A UAV maintenance task scheduling and resource management platform

Abstract

This invention belongs to the field of UAV technology and intelligent operation and maintenance management technology. It discloses a UAV maintenance task scheduling and resource management platform and method, including: a data acquisition and fusion module that acquires and fuses multi-source data; a fault diagnosis and predictive analysis module that uses deep learning models and a CNN-LSTM hybrid model to diagnose faults, predict lifespan, and generate a maintenance task list; an optional task dynamic priority evaluation module that prioritizes the maintenance task list; an intelligent scheduling decision module that uses an improved NSGA-II and VNS hybrid algorithm to generate an optimized allocation scheme based on priority and resource constraints; a resource dynamic management and collaboration module that enables dynamic allocation and collaboration of personnel, spare parts, tools, and sites, and can integrate AR guidance; and an optional visualization monitoring and feedback optimization module that displays status and feedback data to achieve closed-loop optimization. This invention can significantly improve UAV maintenance efficiency, reduce operating costs, and ensure flight safety.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) technology and intelligent operation and maintenance management technology, and specifically relates to a UAV maintenance task scheduling and resource management platform and method. Background Technology

[0002] The rapid development of drone technology has led to its increasingly widespread application in fields such as inspection, logistics, surveying, and security. This has resulted in an expansion of drone fleets and a greater complexity of operational tasks, placing higher demands on drone reliability, availability, and maintenance efficiency. Traditional drone maintenance methods, which rely heavily on human experience for fault diagnosis, task assignment, and resource coordination, have significant shortcomings.

[0003] First, there is a lack of fault diagnosis and prediction capabilities. Current maintenance is mostly a reactive response after a fault occurs, lacking real-time monitoring of the health status of UAVs and early fault warning mechanisms. It is difficult to plan maintenance tasks in advance, resulting in frequent sudden faults that affect mission execution and flight safety.

[0004] Secondly, task scheduling and resource allocation are inefficient. The priority of maintenance tasks often lacks an objective and scientific evaluation system; the allocation of maintenance personnel, spare parts, tools and sites relies heavily on manual coordination, making it difficult to achieve overall optimization, resulting in resource waste or waiting delays, and long average downtime for drones.

[0005] Third, the level of refinement in resource management is insufficient. Spare parts inventory management fails to fully incorporate predictive demand, easily leading to shortages or excessive stockpiling of critical spare parts; the skill level of maintenance personnel is not accurately matched with task requirements, affecting maintenance quality and efficiency; and the utilization rate of maintenance tools and maintenance sites needs to be improved.

[0006] Fourth, information silos are prevalent. Drone status data, historical maintenance records, and available resource information are stored in a scattered manner, making data sharing and comprehensive analysis difficult, and hindering the formation of effective knowledge accumulation and closed-loop optimization.

[0007] In the prior art, for example, Chinese patent application number CN119293682B discloses a high-precision UAV status judgment method and system based on fault search tree. This method mainly focuses on using sensor data for fault identification, but does not delve into intelligent scheduling of maintenance tasks and comprehensive resource optimization management after identification. Another Chinese patent application number CN117369260B discloses a UAV scheduling method combining energy consumption models and task time windows. This method focuses on scheduling in specific scenarios, but lacks a comprehensive and adaptive scheduling and management solution tailored to the characteristics of the entire UAV maintenance process, particularly combining fault prediction, dynamic priority assessment, and complex resource constraints.

[0008] Therefore, there is an urgent need to develop a comprehensive platform that can integrate data from the entire lifecycle of UAVs, enable intelligent fault diagnosis and prediction of remaining service life, and on this basis, conduct efficient maintenance task scheduling and refined resource management to address the challenges currently faced in UAV operation and maintenance support. Summary of the Invention

[0009] The main objective of this invention is to overcome the problems of insufficient accuracy in UAV maintenance diagnosis, weak predictive capabilities, inadequate task scheduling optimization, insufficiently refined resource management, and poor information collaboration in existing technologies, and to provide a UAV maintenance task scheduling and resource management platform. This invention aims to significantly improve the intelligence, automation, and refinement of UAV maintenance, enhance maintenance efficiency, shorten UAV downtime, optimize maintenance resource allocation, and ultimately ensure UAV flight safety.

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] A UAV maintenance task scheduling and resource management platform, comprising:

[0012] The data acquisition and fusion module is configured to acquire UAV operation data, maintenance resource data, and external environment data from UAV systems, maintenance resource management systems, and external environment information sources, and to standardize and fuse the multi-source heterogeneous data to form a standardized dataset.

