62 results about "Multidimensional optimization" patented technology

The invention provides a chip production logistics optimization method and system based on a multi-target genetic algorithm, and relates to the technical field of chip production scheduling, and the method comprises the steps: collecting the real-time operation data of a production line, extracting task execution features and equipment layout features through a dual-channel deep neural network, and carrying out the feature fusion through a recurrent neural network, generating a production state vector containing a multi-dimensional scheduling index predicted value; constructing a multi-dimensional optimization target set based on the production state vector, and generating an initial population by adopting a double-layer gene coding structure; executing non-dominated sorting and crowding degree distance calculation based on the fitness value, and selecting high-quality individuals to form a parent population; and performing adaptive crossover, tabu search optimization and adaptive mutation operation on the parent population, iteratively selecting a new generation of population until convergence, and outputting an optimal scheduling scheme.

The invention relates to the technical field of smart city traffic, and discloses a smart city traffic dynamic optimization system and method based on digital twinning. The system obtains urban traffic network multi-dimensional data from a plurality of heterogeneous data sources through a traffic multi-dimensional data acquisition module and integrates the urban traffic network multi-dimensional data into a traffic related data warehouse; a traffic digital twinning model construction module extracts features from the data warehouse to generate a traffic related feature matrix, and a digital twinning traffic dynamic model is constructed according to the traffic related feature matrix to output a theoretical traffic state value; the traffic flow map construction module determines a dimension link map of each dimension and constructs a traffic flow link map; the traffic core feature screening module screens a traffic core feature sequence based on the map; the traffic multi-dimensional optimization analysis module performs multi-dimensional difference analysis on the theoretical traffic state value and real-time actually measured traffic data, and generates a region-level difference coefficient matrix in combination with the core feature sequence; and the traffic event association positioning module can realize accurate management and dynamic optimization of urban traffic.

The invention relates to the technical field of mine filling, in particular to a filling method of a mine cementing material based on a volume ratio. The method comprises the following steps: for different filling strengths, adaptively generating a filling formula through an adaptive multivariable ratio optimization algorithm, and obtaining a reference formula set; during mine filling operation, process parameters are obtained in real time, and in combination with the reference formula set, the volume ratio of the cementing material to the tailing slurry in the finished product slurry is calculated; calculating the real-time addition amount of the cementing material based on the volume ratio of the cementing material to the tailing slurry; a delay compensation and multi-dimensional optimization predictive control algorithm is introduced, the real-time addition amount of the cementing material is optimized, and the corrected addition amount of the cementing material is obtained; and based on the corrected cementing material addition amount, filling of the mine cementing material is achieved. The technical problems that in the mine filling process, control over the adding amount of the cementing material is not accurate, and self-adaption is poor are solved.

The invention discloses an automatic equipment scheduling method and system for an electromagnetic compatibility test, and the method comprises the steps: monitoring the state and spatial position information of each piece of equipment in real time, and calculating the electromagnetic radiation intensity of the equipment; sorting all the test tasks according to the priorities, and sequentially arranging the equipment from high priority to low priority for testing to form an initial equipment scheduling scheme; calculating the electromagnetic interference gradient of the equipment scheduling scheme, dynamically adjusting the equipment scheduling sequence by adopting the electromagnetic interference gradient, and updating the equipment scheduling scheme; the electromagnetic interference intensity between the devices is updated in real time, and then a device state interference matrix is dynamically evolved; global coupling feedback is calculated; and solving a multi-dimensional optimization objective function which comprehensively considers the total interference variable, the equipment state interference matrix and global coupling feedback under the current scheduling scheme to obtain an optimal equipment scheduling scheme. The method is high in scheduling efficiency, full in resource utilization, accurate in electromagnetic interference control, and capable of effectively meeting complex multi-device collaborative test requirements.

The invention discloses a 16-channel high-precision synchronous data acquisition system based on a multi-parameter model, which belongs to the technical field of data acquisition and comprises a multi-parameter model driving module, a self-adaptive channel scheduling module, a closed-loop calibration compensation module and a collaborative synchronous optimization module. The multi-parameter model driving module establishes a multi-dimensional optimization model fusing the sampling rate, the channel load and the error trend to output optimization parameters; the adaptive channel scheduling module performs real-time quality evaluation and dynamic resource allocation according to the optimization parameters to generate a scheduling instruction; the closed-loop calibration compensation module monitors a synchronization error in real time and calculates a compensation parameter; the collaborative synchronous optimization module realizes inter-channel parameter sharing and collaborative control to generate a synchronous control signal, and feeds back an optimization result to the multi-parameter model driving module to form a closed loop, and through deep coupling and closed loop feedback of the four modules, high-precision synchronization of multi-channel data acquisition is realized, and the synchronization precision reaches + / -35ns.

