47 results about "Combination algorithm" patented technology

The invention discloses a distribution box room environment parameter integrated measurement method, and particularly relates to the field of multi-parameter measurement, and the method comprises the steps: firstly constructing a machine room three-dimensional model and a sensor correlation degree matrix, then building a heat balance, humidity diffusion and airflow motion model, and calibrating parameters; deploying a measurement system and calibrating through dual synchronization and cross check; establishing a sensor confidence evaluation system and an adaptive threshold based on 72-hour reference data; fusing calibration data by adopting a three-level correlation calibration mechanism; four types of measurement modes and conversion logics are designed, and measurement resources are dynamically allocated through environmental risk assessment; multi-dimensional state parameters are extracted, a comprehensive evaluation value is calculated through normalization, dynamic weighting and combinatorial algorithms, and five-level early warning response is achieved; according to the method, multiple models and an intelligent algorithm are integrated, the parameter measurement precision and the environment risk identification efficiency are improved, the response delay is reduced, the fault diagnosis accuracy is improved, and reliable technical support is provided for safe operation and maintenance of the distribution box room.

The invention discloses a zero-trust gateway single packet authentication method and system for a new energy fan control system, computing equipment, a computer storage medium, a computer program product and a chip. The authentication method comprises the steps of generating an authentication material; exchanging and deriving a mixed key; binding Hash calculation is carried out; encrypting and assembling the message; receiving and verifying by the gateway; and binding verification and registration are carried out. The invention relates to a zero-trust gateway single packet authentication method combining a post quantum cryptography algorithm and a national commercial cryptography algorithm. The method is applied to secure communication between a wind power plant and a centralized control center, an SM2, SM3 and SM4 combined algorithm in a national secret system is combined with a post-quantum key agreement algorithm CRYSTALS-Kyber512, identity authentication, encryption and integrity verification are completed through a single message, confidentiality protection and anti-quantum security guarantee under one-time interaction are achieved, and the security of the wind power plant is improved. The method is especially suitable for new energy scenes with wide fan distribution, complex links and high real-time requirements.

In some implementations, a device may receive a request for cryptographic transmission of payload data. The device may obtain the payload data for encryption. The device may encrypt the payload data, to generate encrypted payload data, using a first encryption key, wherein the first encryption key is configured such that the encrypted payload data is decryptable using a second encryption key that corresponds to the first encryption key. The device may generate tokenized data. The device may combine, using a combining algorithm of a set of combining algorithms, the encrypted payload data and the tokenized data to generate combined data, wherein the combined data is not directly decryptable using the second encryption key. The device may provide the combined data and an indicator of the combining algorithm as a response to the request.

The invention discloses an intelligent transient control method and system for a hybrid power assembly, and relates to the technical field of hybrid power engineering machinery, and the method comprises the steps: obtaining a historical working state data set of the hybrid power engineering machinery; training an improved unscented Kalman filtering algorithm by adopting a historical working state data set; the improved unscented Kalman filtering algorithm is a two-stage combined algorithm adopting a whale algorithm and a moss growth optimization algorithm, and a dispersion degree factor and a prior distribution factor with the minimum fitness function value in the unscented Kalman filtering are generated through iterative search; and on the basis of the required power and the working state of the whole machine at the current moment, a trained improved unscented Kalman filtering algorithm is adopted to predict the change of the transient output power of the power battery and the motor and the change of the transient output power of the engine so as to control the output power of the power battery, the motor and the engine. The dynamic response of the whole machine is improved, and adverse effects on an engine, a power battery and a motor are reduced.

The application discloses an automatic planning method and device for a walking route of a land controllable seismic source, and belongs to the field of physical exploration. The application introduces a TSP idea, and finally obtains a seismic source vehicle driving route which is short in path, less in barrier crossing, less in driving restriction, and in line with the driving operation habit of a driver through steps of constructing a map, setting relevant parameters and weight ratios, and solving the walking route of the land controllable seismic source by using a TNH combination algorithm. The application combines the advantages of a two-approximation algorithm, a Heuristic algorithm and an NSGA_II algorithm to propose the TNH combination algorithm for solving the walking route of the land controllable seismic source, so that the automatic planning of the walking route of the controllable seismic source is realized, the operation efficiency is accelerated, the accuracy of operation results is improved, the calculated route is more in line with the demand of field construction, and the exploration operation is further developed intelligently. The application is applied to the field of physical exploration to plan the walking route of the controllable seismic source.

