33results about How to "Mass balance" patented technology

The invention relates to the technical field of information management, in particular to a data query result sorting method and system fusing a large model and incremental learning. The method comprises the following steps: constructing a preliminary query model by utilizing a pre-trained large model, and generating a high-quality preliminary result; then inserting a low-correlation confrontation sample, increasing the diversity of results, and collecting user feedback; based on historical query behaviors, training a user query satisfaction model, and modeling satisfaction from multiple angles; satisfaction prediction is introduced into query optimization, strategies are dynamically adjusted for queries with low satisfaction, and incremental query results are generated and stored in a cache; fusing the preliminary result and the increment result, removing low-correlation items, and dynamically updating a query list; according to the method, the stability is kept while the data quality is improved, the dynamic optimization balance of the query result is realized, and the user experience is continuously improved.

The invention provides an adaptive retrieval enhanced large language model construction and question answering method and system, a storage medium and a program product, and the construction method comprises the steps: collecting question answering data including questions, standard answers and reference contexts; answering the question by using a large language model to obtain an initial answer; judging whether the initial answer is consistent with a standard answer or not, and constructing training data according to a judgment result; and based on the training data, performing fine tuning training on the large language model to obtain the adaptive retrieval enhanced large language model. The question-answering method comprises the steps of inputting questions; self-adaptively retrieving the enhanced large language model to answer questions, and judging whether extra knowledge needs to be acquired or not; and outputting an answer as a final answer according to a judgment result. According to the method, the balance between resource utilization and answer quality can be realized, and the problems of high resource consumption, low question and answer efficiency and poor answer accuracy are solved.

The invention discloses an automobile software defect prediction method based on dynamic feature selection and oversampling, which comprises the following steps: acquiring a data set of automobile software defects, and defining a full feature set of the data set; performing data preprocessing on the data set, and dividing the data set into a training set and a test set; dynamically evaluating the contribution of each feature in the full feature set to a decision feature through a granularity decision entropy algorithm, and reserving important features with high information gain; new samples are generated for the minority class samples by introducing an intelligent resampling algorithm of auxiliary loss and diversity loss, and the representativeness and diversity of the minority class samples are enhanced; training the defect prediction model through the training set and testing the test set; and outputting a corresponding defect prediction result based on the characteristic data of the automobile software through the trained defect prediction model. According to the method, the accuracy of automobile software defect prediction and the generalization ability of the model in new data or multi-class and mixed type data environments can be improved.

The invention discloses a credible surface defect detection method and equipment based on a Bayesian condition diffusion model. The method comprises the following steps: constructing a training data set and a test data set; constructing and training an industrial product surface defect detection model based on uncertainty perception of a conditional diffusion model, inputting a test picture into the model for detection, and respectively obtaining a defect prediction image and an uncertainty prediction image of an input result; sorting the original input image, the defect prediction result and the uncertainty prediction result to obtain a conditional guidance image during second iteration; and inputting the obtained condition guide image and the to-be-detected image into the condition diffusion model to obtain a final defect prediction image and uncertainty measurement. According to the method, uncertainty measurement is carried out on a result generated by the model based on Laplace approximation, a detection result is optimized by utilizing the uncertainty of an output result, and then iteration is carried out to improve the credibility of a final detection result.

The invention discloses an unmanned vehicle high-precision map construction method for a large outdoor scene, an unmanned vehicle at least carries a laser radar, a wheel type odometer and an inertial measurement unit, and the map construction method comprises the following steps: acquiring 3D point cloud information of the unmanned vehicle at a current position by using the laser radar; 3D point cloud preprocessing is carried out through direct filtering and a gradient negative feedback voxel downsampling method; the current pose of the unmanned vehicle is calculated based on wheel type odometer information, inertial measurement unit information and the optimized point cloud registration algorithm, and the layered space screening method for eliminating isolated laser points is an optimization method of the point cloud registration algorithm; selecting key frames according to displacement information and turning information of the unmanned vehicle, and splicing the key frames into a global point cloud map; and the unmanned vehicle performs loopback detection in the area where loopback may exist, and if loopback is detected, loopback constraint is added and graph optimization is executed. According to the invention, the performance expenditure can be reduced, the unmanned system cost is reduced, and the stability of high-precision map construction is improved.

