102results about How to "Avoid double counting" patented technology

The invention discloses an unmanned aerial vehicle target intelligent identification and perception system and method based on deep learning, and relates to the technical field of artificial intelligence, and the method comprises the steps: extracting the importance feature of each target region, calculating a region importance score, generating a target region monitoring plan, and obtaining continuous frame images; an improved YOLOv5 model is adopted, continuous frames of images are input, a bounding box of a suspected target wild animal is output and mapped to an actual geographic coordinate system, and actual position coordinates are obtained; key points of the t-th frame image are detected, coordinates of the key points are extracted, visibility scores and confidence coefficients are calculated, and feature descriptors are formed; performing target detection and key point detection on the (t + 1) th frame of image to obtain actual position coordinates and feature descriptors of candidate targets, performing track matching, updating the actual position coordinates and the feature descriptors of the suspected target wild animals if the actual position coordinates and the feature descriptors are matched, and defining new target animals if the actual position coordinates and the feature descriptors are not matched; counting according to the track matching result to obtain the total number of the target wild animals.

The invention discloses a multi-light-spot centroid extraction method and device based on an FPGA, a medium and a product, and relates to the field of digital image processing, and the method comprises the steps: firstly, caching image line data through a first-stage FIFO and a second-stage FIFO, forming three continuous groups of image line data streams, effectively solving the problems of data synchronization and processing rate matching, and improving the data processing efficiency; then, effective pixels of the light spots are accurately determined by means of a pixel threshold value, and different light spot areas are accurately divided by means of line continuity judgment and boundary marking, so that repeated recognition and error recognition of the light spots are avoided, and the accuracy of light spot recognition is greatly improved. And finally, according to the weighted average positions of the gray values of the effective pixels of all the light spots, the target centroid coordinates of the target light spots are calculated, the internal gray difference of the light spots is fully considered, and compared with a conventional method, the light spot energy distribution center is more accurately reflected, so that the calculation precision of the light spot centroid is remarkably improved.

The invention relates to the field of graphic rendering, in particular to a high-efficiency graphic rendering system based on artificial intelligence core computing power, which comprises an intelligent task decoupling module for performing task decoupling on a graphic rendering task; the AI enhanced texture generation optimization module is used for learning a dynamic scene change rule to generate a texture map in real time, reducing calculation overhead in a repeated texture region by using a multiplexing algorithm, and performing super-resolution compensation on low-resolution textures by using a deep convolutional neural network; the AI driving illumination and shadow calculation module is used for predicting an illumination transmission path in a scene by using AI and reconstructing a dynamic shadow effect so as to realize high global illumination and shadow simulation; the task result synchronization module is used for synchronizing rendering data output by each subtask; according to the method, the high-quality texture map is dynamically generated, super-resolution compensation is carried out on the low-resolution texture, repeated calculation is avoided through a texture multiplexing algorithm, the calculation overhead is remarkably reduced, and the real-time rendering efficiency of a large-scale complex scene is improved.

The invention provides a network node classification method based on a multilayer graph neural network model, and the method comprises the steps: constructing a graph according to network nodes, carrying out the preprocessing of initial node features in the graph, and obtaining an initial embedded vector; according to the initial embedding vectors of the node and the neighbor node thereof, updating the node representation by adopting a layered message passing mechanism; according to the initial embedded vectors of the nodes and the neighbor nodes thereof, the graph attention network uses an attention mechanism to calculate weights for the neighbor nodes of each node and then dynamically updates feature vectors of the nodes; a multi-head attention mechanism is adopted to splice and merge the feature vectors of the nodes output by the graph attention network and the updated node representation, and the comprehensive feature vector of each node is obtained; the comprehensive feature vectors are further optimized through graph contrast learning and a node embedding module, and optimized features are obtained; and obtaining predicted node classification according to the optimization features. According to the method, complex features of heterogeneous graph data can be effectively captured, the method has good robustness for an over-smoothing problem, more accurate heterogeneous network representation learning is realized, and the method has advanced performance and robustness in a node classification task.