The invention belongs to the technical field of wind turbine generators, and particularly provides a wind turbine generator fault early warning method and system based on an optimized deep belief network, and the method comprises the steps: obtaining SCADA data based on a data collection and monitoring control SCADA system installed in a wind field; wherein the SCADA data comprises wind field surrounding environment parameters and wind turbine generator component performance parameters; carrying out abnormal value detection on the SCADA data by utilizing a KNN-box graph combination algorithm; performing parameter optimization on the deep belief network DBN in combination with a sparrow search algorithm SSA and a particle swarm optimization algorithm PSO, and constructing an SSA-PSO-DBN-based wind turbine generator fault early warning model; based on the wind turbine generator set fault early warning model, early warning of the wind turbine generator set fault is completed. The algorithm provided by the invention has relatively high precision in judgment of the running state of the wind turbine generator, can further reduce the risk of misjudgment, and can realize early warning of the fault of the wind turbine generator at the same time.

The present application relates to the technical field of communication security, and in particular to a security authentication method, system and electronic device, comprising: sending an algorithm support list, a hybrid key exchange parameter and a first random number, a first certificate request to a second terminal, so that the second terminal generates a second handshake key and a ciphertext; determining a first handshake key and verifying a target first certificate chain sent by the second terminal using the first handshake key to obtain an authentication result, the present application performs authentication based on the hybrid key exchange parameter, so that the terminal has security against quantum attacks, reduces the leakage of plaintext parameters in the initial handshake message in the whole identity authentication process, uses different encryption processing for subsequent messages, reduces the risk of identity authentication key leakage, and improves the security of identity authentication. Meanwhile, the algorithm selection list is sent in the form of a combination algorithm, which can effectively reduce the algorithm selection matching time of the second terminal and improve the identity authentication efficiency.

A kind of sewer state intelligent diagnosis method and system based on machine learning, the present application relates to the field of digital data processing, in the method, build abnormal data identification model;Abnormal data identification model is obtained based on abnormal data identification result;LGB algorithm and LSTM-Attention combination algorithm are used to clean liquid level data, and the liquid level data after cleaning is obtained;Build rainfall inflow event identification model;Identify high value state;When the duration of high value state is greater than the preset duration, time series correlation analysis is carried out between high value state time period and rainfall data, and the inflow event caused by rainfall is obtained;Trend item data is obtained;Based on trend item data, inflection point detection is carried out to obtain inflection point data;Inflection point data and topological relationship data are associated and analyzed, and the pipe and channel blockage position is obtained in combination with upstream and downstream connection relationship;Diagnosis result is generated based on inflow event and pipe and channel blockage position.The present application improves the accuracy and reliability of the diagnosis of sewer state.

The invention discloses a working condition self-adaptive joint module motor control method for a quadruped robot, which comprises the following steps: S1, after a system is started, loading a preset standard parameter library of six working conditions of constant-speed walking on the flat ground, acceleration / deceleration on the flat ground, climbing, downhill, running and jumping and obstacle crossing, and calibrating a zero point of a multi-source sensor; a flat ground constant-speed walking working condition control mode is entered by default; and S2, collecting motor parameter data in real time through a multi-source working condition sensing module, carrying out fusion analysis on the collected motor parameter data by adopting a combinatorial algorithm, and carrying out working condition identification once every 5-10ms. Core parameters are collected through fusion of the multi-source sensor, a feature extraction and fuzzy neural network combined algorithm is combined, working condition recognition is completed every 5-10 ms, the recognition accuracy of six kinds of working conditions is high, the problem of working condition distinguishing fuzziness caused by dependence of a single parameter in the prior art is solved, and the working condition recognition accuracy is improved. And differential control requirements of complex scenes such as flat ground, climbing, running and jumping and the like can be accurately met.

The invention relates to a risk control rule generation method and system, a readable medium and equipment, and relates to the technical field of natural language processing. According to the scheme, a natural language processing model is adopted to execute reasoning tasks on a risk control file data set to extract structured conditions, and a risk control condition set is obtained; and executing a condition combination algorithm in the risk control condition set, matching the structured condition to generate an initial rule, and outputting the initial rule in a standard formatting manner to provide a risk control rule. According to the scheme, natural language processing and a condition combination algorithm are combined to realize automatic analysis and interpretation of risk control files, accurate extraction of risk control conditions and generation of risk control rules according with file description, so that the conversion cost of the risk control rules is reduced, the rule extraction efficiency is improved, the updating iteration period is shortened, and the risk control efficiency is improved. In addition, the problem of deviation of risk control rule conversion is overcome to a certain extent, the structured degree of the generated risk control rule is improved, subsequent analysis and interpretation are facilitated, and improvement of processing cost and error risk caused by secondary conversion is avoided.