A network security script arrangement method based on LLM enhanced RL comprises the following steps: 1) LLM dynamically expands a to-be-selected strategy atom set based on a security threat scene: firstly, performing semantic similarity retrieval on input security threat intelligence by using a vector library, and matching an optimal defense strategy atom in a knowledge library; the context understanding capability of the LLM is then utilized to implement semantic understanding and reasoning on the candidate atom set, generating new policy atoms complementary to the defense. 2) RL refers to LLM atomic action preference to optimize a script generation strategy, and self-adaptive security arrangement of a complex network environment is realized: firstly, a security script graph model based on strategy atoms is adopted, and multiple constraints corresponding to nodes and edges in the graph model are abstracted on the basis of security arrangement definition; and then, a composite reward function is adopted to model a linear programming function for the multi-constrained security script arrangement problem, so that multi-dimensional optimization of defense efficiency and execution cost is realized.

The invention discloses a multi-beam satellite system transmission optimization method and device fused with image semantic compression, and the method comprises the steps: constructing a multi-beam satellite system model fused with image semantic compression, determining an area irradiated by a multi-beam satellite beam according to a beam hopping mechanism through a multi-beam model, and determining a GPU resource allocation scheme and a power resource allocation scheme; a satellite communication channel state is sensed through a semantic communication model based on a SwinJSCC architecture so as to adjust a compression ratio of image semantic information, and multi-dimensional optimization of image data transmission performance is realized; a multi-agent is constructed based on a multi-beam satellite system model and a semantic communication model, the agent corresponds to a beam hopping mechanism, a GPU resource allocation scheme and a power resource allocation scheme in the multi-beam satellite system model, and the agent learns and optimizes a decision strategy under a multi-agent architecture. Therefore, the resource scheduling in the dynamic hopping beam scene is matched with the service demand.

The invention discloses a federal learning-based cross-device brain grain model training system. The system comprises a terminal device layer, an edge coordination layer and a central server layer, the terminal equipment layer is responsible for collecting original electroencephalogram data and executing local lightweight training; the edge coordination layer is used for aggregating model updates of a plurality of terminal devices in the jurisdiction to generate a regionalized sub-model; according to the central server layer, a central server performs secondary optimization on each regional model to generate a global unified model; the encrypted global model and strategy configuration are pushed to all edge coordination layers; and the edge coordination layer pushes the optimized model and strategy to terminal equipment. Through a three-layer federated architecture and multi-dimensional optimization, the system efficiency is remarkably improved: the terminal equipment only uploads an encrypted model increment, and source data protection is realized in combination with differential privacy and homomorphic encryption; federal transfer learning shortens the cold start time of new equipment.

A training optimization system implements algorithmic solutions to solve the conjoined problem of accelerator architecture search and model partitioning for distributed training. The system makes the multi-dimensional optimization space of architecture search and device placement tractable by reducing the number of accelerator architectures explored through area-based heuristics and employing a novel integer linear program (ILP), the size of which is dependent only on the number of operators. The ILP scheduling optimization also explores the partitioning of operators across cores, known as intra-operator parallelism. Despite the vast space, the ILP described herein requires significantly less time to perform the optimizations across all explored accelerator configurations. Based on the optimal backward and forward pass latencies, the system leverages a novel dynamic programming (DP) approach to determine the device placement and model partitioning scheme.

The invention relates to a multi-path RAG result optimization method and device and electronic equipment. The method comprises the steps that a query instruction of a user is received, information related to the query instruction is retrieved from multi-source data through a standardized interface, and a retrieval result is obtained; performing hybrid coding and duplicate removal processing on the retrieval result to form a candidate information pool; based on the candidate information pool, calling a heterogeneous large language model to generate multiple paths of candidate results in parallel; and calling an independent evaluation large language model to perform multi-dimensional quality evaluation on the candidate result of each path, and based on a quality evaluation result, determining an optimal result from the candidate results of all paths and returning the optimal result to the user. The technical problems that the multi-source RAG result screening of a traditional RAG system is single, and a dynamic and multi-dimensional optimization mechanism is lacked are solved.