[0013] The fault diagnosis and prediction analysis module is configured to diagnose the current fault of the UAV based on the standardized dataset using a preset deep learning-based fault diagnosis model, and to predict the remaining service life of key components using a deep learning-based life prediction model, thereby generating a maintenance task list.

[0014] The deep learning-based lifespan prediction model is specifically a hybrid prediction model that combines convolutional neural networks (CNN) and long short-term memory networks (LSTM).

[0015] The intelligent scheduling decision module is configured to generate a maintenance task allocation scheme that considers multiple constraints and optimization objectives based on the maintenance task list, task dynamic priority, and real-time maintenance resource availability. This scheme employs a hybrid optimization scheduling algorithm that combines an improved non-dominated sorting genetic algorithm (NSGA-II) with an elite retention strategy and a variable neighborhood search (VNS).

[0016] The resource dynamic management and collaboration module is configured to dynamically allocate and coordinate maintenance personnel, spare parts, tools and sites according to the maintenance task allocation scheme.

[0017] Preferably, the platform further includes a task dynamic priority evaluation module, located between the fault diagnosis and predictive analysis module and the intelligent scheduling decision module, configured for:

[0018] Construct a multi-dimensional task evaluation index system, the evaluation indexes including fault severity level, task impact degree, UAV criticality coefficient, spare parts acquisition difficulty coefficient, maintenance window period, and the urgency of predicting the occurrence of faults;

[0019] The method combines the Analytic Hierarchy Process (AHP) with fuzzy comprehensive evaluation to dynamically prioritize each maintenance task in the maintenance task list and sort the maintenance task list according to the scores.

[0020] Preferably, the platform further includes a visualization monitoring and feedback optimization module, configured for:

[0021] The system uses a graphical user interface to display the drone's health status, maintenance task distribution and progress, and maintenance resource scheduling in real time; it also records maintenance process data to form a maintenance knowledge base.

[0022] Furthermore, machine learning algorithms are used to analyze historical maintenance data and scheduling execution effects, and the weights in the fault diagnosis model, the life prediction model, the priority evaluation method, and the strategy of the hybrid optimization scheduling algorithm are adaptively adjusted and optimized to form a closed-loop feedback optimization mechanism.

[0023] Compared with the prior art, the present invention has the following advantages:

[0024] First, this invention, by constructing a data acquisition and fusion module and a fault diagnosis and predictive analysis module, and particularly by employing a life prediction model combining convolutional neural networks (CNN) and long short-term memory networks (LSTM), enables comprehensive and in-depth analysis of UAV operational data and component health status. This not only effectively improves the accuracy and coverage of fault diagnosis, but more importantly, it achieves accurate prediction of potential fault trends and the remaining service life of key components. This makes it possible to shift from reactive maintenance to predictive and proactive maintenance, thereby significantly reducing unplanned groundings and mission interruptions caused by sudden failures, and significantly enhancing the reliability and safety of UAV operations.

[0025] Secondly, this invention overcomes the limitations of traditional manual scheduling and simple rule-based scheduling by establishing a dynamic task priority evaluation module and employing an intelligent scheduling decision module that combines an improved non-dominated sorting genetic algorithm (NSGA-II) with an elite retention strategy and a variable neighborhood search (VNS) hybrid optimization scheduling algorithm. The platform can scientifically calculate the dynamic priority of tasks based on multi-dimensional evaluation indicators. Under the premise of meeting multiple complex constraints such as maintenance personnel skill requirements, spare parts inventory, and tool availability, it quickly generates globally optimized or near-optimal maintenance task allocation schemes with optimization objectives such as minimizing total UAV downtime, minimizing total maintenance costs, and maximizing resource utilization balance. This intelligent scheduling decision ensures the most reasonable allocation of maintenance resources, significantly shortens the average UAV maintenance cycle, and improves the overall fleet uptime and task execution efficiency.

[0026] Furthermore, this invention, through a dynamic resource management and collaboration module, achieves refined and dynamic management and efficient collaboration of core maintenance resources such as maintenance personnel, spare parts, tools, and work sites. The intelligent scheduling and early warning replenishment mechanism for spare parts, the full lifecycle tracking management of tools and equipment, and the integrated augmented reality (AR) remote maintenance guidance function not only improve the utilization efficiency and turnover rate of various resources and reduce delays and waste caused by resource mismatch or shortages, but also enhance the first-time success rate of complex maintenance tasks through technological empowerment, thereby effectively controlling overall maintenance operating costs.