Embodiments of the present application relate to the field of unmanned aerial vehicles, and provide a cache communication and deployment scheme generation method and device for air-ground integrated wireless networks, computer equipment and a storage medium. The method comprises: obtaining initial model parameters; constructing an unmanned aerial vehicle communication and deployment optimization model based on the initial model parameters; solving the unmanned aerial vehicle communication and deployment optimization model by using a combination algorithm of a particle swarm optimization and a Hungarian algorithm; constructing an unmanned aerial vehicle cache optimization model based on a function relationship between optimal values of the unmanned aerial vehicle communication and deployment optimization model and decision variables; solving the unmanned aerial vehicle cache optimization model by using a Monte Carlo simulation and an improved particle swarm optimization algorithm; and generating a cache communication and deployment scheme for air-ground integrated wireless networks according to a communication and deployment sub-scheme and a cache sub-scheme. The method can reduce the implementation cost of unmanned aerial vehicle-assisted wireless networks and improve the implementation effect of air-ground integrated wireless networks.

The invention discloses a UI scheduling optimization method and system of an end-to-end OCR architecture, and relates to the field of software test.The method comprises the steps that text recognition is conducted on a to-be-processed picture based on the OCR image recognition technology, and a plurality of keywords of the to-be-processed picture are obtained; dynamically combining the keywords by using a preset dynamic combination algorithm to obtain combined keywords; on the basis of the combined keywords, UI automatic test cases corresponding to semantics of the combined keywords are obtained; optimizing the automatic test case of the mobile terminal UI according to the case passing rate to obtain an optimized automatic test case, and performing automatic test on the mobile terminal UI according to the optimized automatic test case through a preset automatic test algorithm; according to the method, the test case can be more accurately and comprehensively designed, and the accuracy of automatic testing is guaranteed; the automatic test case is automatically tested, and then the test case is optimized according to the passing rate, so that the more comprehensive test case is realized.

The application belongs to the technical field of path planning, and discloses a robot path planning method based on combination of particle swarm, ant colony and A-Star algorithm, which comprises the following steps: generating a grid map, obtaining positions of obstacles and feasible areas in the grid map, constructing an accessible distance matrix, determining starting positions and target positions of each robot in the grid map in a robot cluster; and using the particle swarm, ant colony combination algorithm and A-Star algorithm to perform path optimization planning on the robot cluster based on the accessible distance matrix, the starting positions and the target positions of the robots in the grid map, so as to obtain an optimal path. The application combines the ant colony algorithm, the algorithm and the simulated annealing algorithm with the particle swarm algorithm, and performs robot obstacle avoidance path optimization, so that the robot can quickly find a path from the starting point to the end point while completing obstacle avoidance.

The application discloses an adversarial training method for improving robustness of a natural language processing model, and comprises the following steps: sending original text into the model for training; generating an adversarial sample of an attack model by using a combination algorithm; and finally adding the adversarial sample into original samples for adversarial training.

The invention discloses a low-temperature plasma jet flow vibration temperature diagnosis method based on machine learning, and belongs to the crossing field of plasma parameter diagnosis and machine learning technologies. The method comprises the following steps: S1, generating a theoretical spectrum through software, collecting an experimental spectrum, and accurately calibrating a temperature label for the experimental spectrum by adopting an Euclidean and SCM combined algorithm; s2, carrying out normalization processing on all spectral data, and dividing the spectral data into training, verification and test sets; s3, constructing and training two neural network models of BPNN and MLP, and learning a mapping relation between the spectrum and the temperature; step S4, optimizing the model through multiple modes such as verification set expansion, regularization and transfer learning to prevent overfitting, and using the optimized model to carry out rapid and real-time temperature diagnosis on the plasma spectrum; the method can solve the problems that the traditional manual fitting nitrogen molecule radiation spectrum diagnosis method is long in time consumption and highly depends on the experience of an operator.