The invention provides a water pump fault diagnosis method and device based on voiceprint recognition and a medium, and the method comprises the steps: obtaining a water pump voiceprint signal, and employing a sparrow search algorithm optimization-based VMD decomposition method to decompose the voiceprint signal to obtain a preliminary decomposition signal; processing the preliminary decomposition signal to obtain an FBANK energy spectrum and a GTFB optimization frequency band energy distribution feature, and then obtaining a multi-dimensional optimization feature vector through a feature weighted fusion strategy; and inputting the multi-dimensional optimization feature vector into a preset fusion model to realize fault classification. According to the technical scheme provided by the invention, accurate identification of various faults of the water pump bearing in a complex industrial scene is realized, and a new technical path is provided for intelligent operation and maintenance of the water pump bearing.

The invention discloses an urban drainage pipe network optimization method and device based on big data, and relates to the technical field of data processing, and the method comprises the steps: constructing a plurality of feature rainfall scenes based on historical rainfall event differential clustering, evaluating the flood drainage response standard reaching performance of a drainage pipe network in each scene, screening and generating a target rainfall scene, and carrying out the optimization of the target rainfall scene. The method comprises the following steps: performing adaptive layout adjustment in combination with overground and underground data, establishing a layout adjustment space, performing multi-degree-of-freedom optimization based on a pipe network loss prediction model, forming an optimization strategy, performing state compensation through real-time monitoring data, and finally generating a dynamic adaptive drainage pipe network optimization scheme. The technical problems that an existing urban drainage pipe network lacks systematic evaluation and a multi-dimensional optimization mechanism when coping with a complex rainfall scene, the flood drainage response capability is insufficient, and the operation efficiency is low are solved, and the purposes of multi-scene optimization and real-time monitoring compensation are achieved. And the technical effects of flood drainage response capability and operation efficiency of the drainage pipe network are effectively improved.

The invention provides a multi-dimensional optimization-based unmanned aerial vehicle data security acquisition method and system, and the method comprises the steps: constructing an unmanned aerial vehicle-assisted Internet of Things data acquisition system which comprises an unmanned aerial vehicle, a plurality of sensors, a plurality of auxiliary devices and a potential eavesdropper, discretizing a task acquisition period into a plurality of time slots; introducing a sensor scheduling variable to perform dormancy wakeup management on the ground sensor, and introducing an interferer selection variable to select an auxiliary device as a friendly interferer based on a channel condition; establishing a path loss model of a ground-air channel and a ground-ground channel to obtain a channel gain, defining a transmission rate and an eavesdropping rate, considering the uncertainty of the position of an eavesdropper to construct a worst-case secrecy rate model of the unmanned aerial vehicle, and setting constraint conditions; a data acquisition problem is formalized into a secrecy rate maximization problem and decomposed into three sub-problems, the sub-problems are solved, and an optimized unmanned aerial vehicle trajectory, sensor scheduling and jammer selection strategy is output.

The invention discloses a low-orbit satellite beam switching method and a communication terminal, and belongs to the technical field of satellite communication, and the method comprises the steps: synchronously collecting real-time multi-source data; preprocessing the real-time multi-source data to obtain a real-time multi-dimensional feature vector, and based on the real-time multi-dimensional feature vector, performing link quality prediction on links of all visible beams in a period of time in the future by adopting a prediction model to obtain an updated candidate beam set, and providing a time window for decision making; and constructing a comprehensive decision function fused with a multi-dimensional optimization target, calculating a comprehensive score of the updated candidate beam set, identifying a global optimal beam, and performing switching judgment and execution. According to the invention, the switching strategy conversion from passive response to active optimization is realized, the time delay and interruption probability of beam switching are obviously reduced, and the system resource utilization rate and the communication quality are improved.

The invention discloses an intelligent thermal power plant unit performance multi-dimensional optimization management and control system, relates to the technical field of thermal power plant units, and solves the problems that the energy consumption states of a plurality of units are not balanced and a thermal power plant does not have an optimal adjustment standard, resulting in large errors of adjusted related parameters. According to the method, the historical operation data of the thermal power plant unit is analyzed, and the power parameters associated with the thermal power plant unit are subsequently adjusted based on the actual analysis result, so that the subsequently adjusted specific parameters are applied to the actual operation state, and the operation optimization effect of the corresponding thermal power plant unit is guaranteed; and the overall operation effect of the thermal power plant unit is improved.