[0027] Furthermore, this invention constructs a transparent and efficient information interaction and decision support environment by setting up a visual monitoring and feedback optimization module. This module can display task progress, resource status, and key performance indicators in real time, enabling managers to fully grasp the operational dynamics and promptly identify and handle anomalies. More importantly, through continuous collection and intelligent analysis of maintenance process data and scheduling execution effects, this module can feed back and optimize the parameters of the diagnostic prediction model, the weights of the priority evaluation system, and the strategy parameters of the scheduling algorithm, forming a continuously learning and self-improving closed-loop optimization system. This ensures that the platform can adapt to constantly changing drone models, operating scenarios, and operational needs, maintaining its long-term advanced nature and applicability. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein:

[0029] Figure 1This is a schematic diagram of the system architecture of a UAV maintenance task scheduling and resource management platform according to the present invention;

[0030] Figure 2 This is a schematic diagram of the initialization and data preparation process of the hybrid optimization scheduling algorithm in the intelligent scheduling decision module of the present invention;

[0031] Figure 3 This is a schematic diagram of the NSGA-II core algorithm flow of the hybrid optimization scheduling algorithm in the intelligent scheduling decision module of the present invention;

[0032] Figure 4 This is a schematic diagram of the VNS local search enhancement process of the hybrid optimization scheduling algorithm in the intelligent scheduling decision module of the present invention. Detailed Implementation

[0033] The following detailed description, in conjunction with specific embodiments, illustrates the specific steps and operation flow of a two-stage potential diffusion system for high-fidelity virtual try-on according to the present invention, so that those skilled in the art can implement the present invention. Through the following specific embodiments, those skilled in the art can fully understand and implement the technical solution of the present invention.

[0034] Example 1

[0035] Please see Figure 1 The UAV maintenance task scheduling and resource management platform disclosed in this embodiment has a core architecture that consists of a tightly integrated system comprised of key modules such as a data acquisition and fusion module 100, a fault diagnosis and predictive analysis module 200, a task dynamic priority evaluation module 300, an intelligent scheduling decision-making module 400, a resource dynamic management and collaboration module 500, and a visualization monitoring and feedback optimization module 600. These modules interact and collaborate via an internal bus or network, forming the foundation for the platform's intelligent operation and maintenance management.

[0036] The core task of the data acquisition and fusion module 100 is to actively collect raw information from multiple data sources, and to perform in-depth processing and integration to provide a data foundation for subsequent analysis and decision-making.

[0037] During the data collection process, the platform communicates with the UAV's flight control system, various onboard sensors, PHM (Health Management Unit), and mission payload via a dedicated interface. This connection ensures that the platform can acquire detailed flight parameters of the UAV in real-time or near real-time, instantaneous operational status data of key components, battery charge / discharge characteristics, drive motor operating parameters, image transmission link quality data, and any fault alarms detected by system self-tests or sensors. Historical flight logs and mission execution records are also fully imported into the platform's central database through corresponding standardized interfaces for subsequent trend analysis and fault tracing.

[0038] To acquire maintenance resource data, the platform connects to the existing maintenance personnel management system to accurately obtain maintenance technicians' identification numbers, professional skill levels, qualification certification information, current workload, real-time geographical location, and shift schedules. Through deep integration with the spare parts inventory management system, the platform comprehensively grasps the unique identifier, name, specifications, current inventory quantity, safety stock threshold, quantity in transit, storage location information, supplier information, and procurement cycle of each type of maintenance spare part.

[0039] The platform also integrates with the tool and equipment management system to obtain identification numbers, types, calibration status, borrowing records, and maintenance plans for special maintenance tools, general maintenance tools, and testing equipment. Regarding maintenance site resources, the platform connects to the maintenance site management system to clearly define the immediate availability status, capacity limits, specific environmental conditions, and configuration of special maintenance equipment for each maintenance workstation or site. Immediate availability status specifically includes whether the site is vacant, occupied, or reserved; capacity limits specifically include the size and number of drones that can be accommodated; specific environmental conditions include temperature, humidity, and cleanliness; and special maintenance equipment specifically includes hoisting equipment and test benches.

[0040] In addition, the platform proactively acquires necessary data from external environmental information sources. Utilizing standard application programming interfaces (APIs), it obtains real-time weather forecasts from professional meteorological service platforms, relevant airspace usage restrictions or temporary flight control notices from air traffic management departments, and future drone mission plans from mission planning systems, providing crucial input for forward-looking planning and scheduling.