This invention provides a bidding information management system and method based on Internet technology. It collects and preprocesses bidding data from participating parties; employs a combined algorithm combining the Analytic Hierarchy Process (AHP) and entropy weighting to perform multi-dimensional real-time comparison and comprehensive scoring of the preprocessed bidding data, yielding bidding results and rankings; it encrypts the bidding data and calculation results using dual encryption technology and constructs a secure transmission channel with firewall protection; it pushes the bidding results to all participants in real time through diversified notification channels; and it displays detailed results, rankings, and scoring criteria through a visual interface, enabling transparent querying of bidding information. This invention solves the problems of delayed result calculation, untimely notification, insufficient data security, and low transparency in existing bidding systems, significantly improving bidding efficiency and enhancing public trust.

The invention discloses an AIoT-based alarm management method and device, and the method comprises the steps: firstly collecting multi-dimensional basic data through an AIoT terminal, enabling a device management system and a user operation platform to be connected to obtain a device historical record and a user operation track, uploading the historical record and the user operation track to a cloud platform after preprocessing, and constructing a multi-dimensional data set; identifying valid and invalid alarm data from the data set by using a random forest and isolated forest combined algorithm; combining the data set and alarm data, analyzing historical fault, equipment aging and environment trend characteristics through a bidirectional LSTM neural network, outputting a fault risk value, and triggering a pre-alarm; then, constructing an equipment association graph by means of a graph neural network, positioning multi-equipment alarm sources, and merging alarms; and finally, the AIoT visual collaborative operation and maintenance platform dispatches an order according to a rule to track the progress, a statistical module generates a report, and alarm full-life-cycle management is realized. According to the invention, the problems of high false alarm rate, lack of fault prediction capability and the like in the prior art are solved.

The application discloses a continuous robot inverse kinematics combination algorithm, and an inverse kinematics solving method comprises the following steps: 1, according to the single joint arm segment parameter, establishing the robot link coordinate system and the tool coordinate system; 2, based on the constant curvature method, according to the geometric relationship, establishing the transformation between the joint rotation angle and the end posture angle; 3, according to the given end posture matrix T en d, solving the end posture angle analytical solution and the inverse kinematics analytical solution; 4, according to the obtained end posture angle analytical solution and inverse kinematics analytical solution and the established transformation between the joint rotation angle and the end posture angle, calculating the inverse kinematics approximate solution; 5, taking the obtained inverse kinematics approximate solution as the initial value, using the numerical iteration method based on the motion Jacobian matrix, calculating the inverse kinematics accurate solution. The inverse kinematics combination algorithm is clear and intuitive, has the advantages of high calculation precision and fast operation speed, and reduces the error generated by the traditional inverse kinematics solving method based on the constant curvature method.

The embodiment of the application discloses a multi-modal data-driven intelligent cabin parameter adaptation method and system, and relates to the technical field of intelligent cabin human-computer interaction. The method synchronously collects physiological data, multi-modal interaction data, driving behavior and scene data, extracts core features after preprocessing, adopts an evidence theory attention mechanism combination algorithm to construct a dynamic fusion adaptation model, realizes accurate and real-time adaptation of multiple cabin parameters such as sitting posture, vision, control and environment, and covers special groups such as the elderly, children and disabled users and all driving scenes. The application solves the defects of the prior art, such as insufficient data fusion dimension, static adaptation model, lack of psychological support and special group adaptation, parameter prediction error and whole-process delay, significantly improves driving safety and driving comfort, and is suitable for parameter optimization of various intelligent cabin products.

This invention discloses a method, system, medium, and product for constructing an elevation datum based on a low-order cosine-corrected Stokes kernel function, relating to the field of physical geodesy. The method includes: calculating model data using an Earth gravity field model, topographic data, and a high-frequency topographic influence combination algorithm; determining the direct and indirect effects of topographic compression on ground gravity anomalies and the elevation datum based on topographic compression; determining empirical and test values ​​of the correction order of the low-order cosine-corrected Stokes kernel function based on the cumulative error of the model gravity anomaly and the test frequency bandwidth; calculating an empirical and test elevation datum using the low-order cosine-corrected Stokes kernel function based on the empirical and test values ​​of the correction order; and determining the final elevation datum based on the empirical and test elevation datums. This invention can improve the accuracy of elevation datum construction.