A training optimization system implements algorithmic solutions to solve the conjoined problem of accelerator architecture search and model partitioning for distributed training. The system makes the multi-dimensional optimization space of architecture search and device placement tractable by reducing the number of accelerator architectures explored through area-based heuristics and employing a novel integer linear program (ILP), the size of which is dependent only on the number of operators. The ILP scheduling optimization also explores the partitioning of operators across cores, known as intra-operator parallelism. Despite the vast space, the ILP described herein requires significantly less time to perform the optimizations across all explored accelerator configurations. Based on the optimal backward and forward pass latencies, the system leverages a novel dynamic programming (DP) approach to determine the device placement and model partitioning scheme.

According to the intelligent optimization design method for the constant-force gripper of the robot provided by the invention, the black-wing algorithm is improved, and the BP neural network algorithm is combined, so that the parameters of the whole gripper and key components are subjected to multi-dimensional optimization, and the comprehensive performance of the gripper can be remarkably improved. Finite element analysis is carried out based on an optimization result, an optimal design scheme meeting industrial requirements can be quickly generated in a virtual environment, a theoretical basis and practical guidance can be provided for production of similar products with light weight and high strength indexes in cooperation with optimization effect experimental verification of an entity prototype, the design time and cost are greatly shortened, and the design efficiency is improved. And the repeated experiment process in the traditional design is also avoided.

The invention relates to an energy scheduling method and device and a medium. Comprising the following steps: acquiring historical original flow data and channel state information of an underwater node; generating a traffic distribution matrix according to the historical original traffic data and the channel state information; obtaining a multi-dimensional optimization model; inputting the flow distribution matrix into a multi-dimensional optimization model to obtain an energy scheduling signaling; and the underwater node receives the energy scheduling signaling and executes energy scheduling operation. According to the technical scheme, a base station obtains historical original flow data and channel state information of an underwater node; and then, the base station generates a flow distribution matrix according to the historical original flow data and the channel state information, and inputs the flow distribution matrix into the multi-dimensional optimization model to generate energy scheduling signaling. Therefore, the base station predicts the flow of the underwater node according to the historical original data and the channel state information of the underwater node, and then dynamically generates the energy scheduling signaling according to the prediction result of the flow, thereby effectively reducing the energy loss of the underwater node.

The application discloses a physical examination user consultation conversion prediction method and system based on a time sequence network, relates to the consultation data conversion prediction field, and forms a precise effect evaluation mechanism by training a prediction model based on historical behavior data, comparing a push result with actual user behavior in real time, and forming a precise effect evaluation mechanism. When the push accuracy is lower than a threshold value, a multidimensional optimization process is automatically triggered: first, the optimal configuration in the limited resources is found by adjusting the hyperparameters iteratively, and if it cannot be satisfied, the software performance index is further optimized, and finally the hardware configuration is upgraded when necessary; the hierarchical progressive optimization strategy ensures the timeliness of model updating and avoids resource waste. By including the training data volume, software and hardware indicators and other key parameters into the effect prediction model, the optimization direction can be predicted in advance, the adjustment efficiency is significantly improved, and finally the user behavior change can be continuously adapted, the push accuracy is ensured, and the resource utilization is maximized.

The invention relates to the technical field of navigation, in particular to a manual navigation method and system based on a knowledge graph, and the method comprises the following steps: obtaining a node version record, extracting an update event, calculating text difference and similarity to generate an update frequency sequence, calculating access fluctuation in combination with user behavior data, generating a score, extracting behavior characteristics to generate a label, and performing manual navigation. And analyzing a label transfer rule to predict an evolution trend, evaluating path stability and credibility by combining scores, and screening and outputting an optimal navigation path sequence for navigation. According to the method, the node updating features are extracted, the text similarity is calculated, the multi-dimensional feature vector is generated in combination with user behavior fluctuation, the label evolution trend is predicted, the optimal path is screened, the information adaptability and navigation flexibility are enhanced, the node matching degree and the jump accuracy rate are improved, the user jump rate is reduced, the search experience is optimized in a multi-dimensional mode, and accurate navigation is carried out.