[0041] During the data fusion and processing phase, the platform standardizes the collected multi-source heterogeneous data. This process includes data format conversion, unit standardization, timestamp alignment, outlier cleaning, and missing value imputation. Outlier cleaning can be handled using methods such as the 3σ principle or the Isolation Forest algorithm, while missing value imputation can be achieved using methods such as mean imputation, regression imputation, or machine learning-based predictive imputation. After preprocessing and fusion, the resulting standardized dataset is stored in the platform's central database or distributed data lake for use by other modules.

[0042] The fault diagnosis and prediction analysis module 200 uses the standardized dataset provided by module 100 to perform in-depth analysis using analysis techniques and algorithm models to accurately identify faults and predict risks.

[0043] At the fault diagnosis level, the platform comprehensively utilizes multiple strategies:

[0044] First, there's the rule-based diagnostic method, which utilizes a built-in expert knowledge base containing FMECA data on common UAV fault modes, causes, and effects, along with preset diagnostic rules. Second, there's the model-based diagnostic method, which establishes digital twin models or mathematical-physical models for key components and detects and isolates faults by comparing the residuals between the model output and actual sensor readings. Third, there's the data-driven diagnostic method, which uses historical fault data and corresponding sensor data to train machine learning classification models, such as Support Vector Machines (SVM), Random Forests, Gradient Boosting Decision Trees (GBDT), or Deep Neural Networks (DNN), to identify current sensor data patterns and determine the presence and possible types of faults.

[0045] In terms of failure prediction, the module focuses on identifying potential degradation trends and estimating the remaining effective life (RUL) of components:

[0046] First, a prediction method based on state parameter trends is employed to perform time-series analysis on the health indicator parameters of key components, predicting their future evolution trends. Second, for components exhibiting significant wear or fatigue characteristics, a hybrid prediction model combining a Convolutional Neural Network (CNN) and a Long Short-Term Memory (LSTM) network is used for RUL estimation. This CNN-LSTM model takes historical operational data and current state data sequences of the component as input; the CNN layer extracts local and spatial features from the data sequences; the LSTM layer captures the temporal dependencies and long-term trends in the data sequences; and the output layer predicts the RUL value of the component or the probability of failure at a specific future time point.

[0047] Specifically, the network structure of the CNN-LSTM hybrid prediction model includes:

[0048] Input layer: Receives time series data of length T. Each time step It contains n-dimensional feature vectors, representing the key components of the UAV in the nth dimension. The state parameters at each point in time.

[0049] CNN Feature Extraction Layer: Employs one-dimensional convolutional operations to extract local feature patterns, specifically including: First convolutional layer: kernel size 3, output channels 64, using ReLU activation function; Second convolutional layer: kernel size 5, output channels 128, using ReLU activation function; Max pooling layer: pooling window size 2, stride 1; Dropout layer: dropout rate set to 0.2 to prevent overfitting.

[0050] LSTM Temporal Modeling Layer: A bidirectional LSTM network is used to capture long-term temporal dependencies: Number of LSTM units: 256; Number of layers: 2 stacked LSTM layers; The outputs of the forward and backward LSTMs are concatenated to form a 512-dimensional feature vector;

[0051] Fully connected output layer: First fully connected layer: 512→128, using ReLU activation; Second fully connected layer: 128→64, using ReLU activation; Output layer: 64→1, outputting RUL prediction values.

[0052] Loss function: The mean squared error loss function (MSE) is used, combined with an L2 regularization term. ,in This is the regularization coefficient, set to 0.001.

[0053] Training strategy: The Adam optimizer is used, with an initial learning rate of 0.001. The learning rate decreases to 0.9 times every 10 epochs. The batch size is 32, and the maximum number of training epochs is 200.

[0054] Taking lithium batteries as an example, their RUL model It can be represented as:

[0055] ,

[0056] in: The mapping function obtained through training, This refers to the number of charge-discharge cycles. The average depth of discharge. Operating temperature This is the discharge current. Initial healthy state, This represents historical sequence data. Model training optimizes network weights using the backpropagation algorithm to minimize the error between the predicted RUL and the actual RUL.

[0057] After completing the diagnostic prediction, the module comprehensively analyzes the results and automatically generates a detailed maintenance task list. Each task includes the UAV ID, task type, fault description or prediction details, information on involved components, suggested maintenance plan, required skill level, estimated man-hours, a list of required spare parts, and a list of special tools. This list serves as the core input for subsequent evaluation and decision-making.

