cost, as well as scenario parameters such as city density, typical flight distance, and fleet size, are combined to form a multi-dimensional input set for large-scale deployment decisions.

[0116] After obtaining the above input set, an initial delivery plan is generated for the given order demand set and low-altitude transportation network. The initial plan includes at least several delivery routes, the execution frequency of each route, the task allocation of different types of drones, and the configuration of take-off and landing points and relay nodes, while meeting basic constraints such as flight safety regulations, battery endurance, and no-fly zones. Subsequently, the cost expenditure data and operational performance data corresponding to the initial plan, along with scenario parameters, are input into the evaluation framework to calculate the overall cost-effectiveness of the plan. The calculated net present value, internal rate of return, unit delivery time, task success rate, idle time ratio, and load balance are compared with the target ranges set above for net present value and internal rate of return, maximum acceptable unit delivery time, minimum task success rate, maximum idle time ratio, and minimum load balance. The deviation values ​​of each indicator are obtained as the basis for subsequent adjustments.

[0117] Based on the evaluation results, an iterative adjustment mechanism is introduced to continuously optimize delivery routes and resource allocation schemes driven by the aforementioned deviations. When the net present value or internal rate of return is lower than the preset financial target range, the total cost is reduced or revenue expectations are increased by shortening low-value long routes, merging redundant routes, adjusting take-off and landing point layouts, or optimizing order aggregation methods. When the unit delivery time exceeds the service time limit, the overall transportation time is compressed by adjusting the route topology, adding relay nodes, increasing the execution frequency of critical routes, or allocating aircraft with higher flight performance to time-sensitive tasks. When the mission success rate is lower than the safety requirements, route segments with severe weather conditions, poor communication quality, or frequent historical failures are eliminated, and time periods and routes with higher stability are selected first. When the idle time ratio is too high or the load balance is too low, tasks are reallocated, migrating tasks from low-load or long-term idle drones to task-concentrated areas, or adjusting the task volume and flight duration among the fleet to achieve more balanced resource utilization. The specific route and resource optimization process can be implemented using heuristic search, integer programming, or simulation-based evaluation methods.

[0118] After each round of adjustments, the updated path planning scheme and resource allocation scheme are re-entered into the evaluation framework to obtain a new round of cost-benefit indicators, which are then compared with the target range again. When the deviations of all key indicators fall within the preset allowable range, or when the number of iterations reaches the preset upper limit, the iteration is terminated, the current scheme is determined as the optimized scheme under the corresponding deployment conditions, and the scheme is solidified as the execution configuration for the large-scale deployment phase. The target range, risk threshold, and allowable deviation range can use the parameters set above, or be updated according to the project planning objectives, service level agreement, and the latest historical statistical data, and are not limited to fixed values.

[0119] During actual system operation, when significant changes occur in order spatiotemporal distribution, weather patterns, cost structure, or equipment status, the cost model output and benefit indicators can be recalculated based on the comprehensive logistics status data obtained from a new round of collection and fusion. The latest results are then input into the evaluation framework, and the iterative optimization process is restarted, allowing delivery routes and resource allocation to be dynamically adjusted according to changes in the operating environment. Through the above application steps, the cost-benefit evaluation framework is expanded from a static evaluation tool into a decision engine driving route planning and resource scheduling. Within a unified structure, indicators such as economy, timeliness, safety, and resource utilization are transformed into executable routes and scheduling schemes, thereby ensuring that the low-altitude logistics system achieves comprehensive optimization of controllable costs, achievable efficiency, and reasonable resource allocation under large-scale deployment conditions.