The embodiment of the invention discloses a control channel monitoring method and device, a chip, equipment and a storage medium, and relates to the technical field of communication. Comprising the steps that downlink control information sent by network equipment is received, the downlink control information comprises configuration selection information, and the configuration selection information is used for indicating adopted antenna switching configuration; determining the number of target receiving antennas based on the antenna switching configuration indicated by the configuration selection information; and performing control channel monitoring through the antennas with the target number of receiving antennas. By adopting the scheme provided by the embodiment of the invention, the balance between the monitoring quality and the power consumption of the control channel can be realized.

The present application relates to the technical field of emergency resource scheduling, and discloses a method, system and device for scheduling the output power of a fuel cell on an emergency power supply vehicle, wherein the method comprises: extracting historical demand data, preprocessing the historical demand data, and obtaining a standardized data set; setting fuzzy control rules between the standardized data set and the output power of the fuel cell, and establishing a fuzzy rule base; constructing an energy scheduling model based on the fuzzy rule base; predicting future electricity demand based on historical demand data, and obtaining a prediction result; using the prediction result as the input of the energy scheduling model, and obtaining an energy scheduling plan; judging whether the deviation value between the real-time demand data and the corresponding prediction result is greater than a set threshold value, and if so, regenerating the energy scheduling plan according to the deviation value, thereby solving the technical problem of optimal scheduling of the output power of the fuel cell and achieving the technical effect of balancing the vehicle driving performance and the power supply quality of the external device.

The deep learning-based video multi-level adaptive compression cross-layer transmission optimization method and system of the application relates to the technical field of video transmission. By designing a compression level selection model based on deep learning, the optimal compression level is output based on the original video data and network environment parameters, and the corresponding compression strategy of the optimal compression level is obtained. A transmission strategy selection model is designed to generate a transmission strategy based on video quality parameters and network environment parameters. A global utility function is defined to calculate the global utility value when the compression strategy and the transmission strategy are assumed to be executed. An optimization objective function is designed, which inputs the compression strategy and the transmission strategy and outputs the optimized compression strategy and the transmission strategy. A condition judgment logic is set to select the final strategy based on the compression strategy and the transmission strategy before and after optimization, thereby achieving global optimization of video compression and transmission.

The invention discloses a motor sweeper operation path generation method and device, and the method comprises the steps: constructing a road connected graph based on a to-be-swept region map road boundary and a topological connection relation; coarsening the connected graph according to a preset merging strategy, and merging the fine-grained nodes into logic region nodes; constructing a final cleaning task area based on the coarsened connected graph; planning a sub-region cleaning sequence in combination with the current position of the cleaning vehicle; and constructing a global operation path. According to the invention, whole-process automation of cleaning area division and sequence planning is realized, manual intervention is not needed, and operation efficiency and reliability are improved; complex road topologies such as roundabout and the like are adapted, and sub-region division is more reasonable; a heuristic weight and an improved threshold acceptance algorithm are fused, the solving efficiency and the planning quality are balanced, and the requirements for the real-time performance and the accuracy of outdoor cleaning are met.

The invention relates to the technical field of cable hot melting control, and discloses a high-voltage cable hot melting control method, a high-voltage cable hot melting control device, high-voltage cable hot melting control equipment and a storage medium. Generating the planned hot melting operation time of each hot melting operation point in the planned execution time table, and generating a hot melting control parameter set of the hot melting operation points by taking the time limit of the hot melting operation task as a limiting condition and taking the highest hot melting operation quality as a target, in this way, the hot melting equipment is driven to execute the hot melting operation task of each hot melting operation point by adjusting the parameters of the integral time constant, the differential time constant or the integral amplitude limiting, and the hot melting operation quality is improved as much as possible while the deadline requirement of the hot melting operation task is met. Meanwhile, by continuously correcting the plan execution time table, the influence of geographical and air temperature environmental factors on the deadline is solved, and the execution rationality and scene adaptability of the hot melting operation task are improved.

The invention provides an RAG-based question and answer method and device, equipment and a medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining a query statement input by a user; performing semantic analysis, intention recognition and entity recognition on the query statement to obtain a query analysis recognition result; a mixed retrieval strategy is adopted, multiple candidate retrieval result sets associated with the query statement are retrieved from a configured retrieval database, and association scores of candidate retrieval results in different candidate retrieval result sets and the query statement are calculated; according to documents to which different candidate retrieval results belong, aggregating the plurality of candidate retrieval results to obtain a document aggregation result; determining a retrieval result generation strategy according to the query analysis identification result; and according to the configured query budget and retrieval result generation strategy, performing adaptive selection on the document aggregation result, and generating a target retrieval result corresponding to the query statement. According to the method and the device, the retrieval accuracy of the RAG question-answering system can be improved.