[0058] The task dynamic priority evaluation module 300 prioritizes the maintenance task list generated by module 200. The module first establishes a multi-dimensional evaluation index system. Primary indicators cover the degree of safety impact, task impact, economic impact, and timeliness impact, using the Analytic Hierarchy Process (AHP) to determine the basic weights of each primary indicator. Secondary indicators include fault severity level, UAV criticality coefficient, spare parts acquisition difficulty coefficient, maintenance window length, and predicted fault occurrence proximity, among others. Fuzzy comprehensive evaluation methods are used to handle their uncertainty and fuzziness. Each maintenance task... Dynamic priority Calculate using the following example formula:

[0059]

[0060] in, For the final dynamic priority score, For the first Weights of each primary evaluation indicator For the first The task in the first The comprehensive evaluation value under each primary indicator The expected completion deadline for the task. For the current time, The acceptable repair window duration for the task. The time urgency factor is the impact coefficient. Estimate the overall resource consumption value for the task. This serves as a reference baseline for the maximum possible resource consumption of a single task in the system. This represents the inverse impact coefficient of resource consumption.

[0061] The specific calculation steps for combining the Analytic Hierarchy Process (AHP) with fuzzy comprehensive evaluation are as follows:

[0062] Step 1: Construct the judgment matrix

[0063] for One evaluation indicator, construct Judgment matrix A, where Indicators relative to indicators Importance: ,in =1 / , =1.

[0064] Step 2: Calculate the weight vector

[0065] The weights are calculated using the eigenvector method. ,in For the weight vector, It is the largest eigenvalue.

[0066] Step 3: Consistency Check

[0067] Calculate the consistency ratio ,in This is a random consistency indicator. When... When the value is less than 0.1, the judgment matrix is ​​considered to have satisfactory consistency.

[0068] Step 4: Fuzzy Comprehensive Evaluation

[0069] For each secondary indicator, a fuzzy evaluation set V = {Excellent, Good, Average, Poor} is established, corresponding to scores {0.9, 0.7, 0.5, 0.3}. Trapezoidal fuzzy numbers are used to represent the uncertainty of the evaluation values.

[0070] Step 5: Final Score Calculation

[0071] The final dynamic priority score is calculated by combining the AHP weights and fuzzy evaluation results.

[0072] This module dynamically updates task priorities based on real-time conditions.

[0073] Please see Figure 2 , Figure 3 and Figure 4 After obtaining the sorted maintenance task list and real-time maintenance resource availability, the intelligent scheduling decision module 400 initiates its core scheduling process. It employs a hybrid optimization scheduling algorithm that combines an improved non-dominated sorting genetic algorithm (NSGA-II) with an elite retention strategy and a variable neighborhood search (VNS) to generate a maintenance task allocation scheme.

[0074] Please see Figure 2 During the algorithm initialization and input phase, the module obtains the sorted list of maintenance tasks and the real-time status of maintenance resources, and sets optimization objectives, including minimizing the total downtime of UAVs, minimizing the total maintenance cost, maximizing the balance of resource utilization, and maximizing the completion rate of high-priority tasks.

[0075] The scheduling schemes are encoded as chromosomes, and an initial population is generated randomly. For each scheduling scheme in the population, the fitness is calculated based on the optimization objective and constraints.

[0076] Please see Figure 3 The core process of genetic evolution operations, namely NSGA-II, includes: performing non-dominated sorting and crowding calculation, dividing the Pareto front and calculating the crowding distance of individuals; performing elite selection to retain superior individuals for the next generation; and performing adaptive crossover and adaptive mutation operations, with the operators and probabilities dynamically adjusted according to population diversity and convergence status.

[0077] The adaptive crossover operator dynamically adjusts the crossover probability and method based on the population convergence state and individual fitness differences:

[0078] Adaptive formula for crossover probability:

[0079] in, =0.9 is the maximum crossover probability. =0.6 is the minimum crossover probability. For the current algebra, For the largest algebra, =2 is the convergence control parameter.

[0080] Crossover operation specific steps: Parent selection: Tournament selection is used, with a tournament size of 3; Crossover point determination: For maintenance task scheduling chromosomes, 1-3 crossover points are randomly selected; Segment exchange: The task allocation segments of parent individuals between crossover points are exchanged; Constraint repair: Check whether the crossover offspring violate constraints such as skill matching and resource capacity. If they do, a constraint repair strategy is adopted.