This invention discloses a satellite communication terminal simulation task orchestration system and device based on intelligent scheduling. It adopts a microservice architecture, with five modules working collaboratively via RESTful APIs or gRPC. The system constructs a multi-dimensional global resource view, combining historical data to drive a regression prediction model to quantify resource requirements. The intelligent scheduling decision module models complex task allocation as a multi-dimensional optimization problem, using the PSO algorithm combined with a multi-objective fitness function to find the optimal solution, supplemented by a task preprocessing mechanism to improve efficiency and effectiveness. The system manages task dependencies using a DAG and deeply integrates with Kubernetes to achieve automated task orchestration, status monitoring, and fault recovery. This effectively solves the pain points of traditional scheduling, such as resource fragmentation and static configuration waste, achieving dual-dimensional global dynamic scheduling of business and resources. This significantly improves cluster resource utilization and simulation execution efficiency, adapting to the needs of large-scale, high-concurrency simulation scenarios.

The invention discloses a universal computing resource deployment method based on a multi-population differential evolution algorithm, and belongs to the technical field of universal computing resource deployment. Comprising the following steps: S101, analyzing power business scene requirements and general inductance network resource performance indexes, and establishing an evaluation index system; s102, combining a BWM method with an entropy weight method, and performing adaptive analysis on the power business demands and the general inductance computing resources to obtain demand weights of different power businesses on the general inductance computing resources; s103, according to an adaptive analysis result of the power business to the general inductance computing resources, establishing a general inductance computing resource deployment multi-dimensional optimization problem model; and S104, adopting a multi-population differential evolution algorithm to calculate a universal computing resource deployment multi-dimensional optimization problem model. Compared with other resource deployment optimization schemes, the method of the invention adopts a scheme of combining power business suitability analysis and a general inductance calculation resource deployment optimization problem, and is more in line with the actual demand of power business, and the proposed optimization problem is more reasonable.

The invention relates to the field of carbon emission optimization, and provides a method, a device and equipment for optimizing carbon emission in a vegetable production place treatment process, and the method comprises the steps: obtaining a plurality of link parameters in the vegetable production place treatment process, and obtaining carbon emission based on a preset carbon emission calculation model; determining a sensitivity analysis result according to the multiple link parameters, and selecting parameters with sensitivity higher than a set threshold and adjustable in actual operation as an optimized parameter set; constructing a deep reinforcement learning environment to operate a carbon emission optimization algorithm, and carrying out iterative optimization on a multi-dimensional optimization target; outputting an optimal optimization parameter combination with the highest reward value and a corresponding carbon emission reduction optimization scheme based on the optimization parameter set; the reward value is an optimization target of the carbon emission optimization algorithm. According to the method, the problems of poor adaptability and low optimization efficiency of a carbon emission optimization method in the vegetable producing area treatment process in the prior art are solved, and rapid identification and dynamic optimization of carbon emission key parameters under complex and changeable working conditions are realized.

The invention discloses a system for processing engineering cost business, belonging to the technical field of data processing management; it is used to solve the technical problem that engineering cost contract formulation business cannot be automatically formulated and multi-dimensional optimization and improvement cannot be implemented in the existing scheme; by utilizing the engineering cost contract database constructed by existing shared data to automatically search and screen according to the engineering type to obtain all qualified engineering cost sample contracts, and determine to obtain the selected draft through target personnel selection; by integrating and processing the sample remark data associated with all the same engineering cost sample contracts, calculating and analyzing the impact degree and impact analysis value of different sample remark regulations, obtaining the remark regulations in the selected draft in the early stage for dynamic supplement and improvement; obtaining the improved transition draft and the processed sample dispute handling table to perform risk avoidance analysis and mark supplement on all the remark regulations in the improved transition draft.

The present invention discloses an energy recovery control method and system for an electric articulated vehicle considering the risk of fishtailing. The method includes the following steps: building a non-linear predictive control model for energy recovery of the electric articulated vehicle based on a symmetric simplification strategy; constructing a multi-dimensional optimization objective function including fishtail suppression optimization and energy recovery optimization by taking balance of safety, energy recovery rate and braking performance as consideration factors; calling the non-linear predictive control model for energy recovery to solve the optimal control sequence based on the multi-dimensional optimization objective function, and calling the control elements of the electric articulated vehicle to control the vehicle according to the first element of the optimal control sequence; and calling the non-linear predictive control model for energy recovery for cyclic predictive control along with the update of the state of the electric articulated vehicle. The present invention can effectively estimate and avoid potential safety hazards such as fishtailing by accurately capturing the dynamic changes during the braking process, and ensure the stability and safety of the braking process.