This invention discloses a method and apparatus for evaluating wind turbine retrofitting schemes based on operational data, relating to the field of wind power engineering technology. The method includes: collecting at least five years of operational data from the target wind farm; preprocessing the data using the DBSCAN-segmented quartile combination algorithm; restoring the free-flow wind speed through primary and secondary correction functions to separate the impact of wind resource fluctuations and obtain the net power generation revenue increment from the retrofit; and establishing a multi-dimensional comprehensive evaluation system to assess the degree to which the technological upgrade improves the overall benefits of the wind farm. The apparatus includes a data acquisition module, a correction module, and an evaluation module, which execute the steps described above. This invention solves the problems of insufficient representative wind measurement data and single evaluation indicators for retrofitting schemes. By utilizing existing operational data from wind farms, it can effectively isolate the impact of wind resource fluctuations, explore the net power generation revenue increment brought about by the technological upgrade of wind turbines, improve the accuracy and comprehensiveness of the evaluation, and provide a scientific basis for the selection of retrofitting schemes.

The application discloses a kind of Web end real-time path planning method, system, storage medium and equipment, it is related to path planning technical field, the method includes: in the driving process of vehicle, traffic data and environmental change information in the application scene of Web end are collected in real time and are converted into the standard data format for algorithm processing;Path initial planning is carried out using hybrid algorithm, and the initial path is obtained;The actual driving information of path is obtained with the user feedback of environmental perception device, the actual driving information of path is compared with initial path, whether there is difference in actual path in the actual driving information of path is judged;If yes, according to the real-time correction of the recommended path in navigation system according to preset proofreading rule, to obtain target navigation path for sharing to other vehicles;Wherein, hybrid algorithm is the combination algorithm of genetic algorithm and ant colony algorithm.The application realizes the real-time optimization of distribution path, efficiently responds to various road condition changes and satisfies customer individualized demand.

The application belongs to the technical field of wind turbine units, and specifically provides a wind turbine unit fault early warning method and system based on an optimized deep belief network, comprising: a data acquisition and monitoring control SCADA system based on a wind farm installation, which obtains SCADA data; wherein the SCADA data comprises wind farm surrounding environment parameters and wind turbine unit component performance parameters; a KNN-box chart combination algorithm is used to detect outliers of the SCADA data; a sparrow search algorithm SSA and a particle swarm algorithm PSO are combined to optimize parameters of a deep belief network DBN, and a wind turbine unit fault early warning model based on SSA-PSO-DBN is constructed; and based on the wind turbine unit fault early warning model, early warning of wind turbine unit faults is completed. The algorithm provided by the application has high accuracy in judging the running state of a wind turbine unit, can further reduce the risk of misjudgment, and can realize early warning of wind turbine unit faults.

The invention discloses an intelligent regulation and control system and method for a coal mine gas extraction pipeline, and relates to the technical field of gas extraction. The method comprises the following steps that S1, extraction drilling parameters are determined, and pipeline layout is completed; s2, deploying a multi-parameter sensor network; s3, determining model hypothesis and variable definition; s4, establishing a multi-objective optimization mathematical model; s5, carrying out initial solution based on an improved genetic algorithm; s6, performing secondary optimization based on the improved BP neural network; step S7, executing branch regulation and control based on the optimal negative pressure; and step S8, dynamic period regulation and model iteration updating. Through the constructed multi-objective optimization model, the genetic algorithm preliminary optimization and the improved BP neural network secondary correction combination algorithm, the negative pressure requirements of all the branches in different extraction stages can be accurately matched, the extraction efficiency of the single branch and the total extraction efficiency of the whole working face are improved, on-demand energy consumption is achieved, and the extraction cost is reduced.

The invention discloses an integrated power supply monitoring control method applied to engineering machinery, and belongs to the technical field of power supply monitoring control. The method specifically comprises the following steps: S1, power supply parameter multi-dimensional time sequence acquisition: acquiring power supply multi-dimensional parameters and working conditions, synchronously recording the parameters, an operation mode and intensity, dynamically adjusting the acquisition frequency according to the working conditions, verifying the integrity of each group of data blocks, and storing and uploading original data to a cloud according to time sequence blocks; s2, preprocessing the collected data: performing multi-domain noise reduction, anomaly repair and standardized preprocessing on the original time sequence data block, and outputting a preprocessed clean time sequence data set; aiming at the working scenes of high electromagnetic interference, dramatic change of temperature and humidity and serious dust pollution of engineering machinery, a combined algorithm of Kalman filtering and wavelet threshold filtering is adopted in a data preprocessing link to carry out noise reduction on electrical parameters such as voltage and current, and temperature parameters are processed in combination with sliding average filtering and sudden change suppression; and more than 90% of high-frequency electromagnetic interference and instantaneous temperature fluctuation can be eliminated.