[0120] The solution in this embodiment first collects multi-source data from the low-altitude logistics system. Satellite navigation positions, meteorological monitoring information, flight status and environmental perception data from airborne sensors, as well as historical delivery efficiency and fault logs are used as the main inputs. Standardized access and preliminary classification are achieved through a unified API interface, and the data is stored in a structured manner on a cloud platform, forming a unified data foundation. Based on this, time and space alignment is performed on the multi-source data. Navigation timestamps are matched with meteorological update times, sensor data and operational records are mapped to a unified geographic reference system, and fused with preset weights to form a unified logistics status dataset. Further, a cost model for large-scale deployment is constructed based on this dataset. Cost variables such as equipment procurement, maintenance, network expansion, and manpower input are parametrically modeled and correlated with deployment scale and operational strategies. Return on investment, operational efficiency, and resource utilization are calculated to form a multi-dimensional set of benefit indicators. Subsequently, the cost model and benefit indicators are integrated into a closed-loop evaluation framework. The framework parameters are dynamically adjusted according to different delivery scenarios. This framework is invoked in low-altitude logistics delivery decisions, and the evaluation results serve as the basis for route planning and resource allocation decisions. Iterative optimization achieves cost-effectiveness unification under large-scale deployment conditions.

[0121] Example 2: Figure 2 This invention presents a low-altitude logistics delivery decision-making system based on multi-source data fusion, comprising:

[0122] Data acquisition module: Collects multi-source data from the low-altitude logistics system and performs standardized access, classification, and cloud storage through a unified API interface;

[0123] Data fusion module: It integrates the collected data by merging them, aligning them with timestamps and spatial coordinates, and applying a weighting mechanism to create a unified logistics status dataset.

[0124] Cost model building module: Based on the fused dataset, a cost model for large-scale deployment is established. Equipment procurement, maintenance, network expansion and human resources are used as cost variables. Parametric modeling is performed and correlation mapping is established to form a scalable cost framework, providing a quantitative basis for the generation of benefit indicators.

[0125] Benefit indicator generation module: Generates benefit indicators using the cost model, and quantifies each indicator based on the relationship between the above variables to form a multi-dimensional benefit indicator set;

[0126] Evaluation framework integration module: Integrates cost models and benefit indicators into a closed-loop evaluation framework, links various evaluation elements through feedback loops, and dynamically adjusts framework parameters according to scenario parameters to ensure that the evaluation framework adapts to different deployment needs;

[0127] Decision Application Module: The output of the evaluation framework is used as the input for delivery decisions. By iteratively adjusting and refining route selection and resource allocation, a business closed loop from data collection to decision optimization is formed.

[0128] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0129] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0130] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0131] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0132] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0133] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0134] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0135] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0136] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A low-altitude logistics distribution decision-making method based on multi-source data fusion, characterized in that, include: S1: Collect multi-source data from the low-altitude logistics system, configure a low-altitude logistics data collection scheme, and use satellite navigation position, meteorological elements, airborne sensor status and historical efficiency and fault data as multi-source input sources. Use the collected multi-source data as input sources, access, preliminarily classify and store it through a unified interface to form a unified data foundation. S2: The collected multi-source data is fused and processed, and the data from different sources are aligned according to timestamps and spatial coordinates, and integrated into a unified logistics status dataset through a preset weight allocation mechanism; S3: Based on the fused logistics status dataset, a cost model for large-scale deployment scenarios is established. The equipment procurement cost variable is defined by extracting hardware specification requirements from the fused dataset and combining configuration quantity and unit price. The maintenance cost variable is defined, and the maintenance frequency and maintenance expenditure type are extracted based on environmental data, real-time operation data and historical fault records. Based on location data, environmental data and historical data, the communication coverage density and infrastructure configuration parameters are statistically analyzed, and the network expansion cost variable is defined. The various cost variables are parameterized and modeled, and a correlation mapping relationship is established with the deployment scale and operation strategy. S4: Use the cost model to generate benefit indicators, taking return on investment, operational efficiency and resource utilization as the basis for calculating benefit indicators, and quantify the indicators through the relationship between variables; S5: Integrate cost models and benefit indicators to form an evaluation framework, link all elements into a closed-loop structure within the evaluation framework, and adjust the framework parameters according to the parameters of different delivery scenarios. S6: Apply the evaluation framework to large-scale delivery decisions in low-altitude logistics systems, use the framework output as decision input, and optimize delivery routes and resource allocation under large-scale deployment conditions through iterative adjustments.