The invention discloses a multi-dimensional optimization screening system for target interaction. The multi-dimensional optimization screening system comprises the following steps: S1, preprocessing drugs and target information in a pharmacomics database; s2, converting the molecular structure information of the medicine into topological graph representation; s3, performing embedding representation on the target spot sequence information; s4, inputting the extracted drug feature information and target feature information into a multi-task learning classification model for training, and storing the trained model; and S5, loading the trained model, inputting to-be-predicted drug and target spot information, and outputting a prediction result of the interaction relationship between the drug and the target spot. Through combination of a dynamic feature enhanced graph convolutional network (DFE-GCN) and a hierarchical attention module (HAM) of the multi-dimensional optimization screening system for target interaction, the extraction capability of drug molecular structure features is remarkably improved, and the system is particularly excellent in the aspect of capturing local and global features of drug molecules.

The invention discloses an existing operation station shield receiving construction step sequence optimization method and system based on digital twinning, and the method comprises the steps: collecting the associated data of an existing operation station, constructing a digital twinning basic model, achieving the full-factor digital mapping of entity engineering, and carrying out the parameterization and lightweight optimization. And establishing dynamic association between the model component and the original data. Dynamic monitoring and construction process data are collected in real time and are uniformly accessed through an integrated platform, and a real-time synchronization mechanism of the data and a model is established after preprocessing and screening. Based on the model and integrated data, setting a multi-dimensional optimization constraint, embedding a dynamic optimization algorithm, simulating, deducing and optimizing a construction step sequence by taking safety, high efficiency and low risk as targets, and generating an initial scheme; and when a preset trigger event occurs, re-optimization is carried out, and an adaptive scheme is output. And the scheme is pushed to a construction execution end, an execution state is tracked, actual and preset data deviations are compared, and an early warning mechanism is established. And smooth construction of the existing operation station is ensured.

The application discloses a kind of ground well transient electromagnetic multi-parameter three-dimensional fast inversion method and device, and main technical scheme is: first, the original transient electromagnetic response signal containing magnetic field intensity and magnetic field rate of change is collected, after being converted into magnetic field intensity data, high-precision magnetic field cumulative response data is obtained by time integration and error correction;Subsequently, using the inversion model of pre-built containing Cole-Cole model magnetic dipole forward equation, layered background field module and adaptive three-dimensional grid, extract only the observation magnetic anomaly data containing target geological body, build composite objective function and adopt steepest descent method iterative solution;Finally, through geological constraint, depth weight, variable regularization, multidimensional optimization is carried out to initial three-dimensional Cole-Cole parameter model, and the relative root mean square error of magnetic field cumulative response is used as convergence determination index, and iteration is carried out until convergence, and the final three-dimensional Cole-Cole parameter model is output as the inversion result.

The invention relates to the technical field of wind power generation research, and discloses an energy storage coordination management method, system, equipment and medium for wind power generation power control, and the method comprises the steps: building a dynamic prediction model according to the power fluctuation characteristics of a wind power generation system; designing a multi-dimensional optimization objective function of the energy storage system based on the dynamic prediction model; performing global optimization solution on the multi-dimensional optimization objective function through an upper-layer controller in hierarchical cooperative control, and performing local optimization solution on each energy storage unit through a lower-layer controller by adopting a distributed optimization algorithm; and the optimization solution result at the current moment is applied to the energy storage system, real-time data at the next moment are collected for optimization solution again, and real-time dynamic adjustment of the energy storage system is achieved in a rolling optimization mode. According to the method, the response speed and the adjustment precision of the energy storage system to high-frequency power change are remarkably improved, the problem of dynamic response lag in the prior art is practically solved, and the method can better adapt to the characteristic of high fluctuation of wind power generation power.

The invention provides a surface mine transportation scheduling method and device, and relates to the technical field of transportation scheduling, and the method can obtain engineering data of a target mining area, and employs a global search algorithm based on a gradient optimizer according to the engineering data and a target function constructed with the lowest total transportation cost as a target. And carrying out iterative optimization on the number of transportation times of each transportation route in the target mining area to obtain a target number of transportation times scheme, and further determining a vehicle scheduling scheme of the target mining area. Therefore, when the vehicle scheduling scheme of the target mining area is determined, a global search algorithm based on a gradient optimizer is adopted, and the problems of low operation efficiency and easy premature convergence of a particle swarm type method for a multi-dimensional optimization problem can be avoided through a gradient search rule and a local escape operator; the early-stage exploration capability and the later-stage development capability can be better balanced, and the optimal scheduling scheme can be quickly and accurately obtained.