The invention relates to the technical field of novel power system optimization operation, in particular to a new energy high-uncertainty scene generation method and device based on a CWGAN-GP model, and the method comprises the steps: correcting an initial generative adversarial network through employing a Wasserstein distance and a gradient penalty term, so as to obtain a final generative adversarial network, and the target convolutional neural network is introduced into the final generative adversarial network to obtain a conditional deep convolutional generative adversarial network of improved deep convolution, so that WGAN-GP is introduced into the conditional deep convolutional generative adversarial network of improved deep convolution to obtain a CWGAN-GP model of improved deep convolution. And a target L-ISODATA clustering algorithm is applied to cluster the generated wind and light scenes so as to generate a high-uncertainty new energy power generation output scene. Therefore, the problems that in the related technology, when the GAN processes high-dimensional data, complex space-time characteristics are difficult to effectively capture, and due to lack of prior information, the training direction is prone to deviation, and the scene generation effect and stability are affected are solved.

The invention relates to the technical field of underwater static target object detection and processing, in particular to an unmanned cross-domain positioning and acoustic fingerprint processing method and system for an underwater static target. Comprising the following steps: performing wide-area scanning on a task sea area, and scheduling an AUV and unmanned aerial vehicle cluster to a target area after discovering a suspicious target; the unmanned aerial vehicle cluster receives an acoustic signal of the AUV and transmits the acoustic signal to the cooperative resolving center in combination with self-positioning data so as to provide accurate coordinate guidance for the AUV; the AUV sails to a target area, parallel processing map construction and accurate positioning, target identification and dual-mode fingerprint generation are carried out, and a target task package is packaged; the ROV receives and analyzes the task packet, autonomously plans a path and sails to a target area, a sonar is started to collect data and generate real-time feature fingerprints, and the real-time feature fingerprints are matched with fingerprints in the task packet; and after matching succeeds, a specific job task is autonomously executed. According to the invention, a complete automatic solution is provided for scenes such as deep and far sea detection, emergency salvage, pipeline maintenance and the like.

The invention provides a student classroom behavior detection method based on Transform and task dynamic alignment. On one hand, the GLTB is added into the backbone network to enhance the feature extraction capability. Wherein the CGLU is used for enhancing the non-linear feature expression capability, and the global feature extraction capability of the MHSA is combined, so that the method has higher feature extraction capability. And on the other hand, a neck structure network improved by GLSA is adopted, global features can be effectively extracted, fine-grained local features can be captured, and interaction between global information and local information can be promoted. According to the method, when a complex task is processed, key information in the image can be identified and understood more accurately. In addition, a novel detection head ETADH is adopted, interactive features of classification and positioning tasks are learned through a feature extractor, and task decomposition is performed through a task decomposition module. In this way, the interactivity between the two tasks is enhanced, and the detection precision is improved by effectively utilizing the associated information between the two tasks.

The invention provides a route-independent network measurement system which comprises a plurality of network measurement points and a centralized controller, each network measurement point maintains a route-independent data sketch, each network measurement point is of a multi-layer structure, each layer comprises a plurality of units provided with bitmaps, and initial values of all bits in the bitmaps are first numerical values; the network measurement point is configured to obtain a packet keyword of a received data packet and a flow keyword of a data flow to which the data packet belongs according to a measurement task; updating a data sketch of the network measurement point according to the keyword, determining an associated layer according to the packet keyword, and determining an associated unit in the associated layer according to the stream keyword; mapping the packet keyword to a position of the bitmap of the associated unit, and updating the value of the position to a second value; and the centralized controller is configured to estimate the number of data packets transmitted by each data stream in the network according to a global data sketch obtained by collecting and aggregating the updated data sketches of the plurality of network measurement points so as to complete a measurement task.

The invention discloses a power grid business data feature extraction method, which belongs to the technical field of power grids, and comprises the following steps: periodically collecting power grid business data; performing primary processing and feature extraction on the power grid service data to obtain basic feature data; performing feature re-extraction and analysis on the basic feature data to obtain advanced feature data; performing anomaly detection on the basic feature data and the advanced feature data, and judging whether an anomaly detection result is abnormal or not; and when the anomaly detection result is abnormal, triggering an alarm mechanism, and adjusting a feature extraction strategy. According to the power grid business data feature extraction method, multi-level feature extraction and real-time anomaly detection are adopted, and efficient, accurate and real-time power grid business data processing can be achieved.