[0081] Skill mismatch: Reassign to personnel with the corresponding skills; Resource conflict: Adjust the task start time or replace resources; Temporal dependency violation: Reorder according to dependency relationships.

[0082] Local search enhancement, also known as the VNS process (see...) Figure 4 After NSGA-II has evolved to a certain stage, it starts by selecting representative solutions from the Pareto optimal solution set as initial solutions and performing local deep optimization by searching in multiple predefined neighborhood structures (such as task order swapping, resource reallocation, and start time fine-tuning).

[0083] The VNS algorithm defines five neighborhood structures for local search optimization:

[0084] Neighborhood structure -Task order swapping: Randomly select two adjacent tasks and swap their execution order; suitable for optimizing the timing arrangement between tasks.

[0085] Neighborhood structure - Task redistribution: Randomly select one task and redistribute it to other maintenance personnel who meet the skill requirements; this is used to balance the workload of personnel.

[0086] Neighborhood structure -Time window adjustment: Randomly adjusts the start time of the selected task within a range of ±2 hours; used to optimize resource utilization.

[0087] Neighborhood structure -Resource replacement: Replace the spare parts, tools or facilities required for the task with equivalent resources; suitable for mitigating resource conflicts.

[0088] Neighborhood structure -Task decomposition and merging: Decompose complex tasks into multiple sub-tasks, or merge similar sub-tasks; used to improve maintenance efficiency.

[0089] VNS search process: Step 1, initial solution s = selected from the Pareto optimal solution set; Step 2, k = 1 (neighborhood structure index); Step 3, While k ≤ 5: randomly generate a solution s' in the neighborhood Nk(s); improve s' using local search to obtain s''; If f(s'') < f(s): s = s'', k = 1; Else: k = k + 1; Step 4, return the optimal solution s.

[0090] Throughout the process, various constraint conditions are strictly checked, such as the matching of maintenance personnel skills and task requirements, the consistency of spare part model specifications, the specialization of special tools, the capacity and environmental adaptability of the maintenance site, the continuous operation duration limit, and the temporal dependencies between tasks. The algorithm terminates when it reaches the preset maximum number of iterations or there is no significant improvement in the Pareto front, and outputs the final Pareto optimal solution set or recommends a comprehensive optimal solution, specifying the planned time, executor, required resources, and site for each task.

[0091] Resource dynamic management and collaboration module 500: responsible for executing the scheduling plan generated by the execution module 400 and managing resources. Its functions include: automatically assigning maintenance tasks to maintenance personnel with corresponding skills and currently available, and planning the optimal path; after task confirmation, the system automatically reserves or applies for the required spare parts, triggers urgent procurement or transfer if the inventory is insufficient, and tracks the spare parts; maintenance personnel can reserve or apply for the use of tool equipment through the platform, and the system displays their status and location in real time; the use of the maintenance site is incorporated into the scheduling, and the system allocates according to task characteristics and avoids conflicts; provides remote maintenance guidance based on augmented reality AR, allowing experts to remotely guide on-site maintenance in real time, and all guidance processes can be recorded.

[0092] Visualization monitoring and feedback optimization module 600: provides a unified monitoring and management interface and drives system optimization. Its functions include: real-time displaying key performance indicators such as the fleet health status, maintenance task queue and progress, resource load, spare part inventory, and personnel efficiency through a customized dashboard; users can query the full life cycle status of any maintenance task, and the system automatically warns of tasks with lagging progress or anomalies; collects and organizes historical maintenance cases to form a structured maintenance knowledge base to assist in diagnosis and training; realizes closed-loop feedback and continuous optimization, specifically by comparing the actual fault information with the model results to update the parameters of the fault diagnosis and prediction model, analyzing the completion of high-priority tasks to adjust the weights of the priority evaluation model, and comparing the differences between the scheduling plan and the actual execution to optimize the scheduling algorithm parameters or strategies.

[0093] Embodiment 2

[0094] This embodiment provides a method for unmanned aerial vehicle maintenance task scheduling and resource management implemented based on the platform described in Embodiment 1. Its main steps include:

[0095] Step S10: Data Acquisition and Initialization The fusion platform continuously collects UAV flight data, sensor data, component status data, real-time maintenance resource information and historical data through the data acquisition and fusion module 100, and stores them in the database after fusion processing.

[0096] Step S20: Fault Monitoring, Prediction, and Task Generation. The fault diagnosis and predictive analysis module 200 monitors the UAV's health status in real time. When a real-time fault alarm is detected, or a current fault is identified through model analysis, a fault repair task is immediately generated. Simultaneously, this module periodically scans the fleet based on the CNN-LSTMRUL prediction model, generating predictive maintenance tasks for components with RUL below the warning threshold. Combined with a preset periodic maintenance plan, planned maintenance tasks are generated. All tasks are added to the initial maintenance task list.