The invention discloses a view finding control method of head-mounted equipment, and relates to the technical field of intelligent wearable equipment. The method comprises the steps of determining a fixation point of a wearing user in a three-dimensional space based on a sight line direction and a fixation depth; projecting the fixation point to a pixel plane of a main camera of the head-mounted device to obtain a corresponding pixel coordinate; and based on the pixel coordinates and a preset shooting composition rule, adjusting a target corresponding to the fixation point to a shooting picture framed by the electronic cutting window according with the preset shooting composition rule by adjusting the position of the electronic cutting window in the shooting picture of the head-mounted device or adjusting the orientation of a main camera of the head-mounted device. By fusing the eye physiological signals, the head space posture and the camera space geometry, the three-dimensional coordinates of the fixation point of the wearing user are accurately calculated, the head-mounted equipment is intelligently controlled to accurately shoot the image with the target corresponding to the fixation point of the wearing user as the center, and the function of automatically shooting high-quality pictures is achieved.

The invention relates to the field of digital work data processing, and discloses a digital work data processing method, which comprises the following steps that: a robust watermark and a hidden prompt signal corresponding to the robust watermark are generated through an antagonism optimization algorithm, and the hidden prompt signal is used for guiding a model to hide the watermark in a process of generating an image work by a diffusion model; implanting the generated strong watermark in a middle step of a reverse diffusion stage of generating the image work by the diffusion model; in the residual reverse diffusion step after the watermark is implanted, jointly guiding the diffusion model to generate an image work which hides the strong watermark and meets semantic conditions by using text prompt and prompt signals, and reducing the influence of the watermark on the quality and visual effect of the finally generated image; when the source and ownership of the generated image digital work need to be verified, the existence of the robust watermark is verified through the inverse process of the inverse diffusion of the diffusion model. According to the method, the robust watermark can be secretly implanted into the generated content without retraining the model.

The application provides a large language model social media event analysis capability evaluation method based on a standard knowledge graph, and relates to the technical field of the large language model. First, text data related to a social media event is collected, the event-related text data is input into a pre-set and trained standard knowledge graph construction model to generate a standard knowledge graph draft; the generated standard knowledge graph draft is modified through human-machine cooperation and a conflict reservation mechanism to generate a final standard knowledge graph; based on the final standard knowledge graph, an event analysis report and its knowledge graph output by the large language model are evaluated; the scheme enhances the objective evaluation level of the anti-illusion and deep reasoning capability of the large language model in complex event analysis, and balances the efficiency and quality of constructing the standard knowledge graph in a complex social media event background.

The present invention provides a text recognition search method and system for images, which relates to the field of image search technology, and includes preprocessing an image to be searched to obtain a preprocessed image; performing text recognition on the preprocessed image using a text recognition model to obtain a text recognition result; extracting a text-image feature vector of a corresponding position in the preprocessed image according to the position of the text, and determining a text confidence content vector based on the text content and in combination with the text confidence; constructing a text-image fusion feature based on the text-image feature vector and the text confidence content vector, and constructing an image search index; performing word segmentation and semantic analysis on the search text to obtain a query fusion vector; in the image search index, determining a matching result by calculating the similarity between the query fusion vector and the text-image fusion feature, and extracting an image to be searched corresponding to the preprocessed image as an image search result according to a mapping relationship between the text-image fusion feature and the preprocessed image.

The invention discloses an Internet of Things perception-driven offshore facility group toughness maintenance scheduling and ship path collaborative optimization method, and relates to the technical field of ocean engineering operation and maintenance and intelligent optimization. The method comprises the following steps: step 1, constructing a maintenance decision hierarchical architecture supported by Internet of Things perception; 2, performing structured modeling on the operation and maintenance network of the offshore facility group; 3, performing uncertainty parameter calibration on the navigation time, the operation duration and the reachable time window based on the sensing data of the internet of things; 4, constructing a time integral optimization target based on toughness guidance; step 5, establishing a constraint system of cumulative propagation of time uncertainty along the path and feasibility guarantee; and 6, outputting a scheme by adopting a hierarchical iteration solution and local enhancement strategy. According to the method, closed-loop fusion of Internet of Things perception and decision optimization can be realized, the execution robustness of a scheduling scheme and the system toughness recovery capability are effectively improved, and decision support is provided for operation and maintenance of facilities such as an offshore wind plant and an oil and gas platform group.