The invention discloses a model quantification method combined with efficient parameter transfer learning, which comprises the following steps: constructing an M1 model based on a pre-trained ViT model, and adding a low-rank adaptive matrix in an attention layer of the M1 model to obtain an M2 model added with a LoRA module; a plurality of sub-modules in the M2 model are converted into quantization modules, and a quantized M3 model is obtained; setting a weight quantization bit number and an activation quantization bit number, and calibrating the model M3 combined with the preset transfer learning mode to obtain each calibrated model; respectively training each calibrated model according to training logic by utilizing a preset training set; and evaluating the trained QATL-M3 model, the trained PTLQ-M3 model and the trained GQTL-M3 model by using a preset verification set to obtain an evaluation result. The invention relates to the field of deep learning, and solves the technical problems that an existing model quantification method is high in model calculation cost and prone to overfitting under the condition of low-bit quantification.

The invention provides a slipper sorting and counting method and system based on AI in the technical field of slipper sorting management. The method comprises the steps that S1, a large number of historical slipper images are collected to construct a data set; s2, creating a slipper identification model; s3, training a slipper identification model by using the data set, and deploying the slipper identification model to a sorting robot; s4, the industrial personal computer obtains the production instruction list and analyzes the production instruction list to obtain SKU codes and the sorting number, corresponding induction counting tables are started, all the SKU codes, the sorting number and the number of the counting tables are bound to generate a binding relation, and the binding relation is sent to a sorting robot; s5, the sorting robot collects real-time slipper images and inputs the real-time slipper images into the slipper recognition model to obtain SKU codes, and the slippers are sorted to the corresponding induction counting tables based on the SKU codes and the binding relation; and S6, counting the slippers by the sensing counting table through a photoelectric sensor, and checking a counting result through a visual counting camera. The slipper sorting and counting device has the advantage that the efficiency and quality of slipper sorting and counting are greatly improved.

The invention relates to the technical field of reinforcement learning, in particular to a self-learning method and device for a non-antagonistic task, equipment and a medium, and the method comprises the steps: 1, initializing a decision neural network of the non-antagonistic task; step 2, executing an MCTS state search algorithm based on the current decision neural network, obtaining a third-party efficiency score and recording path information; step 3, performing training optimization on a decision neural network according to the third-party efficiency score and the path information; and step 4, repeatedly executing the steps 2-3 in parallel until the decision neural network meets a trigger termination condition. The self-learning method for the non-antagonistic tasks provided by the invention is not limited to task scenes, can obtain better results for all the non-antagonistic tasks, and can perform adaptive modification according to the task scenes; the shortest solution can be quickly and gradually explored from 0 for different non-antagonistic tasks, repeated calculation is avoided to the greatest extent, and the exploration efficiency is improved.

According to the large-view-angle shooting assembly of the granary insect situation monitoring device and the insect pest image processing system, the first camera and the second camera are arranged, and the second camera is arranged in an inclined mode, so that shot pictures are partially overlapped to form continuous large-view-angle shooting pictures, full coverage of an insect pest activity area in a granary is ensured, and the insect pest situation monitoring effect is improved. According to the design, pests in a free moving state can be captured, and the problem that the same pest is repeatedly counted due to the limitation of a shooting angle can be avoided, so that the accuracy and comprehensiveness of pest monitoring are improved; through the design that a third camera vertically faces the lower portion of the barrel to shoot the interior of the trap, the trapping shooting mode of the trap is reserved, and therefore the device combines the trapping shooting mode and the luring lamp shooting mode; pests entering the trap can be monitored, and larvae living in grains can be effectively recognized.

This application relates to the field of image measurement technology and proposes an automatic detection, tracking, and positioning method for aerial targets, applicable to high-precision perception and stable locking of low-altitude and near-altitude moving targets in complex environments. This invention employs a system redundancy and fusion detection approach, utilizing both a visible light camera and a lidar sensor to construct two sets of detection sensors with different modes for target detection. The data from these different modes are then fused, enabling automatic detection, tracking, and positioning of small aerial targets at distances greater than 500 meters in the far field. This provides target positioning information for subsequent precise detection, focusing, and real-time tracking display.