[0097] Step S30: Dynamic evaluation and sorting of task priorities. The dynamic priority evaluation module 300 calculates the dynamic priority score of each task in the maintenance task list generated in S20 according to the preset evaluation model, and sorts the list according to the score.

[0098] Step S40: Intelligent Scheduling and Resource Allocation. The intelligent scheduling decision module 400 obtains the sorted maintenance task list and the latest resource availability status provided by the resource dynamic management and coordination module 500. It then initiates the NSGA-II and VNS hybrid optimization scheduling algorithm, iteratively searching for the optimal maintenance task allocation scheme under multiple constraints and with a preset optimization objective.

[0099] Step S50: Task Execution and Resource Collaboration. The resource dynamic management and collaboration module 500 automatically pushes tasks to relevant maintenance personnel based on the selected maintenance task allocation plan, and coordinates spare parts issuance, tool preparation, and site reservation. Maintenance personnel execute operations according to instructions, record working hours and consumption, fill in reports, and can initiate AR remote collaboration as needed.

[0100] Step S60: Monitoring, Feedback, and Optimization. The visualization monitoring and feedback optimization module 600 monitors the task execution progress and resource usage throughout the process. After the task is completed, complete maintenance records are collected, and the knowledge base is updated. The system periodically analyzes historical scheduling execution effects and actual operation and maintenance data, and adaptively adjusts the fault diagnosis model, priority evaluation weights, and scheduling algorithm parameters to achieve closed-loop optimization.

[0101] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result according to substantially the same method falls within the scope of the present invention. Therefore, the scope of the present invention is defined only by the appended claims.

Claims

1. A UAV maintenance task scheduling and resource management platform, characterized in that, include: The data acquisition and fusion module is configured to acquire and fuse UAV operation data, maintenance resource data, and external environment data to form a standardized dataset. The fault diagnosis and prediction analysis module is configured to diagnose the current fault of the UAV based on the standardized dataset using a first preset deep learning-based fault diagnosis model, and to predict the remaining service life of key components using a hybrid prediction model combining convolutional neural networks and long short-term memory networks. The hybrid prediction model can output the predicted value of the remaining service life with physical time units, thereby generating a maintenance task list. The task dynamic priority evaluation module is configured to prioritize maintenance tasks based on the urgency of failures reflected in the predicted remaining useful life values ​​with physical time units; specifically, it includes: Construct a multi-dimensional task evaluation index system, the evaluation indexes including fault severity level, impact on subsequent tasks, UAV criticality coefficient, spare parts acquisition difficulty coefficient, maintenance window period, and the urgency of predicting the time of fault occurrence; A combination of the analytic hierarchy process (AHP) and fuzzy comprehensive evaluation is used to dynamically prioritize each maintenance task; specifically, the following formula is used to calculate the priority of each task. Dynamic priority of each maintenance task: in, For the first The final dynamic priority score for each maintenance task. For the first The weight of each primary evaluation indicator, For the first The task in the first The comprehensive evaluation value under each primary indicator For the first The expected completion deadline for each task. For the current time, For the first Acceptable repair window duration for each task. The time urgency factor is the impact coefficient. For the first The estimated total resource consumption for each task. This serves as a reference baseline for the maximum possible resource consumption of a single task in the system. The estimated total resource consumption value for each task; The maintenance task list is sorted according to the calculated priority scores; The intelligent scheduling decision module is configured to generate a maintenance task allocation scheme based on the maintenance task list, dynamic task priority, and real-time maintenance resource availability, using a hybrid optimization scheduling algorithm that integrates an improved non-dominated sorting genetic algorithm with an elite retention strategy and a variable neighborhood search. The resource dynamic management and collaboration module is configured to dynamically allocate and coordinate maintenance personnel, spare parts, tools and sites according to the maintenance task allocation scheme.

2. The UAV maintenance task scheduling and resource management platform according to claim 1, characterized in that, The UAV operation data acquired by the data acquisition and fusion module includes: UAV flight parameter data, sensor telemetry data, component health status information, historical maintenance records, and fault code data; Furthermore, the maintenance resource data acquired by the data acquisition and fusion module includes: skill level information, qualification certification information, real-time location information, busy / idle status information of maintenance personnel, model parameters, inventory quantity, storage location information, and estimated replenishment time information of spare parts, type information, availability information, and calibration cycle information of maintenance tools, as well as capacity parameters, occupancy information, and environmental condition information of the maintenance site.