The application discloses a phase change direct cooling system and belongs to the technical field of battery cooling, comprising a refrigerant pipeline, the refrigerant pipeline comprising a liquid pipeline, a gas-liquid two-phase pipeline and a gas pipeline; the final pipe diameter of the refrigerant pipeline is obtained in the following manner: a, the gas-liquid two-phase pipeline selects a standard model pipe with the minimum absolute difference value according to the calculated theoretical two-phase pipeline pipe diameter, then the minimum standard mass flow rate and the standard saturation temperature drop are obtained and scoring judgment is performed until the final pipe diameter meeting the requirements is obtained; b, the liquid pipeline selects the pipe diameter of the standard model pipe with the minimum absolute difference value as the final pipe diameter according to the calculated theoretical liquid pipeline pipe diameter; c, the gas pipeline selects the pipe diameter of the standard model pipe with the minimum absolute difference value as the final pipe diameter according to the calculated theoretical gas pipeline pipe diameter. The application realizes uniform distribution of refrigerant and improves the overall temperature uniformity of the battery under the premise of ensuring the reliability of oil return by reasonably determining the pipeline structure and the pipe diameter.

The invention belongs to the technical field of machining process monitoring and quality, and particularly relates to a scrap quality feedback control system of an aluminum profile flexible machining center, which comprises the following steps: S1, in the machining process, synchronously triggering a multi-source sensor array deployed in a scrap planing area, and acquiring multi-modal original data of scraps by taking a configurable time window as a unit; performing preprocessing and feature fusion on the data in each time window to generate a current multi-dimensional feature vector; when a new working condition is triggered, collecting continuous shaving data to generate a dynamic reference feature vector, performing intelligent matching and weighted fusion on the dynamic reference feature vector and historical reference features with similar process conditions in the expert knowledge base, forming a standard feature vector of the working condition, and updating the standard feature vector to the knowledge base; according to the method, self-adaptive optimization, abnormal early warning and safety intervention in the machining process can be achieved, so that the machining quality is improved, the service life of a tool is prolonged, equipment safety is guaranteed, and dependence on experience of operators is reduced.

The invention relates to the technical field of microscope imaging, in particular to a microscopic imaging method and device based on grating illumination, equipment and a medium. According to the application, stripe-shaped illumination light is formed through the grating device with alternately arranged light-transmitting areas and light-proof areas, the imaging camera is automatically triggered to capture the sample image after each movement of the grating device, the background picture and the highlight picture are created, and the gray value difference of corresponding pixel points of any two captured images is calculated, so that the image quality is improved. And when the gray value difference is smaller than or equal to the preset threshold value, the corresponding pixel values in the background picture and the highlight picture are updated, so that the imaging time is greatly shortened, the imaging efficiency is improved, the complex calculation processing is reduced, and the calculation complexity is reduced.

The utility model relates to the technical field of laminated slabs, and discloses a prefabricated laminated slab structure which comprises a slab body, mesh steel bars are arranged in the slab body, and truss steel bars with one ends penetrating to the outer side of the slab body are fixedly installed in the slab body. By arranging the mesh reinforcing steel bars and the truss reinforcing steel bars and reasonably arranging the staggered intervals, the overall performance is not changed while part of the reinforcing steel bars are reduced, the processing time of the laminated slab is shortened, the balance of quality, efficiency and cost is achieved, and meanwhile the connecting angle of the web member reinforcing steel bars is adjusted, so that the overall performance of the laminated slab is improved. Supporting points during installation of a large number of truss steel bars can be reduced, the construction efficiency can be improved, meanwhile, the diameter and the overall height of the steel bars are adjusted and used according to the load-bearing requirement, the overall quality is guaranteed, meanwhile, needed work during production of the laminated slab is reduced as much as possible, and dual optimization of the structural performance and the construction efficiency is achieved.

The invention belongs to the technical field of multi-UUV formation collision avoidance, and particularly relates to a careful formation collision avoidance method based on navigation UUV local coordination planning, a program, equipment and a storage medium. Firstly, parameters and task target points are initialized; the navigation UUV carries out obstacle detection and sensing and construction of a rolling probability occupied grid local map by using a multi-beam sonar installed at the bow of the navigation UUV; performing map rolling updating according to the formation scale; calculating a local optimal sub-target point based on the virtual potential field; formation constraint-oriented improved IRRT * path planning is carried out, formation constraint collision detection is completed, and a local optimal path is obtained; and performing path execution and formation coordination control so as to realize formation coordination collision avoidance and complete a task. The multi-UUV formation collision avoidance method realizes safer, more real-time and more coordinated multi-UUV formation collision avoidance, is especially suitable for an underwater unknown complex environment with limited resources and difficult communication, and has important practical application value.