The invention provides an electric vehicle modal analysis and balance control method, system and device and a storage medium, and the method comprises the steps: solving a characteristic value based on the rigidity and quality of an electric vehicle through employing a finite element method when the inherent frequency of the electric vehicle meets the design index requirements; when the inherent frequency of the electric vehicle does not meet the design index requirement, the characteristic value sensitivity of the electric vehicle is obtained based on a finite element perturbation formula of the characteristic value sensitivity, the inherent frequency of the electric vehicle is adjusted according to the characteristic value sensitivity of the electric vehicle, and a characteristic value is obtained. According to the method, the most sensitive part of the electric vehicle can be directly found out, the inherent frequency of the electric vehicle is greatly improved through small modification, and the balance of the electric vehicle is rapidly achieved; repeated calculation can be avoided, the speed is high, and the method is simple and low in cost; the stability is high, the safety is high, and the application range is wide.

The application relates to electronic data processing of engineering cost, and discloses a method for sharing and canceling nodes in the scope of engineering cost equivalence request. A processor parses a candidate variable into a normalized scope semantic tuple containing a project version, a hierarchical node range, a source field path, filtering, aggregation, transformation, a unit of measurement and precision, to request a summary to locate a sharing bucket, and compare complete tuples in the bucket, and only merge requests that are completely equivalent in semantics, have certainty and have no external side effects. A shared node maintains a node generation number and an active candidate reference set; when a candidate is canceled, only the reference is removed; only when the reference is empty, the node is to be executed and the generation numbers match, the node is canceled. A node executes a snapshot of an immutable variable once, writes a barrier to prevent distribution to canceled candidates, and invalidates affected nodes according to a change set, so that range missharing, cross-candidate cancellation interference and old snapshot reuse are avoided.

This invention discloses a method and system for predicting the structural response of historical buildings under atmospheric temperature fields. The method includes: acquiring atmospheric temperature field information, structural temperature field information, and structural response information over a past period; using the atmospheric temperature field information, structural temperature field information, and structural temperature field information and structural response information as training sets, respectively, to establish and train an atmospheric temperature prediction model for structural temperature and a structural temperature prediction model for structural response based on artificial intelligence algorithms; predicting future structural temperature field information based on future atmospheric temperature field information and historical structural temperature field information using the atmospheric temperature prediction model for structural temperature; and predicting the future structural response information of the historical building structure over a future period based on future structural temperature field information and historical structural response information using the structural temperature prediction model for structural response. This invention improves the accuracy and precision of predicting the structural response of historical buildings as atmospheric temperature changes, and has a wide range of applications.

The invention provides a device, a method and equipment for small target image detection and classification. The device for small target image detection and classification comprises an image input module, a feature extraction module and a detection or classification module. The image input module is used for processing an acquired image and inputting the image to the feature extraction module; the feature extraction module is used for extracting the features of the image through a feature extraction model to obtain the coordinate information of the small target; and the detection or classification module is used for cutting out local feature blocks on the feature map according to the coordinate information of the small targets and inputting the local feature blocks into a decision head to obtain the feature information of all the small targets in the image. The method can avoid the repeated calculation of the same features, thereby reducing the calculation amount, and improving the utilization rate of calculation resources.

The present application relates to a kind of laser radar and camera online self-calibration method and system without induction, the method includes: S1.in the process of vehicle driving, through multi-source trigger mechanism acquisition laser radar point cloud data and camera image data, simultaneously obtain vehicle motion state information and automatic driving perception system output result, the multi-source trigger mechanism includes time trigger, event trigger and cloud data mining trigger parameter degradation model;S2.the pre-processing and feature extraction of the multi-source data collected, obtain laser radar point cloud feature and camera image feature;S3.based on geometric constraint, motion consistency and perception result fusion, constructs multi-modal verification model, and carries out multi-dimensional verification to calibration parameter.The present application not only reduces the artificial maintenance cost, through automatic means maintenance vehicle state to achieve sensor calibration parameter degradation self-detection and self-calibration without induction, and improves the reliability and attendance rate of automatic driving vehicle.

This invention relates to the field of vehicle design technology and discloses a method and apparatus for determining the display area of ​​a combination instrument cluster. The method includes: acquiring first simulated data of the steering wheel, second simulated data of the user's eye ellipse, and third simulated data of the combination instrument cluster display surface; determining a central field of view region based on the first simulated data, the third simulated data, and simulated data of the central eye point; translating the central field of view region based on the simulated data of the central eye point, at least one simulated data of an extreme eye point, and the third simulated data to obtain at least one extreme field of view region corresponding to each of the at least one extreme eye point simulated data; and determining the display area of ​​the combination instrument cluster based on the central field of view region and the at least one extreme field of view region. Applying the technical solution of this invention can meet the vision requirements of users of different body types, making the design position of the combination instrument cluster universally applicable.