The invention relates to the technical field of concrete strength detection, in particular to a method and a device for predicting concrete degradation in a plateau environment based on a combinatorial algorithm. The method comprises the following steps: constructing a first data set; decomposing the first data set according to an improved EMD (Empirical Mode Decomposition) model to obtain a second data set; performing single prediction on the second data set to obtain a single prediction value of the second data set; integrating the single predicted values of the second data set to obtain the single predicted values of the first data set; constructing a variable weight combination prediction model according to the single prediction value of the first data set; and verifying the effectiveness of the variable weight combination prediction model according to a model advantage and disadvantage criterion. The concrete degradation in the plateau environment is predicted by using the combinatorial algorithm, and the prediction precision can be effectively improved.

The invention discloses a data transmission display method between a mobile device and an external display device, and the method comprises the steps: uniformly converting to-be-displayed message contents into binary dot matrix data through the mobile device, and carrying out the efficient compression through employing a combined algorithm including stroke coding and dictionary compression; then, the compressed data and the display parameters are packaged into a special binary protocol packet, and the special binary protocol packet is sent to display equipment through wireless communication; and the display equipment directly analyzes and decompresses the dot matrix data, and controls the screen to perform self-adaptive display according to the display parameters. According to the method, the dependence of a traditional text transmission mode on a display equipment word stock is overcome, the hardware resource overhead is remarkably reduced, meanwhile, efficient transmission and dynamic and diversified display effects of a complex message format are achieved through efficient compression and a flexible protocol, and the method is particularly suitable for embedded scenes such as an intelligent display and a vehicle-mounted screen.

This application provides a path planning method and system for collaborative coverage search of unmanned surface vessel (USV) swarms. The method includes: dividing the USV swarm into several subsets and determining the candidate path points and corresponding search intensities for each subset; using an ant colony algorithm based on multi-ant combinations to determine the optimal global path for each subset to conduct collaborative coverage search of the target water area, wherein the multi-ant combination algorithm is implemented through several multi-ant combinations, each multi-ant combination consisting of several ants representing each subset, and in each iteration, each ant in each multi-ant combination sequentially and cyclically selects the next path point from the candidate path points corresponding to the subset it represents; and assigning the optimal global path corresponding to each subset to all USVs in that subset. The technical solution of this application can rationally plan the path for collaborative coverage search of the target water area by multiple USV swarms.

The application discloses a sorting method and system of liquid metal batteries, and belongs to the technical field of secondary battery applications. The method comprises the following steps: reconstructing sampled liquid metal battery discharge voltage data to obtain a smooth voltage curve; identifying a voltage curve inflection point to generate a curve representation index; repeatedly mapping and screening the curve representation index to form a sorting index; filtering sample outliers in the sorting index by using an improved DBSCAN clustering algorithm; and then adopting a Mean Shift algorithm to optimally divide the remaining sample space to obtain a battery sorting result. The improved DBSCAN clustering algorithm is used to filter sample outliers, and then the Mean Shift algorithm is used to optimally divide the remaining sample space, so that the combined algorithm can simultaneously achieve the purposes of outlier detection and clustering, thereby improving the sorting precision.

The invention provides an unmanned aerial vehicle and supply vehicle heterogeneous task joint scheduling method for agricultural plant protection. The method comprises the following steps: performing task plot partitioning according to adjacent spaces and similar task types; performing task performability verification and task unique allocation; performing task sub-segment division on the task block; determining an operation time sequence of the task sub-segments in the same task plot; calculating the soft constraint penalty of the operation time window of the task sub-segment based on the operation time sequence; gSV-UAV binding and GSV route optimization are carried out on the basis of the task sub-segments; on the basis of supply waiting penalty and same-vehicle non-overlapping constraint, supply time sequence and operation time sequence coordination of task subsections is carried out, and UAV continuous operation interruption statistics is carried out; constructing a comprehensive scheduling model, and solving the comprehensive scheduling model based on a GA-SA-LKH combinatorial algorithm to obtain a baseline scheduling scheme; and executing and monitoring according to a baseline scheduling scheme, if real-time disturbance exists, performing local re-optimization in a rolling window through multi-agent reinforcement learning, and stopping scheduling when no real-time disturbance exists or a scheduling task is completed. According to the invention, the overall operation efficiency and scheduling robustness of the system are effectively improved.