2. The low-altitude logistics distribution decision-making method based on multi-source data fusion according to claim 1, characterized in that, Collect multi-source data from the low-altitude logistics system, use the collected multi-source data as input, access, initially classify and store it through a unified interface to form a unified data foundation, including: Standardized access is performed through a unified API interface. Input data is sequentially checked for format, numerical range and integrity. A queue buffering mechanism is configured to queue and cache data arriving during peak periods. The data is initially classified, and metadata tags containing data source identifiers, collection timestamps, and priority markers are generated for each group of data. In the cloud storage platform, location groups, environment groups, real-time groups, and historical groups are stored in geographically indexed partitions, time-series engines, low-latency storage areas, and relational databases, respectively. Sensitive fields are encrypted and access controlled, and data management is carried out in conjunction with backup strategies and audit logs.

3. The low-altitude logistics distribution decision-making method based on multi-source data fusion according to claim 1, characterized in that, The collected multi-source data is fused, and the data from different sources are aligned according to timestamps and spatial coordinates, including: Align multi-source data in the time dimension, using the timestamp of satellite navigation data as a benchmark, perform synchronous matching on meteorological monitoring data, real-time sensor data and historical operational data, and retrieve the most recent valid record of each and associate it with the corresponding navigation time point; After time alignment, spatial coordinate mapping is performed on the multi-source data. A unified geographic reference system is adopted to convert the location of meteorological monitoring stations, the relative orientation information of sensors, and historical path records into unified global latitude and longitude coordinates. Altitude is introduced as a third dimension to interpolate the height data. A weight allocation mechanism is introduced to assess the reliability of each data source to determine its relative weight, and to dynamically adjust the ratio between the weights according to the application scenario.

4. The low-altitude logistics distribution decision-making method based on multi-source data fusion according to claim 1, characterized in that, The data is integrated into a unified logistics status dataset through a pre-defined weight allocation mechanism, including: By using weights to hierarchically integrate aligned data, fusion of navigation and sensor data is used to generate flight status data, and meteorological data is overlaid and linked with historical records to generate risk markers. The integrated dataset output is used as input to the cost model, and historical feature fields of path meteorological equipment are provided. The fused data is used as a unified input for cost model building and benefit index calculation, and is processed continuously between the data layer and the model layer in a predetermined order.

5. The low-altitude logistics distribution decision-making method based on multi-source data fusion according to claim 1, characterized in that, The various cost variables are parametrically modeled and their correlation mapping relationships with deployment scale and operational strategies are established, including: Define the human resource input cost variable, and determine the monitoring positions, training cycles, and personnel configuration parameters based on flight status records and operational records; Parametric modeling of cost variables is performed, and each cost variable is expressed as a functional relationship with quantity, unit price, period and frequency as parameters; Establish correlation mapping relationships to associate each cost variable with the deployment scale and operation strategy, and construct a cost framework structure with total cost and sub-cost views.

6. The low-altitude logistics distribution decision-making method based on multi-source data fusion according to claim 1, characterized in that, The cost model is used to generate benefit indicators, with return on investment, operational efficiency, and resource utilization rate as the basis for calculation. These indicators are then quantified through the relationships between variables, including: Construct an investment return rate indicator, establish a cash flow series from cost variables and historical revenue data, and calculate net present value and internal rate of return; Construct operational efficiency metrics, extract relevant variables from the dataset output by the model, and calculate unit delivery time and task success rate; Construct a resource utilization index, combine scheduling records with the human resource input and equipment configuration variables in the cost model, calculate the idle time ratio of each drone, and calculate the load balance degree based on the dispersion of task allocation. The benefit indicators are calculated based on the functional relationships between variables, and the resulting multidimensional benefit indicator set is used as the input parameters of the evaluation framework.