The present application discloses an underwater noise processing method, device, equipment and medium. The method comprises: in response to an underwater noise processing request received in real time, converting the latitude and longitude coordinates of the noise source entity and the detection entity in the target sea area contained in the underwater noise processing request into simulated geographic grid coordinates respectively; calculating the grid standard distance corresponding to the noise source entity and the detection entity according to the simulated geographic grid coordinates; and determining the underwater noise value corresponding to the grid standard distance according to the grid standard distance. The embodiment provided by the present application maps the target entity to the simulated geographic grid, simplifies the calculation of the grid standard distance between the target entities, and then uses the grid standard distance to determine the underwater noise value corresponding thereto, avoiding multiple repeated calculations, saving a lot of calculation time, and thus improving the response speed of the system.

The application provides a three-dimensional part retrieval method based on graph similarity search, relates to the field of computer graphics, and solves the technical problem that the prior art cannot fully utilize the geometry and design semantic information of CAD, and it is difficult to capture and display the topological relationship, resulting in low retrieval efficiency. The method comprises the following steps: obtaining part three-dimensional data of a computer-aided design (CAD) model, and constructing a training data set; constructing an edge-face connection graph based on the part three-dimensional data; calculating a graph edit distance (GED) matrix of all edge-face connection graphs in the training data set as a supervision signal; training a ranking model based on the GED matrix, mapping the edge-face connection graph to a hidden space, and constructing a part vector database according to the output graph-level embedding vector; the ranking model is constructed based on a graph attention network; inputting a CAD part to be queried into the trained ranking model to obtain a feature vector, performing nearest neighbor search in the part vector database, and returning a similar part result.

This invention discloses a multi-task operation ticket continuous printing system and method. The system includes: a status management module for centrally maintaining and managing the status of tasks, steps, and page numbers during the printing process; a dynamic layout calculation module for real-time measurement and pagination decisions on the operation ticket content, including a paginator unit to determine whether drawing is allowed based on the real-time remaining space on the page; and a structured drawing module for executing page drawing based on the layout calculation results, with its drawing cycle following a "measurement, decision, execution" process and synchronously updating the status. The method, based on the above system, forms a closed-loop control process through four steps: initializing the printing status, performing dynamic layout calculation, performing structured page drawing, and updating the printing status, which is executed cyclically until all tasks are completed. This invention achieves accurate dynamic content adaptation, intelligent pagination, globally continuous page numbers, and unified format output for multi-task, multi-step operation tickets, solving the problems of chaotic layout and difficult status management in existing technologies. The system architecture is clear and easy to maintain and expand.

The invention relates to a numerical simulation method, device and equipment for a convective cloud influenced by explosion, and a medium. The method comprises the following steps: simulating a convective cloud before explosion in a WRFLES mode based on downloaded sounding data; in the explosion mode, a partial explosion field is established with the explosive as the center, and explosion shock waves are simulated and calculated; the operation result of the explosion mode is checked through visual software, the coupling time of the WRFLES mode and the explosion mode is determined according to a preset distance and a pressure constraint condition, part of the explosion field is averaged and spliced in sequence at the moment where the coupling time is located, and a complete explosion field is obtained; the complete explosion field is coupled into the WRFLES mode, and the convection cloud after explosion is simulated by running the WRFLES mode again. By adopting the method, the influence of explosion on the convective cloud can be accurately simulated and described.

Embodiments of the present application relate to a kind of processing method and system for lane line identification based on image, the method comprises: receiving first image;According to preset image size, the size of first image is adjusted to obtain corresponding second image;Second image is input into preset lane line prediction model to carry out lane line instance and attribute identification processing to obtain corresponding third image.Through the end-to-end lane line prediction model provided in the present application, the identification efficiency can be improved, and system algorithm power resource is saved.