3. The UAV maintenance task scheduling and resource management platform according to claim 1, characterized in that, The fault diagnosis and prediction analysis module also incorporates a fault propagation path analysis method based on dynamic Bayesian networks to identify potential fault modes and their probability of occurrence.

4. The UAV maintenance task scheduling and resource management platform according to claim 1, characterized in that, The intelligent scheduling decision module employs a hybrid optimization scheduling algorithm with multiple optimization objectives: minimizing the total downtime of UAVs, minimizing the total maintenance cost, maximizing the balance of resource utilization, and maximizing the completion rate of high-priority tasks. In the evolutionary process of the non-dominated sorting genetic algorithm, an adaptive crossover operator and an adaptive mutation operator based on the characteristics of the maintenance task and the resource status are introduced. After obtaining the Pareto optimal solution set, the variable neighborhood search is used to perform local deep optimization of the key scheduling schemes in the solution set.

5. The UAV maintenance task scheduling and resource management platform according to claim 1, characterized in that, When generating a maintenance task allocation scheme, the intelligent scheduling decision module also considers the following constraints: matching of maintenance personnel skills with task requirements, consistency of spare parts models and specifications, specialization of special tools, maintenance site capacity and environmental adaptability, continuous operation time limit, and time dependence constraints between different maintenance tasks.

6. The UAV maintenance task scheduling and resource management platform according to claim 1, characterized in that, The resource dynamic management and collaboration module performs the following functions: The system automatically assigns identified maintenance tasks to available maintenance personnel with the appropriate skills and plans the optimal route to the maintenance location or the location of the drone. Automatically reserve or request the required spare parts based on task requirements, and trigger intelligent replenishment or emergency procurement processes based on spare parts inventory and transit damage risks. Track the real-time location and usage status of repair tools to ensure tool availability and schedule timely maintenance; Schedule and allocate maintenance site resources, optimize site turnover rate, and coordinate site sharing when multiple tasks are carried out in parallel. It also provides augmented reality-based remote maintenance guidance, allowing senior technical experts to remotely guide on-site maintenance personnel in performing complex maintenance tasks in real time.

7. The UAV maintenance task scheduling and resource management platform according to claim 1, characterized in that, It also includes a visualization monitoring and feedback optimization module, which is configured to: The geographic information system interface displays the drone's location, maintenance task distribution, maintenance personnel dynamics, and resource scheduling in real time. It provides status tracking and early warning functions for the entire lifecycle of maintenance tasks; it records actual working hours, spare parts consumption data, and fault resolution data during the maintenance process to form a maintenance knowledge base. By using machine learning algorithms to analyze historical maintenance data and scheduling performance, the parameters of the fault diagnosis model, the parameters of the hybrid prediction model, the evaluation weights of the task dynamic priority evaluation module, and the strategy of the hybrid optimization scheduling algorithm are adaptively adjusted and optimized.

8. A method for scheduling and managing resources for unmanned aerial vehicle (UAV) maintenance tasks based on the platform described in any one of claims 1-7, characterized in that, Includes the following steps: Step S1: Acquire and fuse UAV operation data, maintenance resource data, and external environment data through the data acquisition and fusion module to obtain a standardized dataset; Step S2: The fault diagnosis and prediction analysis module processes the standardized dataset. The first preset deep learning-based fault diagnosis model is used to diagnose the current fault of the UAV, and a hybrid prediction model combining convolutional neural network and long short-term memory network is used to predict the remaining service life of key components, so as to output the predicted value of the remaining service life with physical time unit and generate a maintenance task list. Step S3: Using the task dynamic priority evaluation module, prioritize the maintenance task list based on the urgency of the fault as reflected by the remaining service life prediction value with physical time units generated in step S2. Step S4: Based on the sorted maintenance task list and real-time maintenance resource availability, the intelligent scheduling decision module generates a maintenance task allocation scheme by using a hybrid optimization scheduling algorithm that combines an improved non-dominated sorting genetic algorithm with an elite retention strategy and a variable neighborhood search. Step S5: Dynamically allocate maintenance resources and coordinate operations according to the maintenance task allocation scheme through the resource dynamic management and collaboration module; Step S6: If the platform includes a visualization monitoring and feedback optimization module, then the module is used to monitor task execution and provide feedback optimization to the system.

Images