7. The low-altitude logistics distribution decision-making method based on multi-source data fusion according to claim 1, characterized in that, The cost model and benefit indicators are integrated to form an evaluation framework. Within this framework, the various elements are linked into a closed-loop structure, and the framework parameters are adjusted according to parameters of different delivery scenarios, including: Cost variables and benefit indicators are aggregated to form an evaluation element set, and the corresponding correlation between equipment procurement costs, maintenance costs, network expansion costs, human resource input and return on investment, operational efficiency and resource utilization rate are established in the element set; The evaluation framework constructs a cost-benefit linkage structure, sets preset weights for the mapping paths between cost variables and benefit indicators, and adjusts the influence coefficients of the weights according to scenario parameters. A closed-loop feedback mechanism is introduced to set target ranges for each benefit indicator and adjust cost variables and path configuration parameters based on the deviation between the evaluation results and the target ranges. Scenario parameters are introduced into the evaluation framework to dynamically adjust the weights of each evaluation indicator, and the corresponding framework configuration is selected or updated based on city density, flight distance, and fleet size.

8. The low-altitude logistics distribution decision-making method based on multi-source data fusion according to claim 1, characterized in that, The evaluation framework is applied to large-scale delivery decisions in low-altitude logistics systems. The framework output is used as the decision input, and iterative adjustments are made to optimize delivery routes and resource allocation under large-scale deployment conditions, including: The quantitative evaluation results are extracted from the evaluation framework as decision inputs and combined with cost variables and scenario parameters to generate a multi-dimensional set of decision inputs. An initial delivery plan is generated based on order demand and the low-altitude transportation network. The data corresponding to the delivery plan is then input into an evaluation framework to calculate various benefit indicators and compare them with preset target ranges. An iterative adjustment mechanism is introduced in the decision-making process. Based on the evaluation deviation, the delivery route and resource allocation are cyclically optimized. The delivery route is reduced, merged and its layout is adjusted, and the order aggregation method is adjusted. The resource allocation is subject to task migration and fleet rotation operations. After each round of path and resource adjustments, the solution is re-entered into the evaluation framework to obtain metrics. The iteration is terminated when the deviation is within the allowable range or the number of iterations reaches the upper limit. When order distribution, environmental parameters, or cost parameters change during system operation, cost and benefit indicators are recalculated based on updated multi-source data. The recalculated indicators are then input into the evaluation framework, and the optimization steps of path planning and resource allocation are re-executed.

9. A low-altitude logistics distribution decision-making system based on multi-source data fusion, used to implement the low-altitude logistics distribution decision-making method based on multi-source data fusion as described in any one of claims 1-8, characterized in that, include: Data acquisition module: Collects multi-source data from the low-altitude logistics system and performs standardized access, classification, and cloud storage through a unified API interface; Data fusion module: It integrates the collected data by merging them, aligning them with timestamps and spatial coordinates, and applying a weighting mechanism to create a unified logistics status dataset. Cost model building module: Based on the fused dataset, a cost model for large-scale deployment is established. Equipment procurement, maintenance, network expansion and human resources are used as cost variables. Parametric modeling is performed and correlation mapping is established to form a scalable cost framework, providing a quantitative basis for the generation of benefit indicators. Benefit indicator generation module: Generates benefit indicators using the cost model, and quantifies each indicator based on the relationship between the above variables to form a multi-dimensional benefit indicator set; Evaluation framework integration module: Integrates cost models and benefit indicators into a closed-loop evaluation framework, links various evaluation elements through feedback loops, and dynamically adjusts framework parameters according to scenario parameters to ensure that the evaluation framework adapts to different deployment needs; Decision Application Module: The output of the evaluation framework is used as the input for delivery decisions. By iteratively adjusting and refining route selection and resource allocation, a business closed loop from data collection to decision optimization is formed.

Images