This application relates to the field of natural language processing, and more particularly to a method for ultra-long context tensor parallel incremental attention and differential compression transmission. It employs a multi-level precision sliding window attention mechanism, dynamically allocating precision levels such as fp16 / int8 / int4 based on the distance between the window and the current computation position, and the information entropy of the window content. A tensor parallel incremental attention engine enables cross-GPU / node incremental attention computation and asynchronous All-Reduce synchronization, improving computational efficiency. Semantic differential compression transmission technology is used, transmitting only the differential data based on similarity detection, significantly reducing cross-node transmission overhead. Combined with an extreme performance optimization and security control strategy, a balance between performance and security is achieved through hardware monitoring and dynamic power consumption / frequency adjustment. This invention can significantly reduce memory usage and transmission overhead in ultra-long context processing, improve computational efficiency and hardware utilization, enhance the understanding ability of large models for long texts, adapt to heterogeneous hardware environments, and is suitable for large model inference and training scenarios.

The invention discloses a method and a system for taking materials from a rectangular tray by a gantry mechanism, which can accurately calculate the storage specification of the materials in the tray and the material taking parameters of the gantry mechanism to ensure that the material taking operation is accurate each time. The method is suitable for rectangular material trays of different specifications and numbers, and only relevant parameters need to be adjusted. The system adopts an instructed operation mode, so that the workload of code editing and equipment debugging is reduced, and the code logic is clear and easy to expand and maintain. Through programming and instruction control, automatic material taking of the gantry mechanism is achieved, and the production efficiency is improved. And meanwhile, the system records and resets the accumulated material taking frequency, and subsequent process control and data analysis are facilitated. In addition, the system avoids repeated calculation of X-axis and Y-axis coordinates, the overall efficiency is improved, and subsequent processing is simplified through a data set. According to the method and system, the position of the gantry mechanism can be accurately controlled, high-precision material taking is achieved, configuration is flexible, and the method and system are suitable for various application scenes.

This invention discloses a space-ground integrated monitoring method and system for carbon emissions from transportation infrastructure, belonging to the field of carbon emission monitoring and accounting technology for transportation infrastructure. The method includes establishing a list of monitoring objects for transportation infrastructure projects, defining Scope1 as the main direct emission ledger and Scope2 and Scope3 as hierarchical supplementary ledgers; accessing and preprocessing ground-based multi-source data; identifying emission points for nodal infrastructure using upwind background, downwind enhancement, and Gaussian plume inversion, and spatially allocating linear infrastructure using key point inversion and representative segment extrapolation; verifying and filling in missing data using activity intensity data, and forming uncertainty and confidence levels. This invention realizes a closed-loop transformation of carbon emission monitoring for transportation infrastructure from single-ledger estimation to space-based remote sensing discovery, ground-based multi-source verification, activity intensity accounting, and quality classification output, improving the spatial interpretability, auditability, and engineering applicability of monitoring results.

This invention discloses a method and system for collaborative optimization of task scheduling and data caching, belonging to the fields of edge computing and mobile network technology. The method includes: constructing an edge computing environment through multiple edge servers, cloud servers, and a data cache; performing service chain-aware matching on arriving tasks within the edge computing environment to reuse computation results; scheduling tasks using a particle swarm optimization algorithm; obtaining a task scheduling scheme based on a comprehensive fitness function and task scheduling; executing the task scheduling scheme to obtain intermediate result data items; evaluating the value of the intermediate result data items using a multi-dimensional evaluation model to obtain a value evaluation result; updating the content of the data cache based on a genetic algorithm and the value evaluation result; and achieving collaborative optimization between repeated task scheduling and data caching. This invention achieves effective processing of tasks combining multiple services through a service chain-aware matching mechanism and establishes a bidirectional feedback collaborative optimization mechanism for task scheduling and data caching.

The application discloses a mobile device path planning method and an electronic device. The mobile device path planning method comprises the following steps: constructing a fitness function of mobile device path planning; initializing a particle swarm population, and a particle position is used for encoding a candidate path composed of multiple intermediate path point coordinates; performing multiple iterations of optimization, in each iteration, after each velocity and position update is completed, performing a dimension-by-dimension optimization operation on the entire population, the dimension-by-dimension optimization operation searches the path point coordinates of the candidate path encoded by all particle positions in the population dimension by dimension, and optimizes the corresponding path point coordinates in the global optimal position; after the iterations are completed, determining an optimal path based on the path encoded by the global optimal position. The application overcomes the defect that the effective local feature is lost due to simple elimination of low fitness particles in the existing PSO algorithm; meanwhile, repeated calculation of full path evaluation is avoided, and the real-time performance and high efficiency of algorithm operation are ensured while the optimization precision is effectively improved.