This invention belongs to the field of coffee extract processing and purification technology, and discloses a method for inhibiting white spots in coffee concentrate. The method includes: first, diluting a pre-prepared coffee concentrate with deionized water to a soluble fixed content of 10-15°Brix and a natural pH of 5.0-5.8; then, taking a certain amount of the diluted coffee concentrate and preheating it to 40-55°C, while simultaneously adding hemicellulase, pectinase, neutral protease, and acid lipase, and performing enzymatic hydrolysis at a reaction temperature of 45-50°C for 2-3 hours, with a stirring speed of 50-80 rpm during the hydrolysis; then, raising the temperature of the hydrolyzed system to 90°C and maintaining it for 8-10 minutes to completely inactivate all enzyme activities; finally, concentrating the inactivated material at a concentration temperature ≤60°C using low-temperature vacuum to the target concentration and filling it to obtain the final product. This invention, based on existing equipment, adds a dilution step and an acid lipase to target the hydrolysis of triglycerides, while extending the hydrolysis time, effectively inhibiting the formation of white spots and doubling the shelf life.

The invention provides a test support device mass inertia balancing system and a mass inertia balancing method. The test support device mass inertia balancing system comprises an oil conveying pipe, two hydraulic cylinder oil pipes, an oil return pipe, a hydraulic cylinder, a heaving motion guide rail and a test model support component, the output rod is provided with a displacement sensor, and the test model supporting part is provided with a speed and acceleration sensor. A force sensor is arranged on the test model supporting part so as to detect the acting force between the test model supporting part and a tested piece mounted on the test model supporting part; and according to the acting force and the speed and acceleration detected by the speed and acceleration sensor, at least the servo valve is controlled so as to control the lifting of the test model supporting part and enable the acting force to be within a preset value, thereby balancing the mass and inertia of the test model supporting part. Therefore, the quality and inertia of the test model supporting part do not affect the model test.

The invention provides a reply generation method, device and equipment of a large language model based on retrieval enhancement, relates to the technical field of artificial intelligence, and aims to generate a high-quality reply statement. The method comprises the following steps: acquiring an initial knowledge feature vector sequence corresponding to input information; the input information at least comprises a query statement and a knowledge document retrieved according to the query statement; performing multi-stage attention interaction according to the initial knowledge feature vector sequence, and removing document knowledge feature vectors with low correlation with the query statement in different stages of attention interaction by adopting different discarding strategies according to the importance degree of the document knowledge feature vectors in the attention interaction process; obtaining a final knowledge feature vector sequence; each stage of attention interaction comprises at least one layer of attention interaction; and generating a reply statement of the query statement according to the final knowledge feature vector sequence.

The present invention relates to the technical field of energy efficiency optimization, specifically to a method for optimizing the energy efficiency of a PET bottle production line, comprising the following steps: based on the process parameters of the PET bottle production line, analyzing the energy consumption characteristics of the heating, blow molding, and cooling links, identifying key links with low energy efficiency, including heating time exceeding a threshold value and unstable blow molding pressure, and obtaining key link energy efficiency analysis results. The present invention, by analyzing the process parameters of the production line, can accurately identify the key links with low energy efficiency, and adjust the parameters in a targeted manner to optimize energy consumption, and adopt a multivariate linear regression method to evaluate the impact of different parameter settings on energy efficiency, so that the optimization measures are more scientific and accurate, and by refining the operation priority and adjusting the production parameters using a deep deterministic policy gradient model, the energy utilization rate is optimized, and the cost and product quality are balanced, so that the entire production process is more efficient and sustainable, product quality is guaranteed, and the market competitiveness of the enterprise is significantly enhanced.

The invention discloses a category knowledge graph construction and reasoning method for commodity classification. The method comprises the following steps: generating an attribute-category structured data pair through source collection commodity data; performing association mining to construct a knowledge graph and a calibration reference library; three types of abnormal samples are generated based on the atlas weight to construct a training set, and a strategy threshold is calculated; constructing a map reasoning and deep learning dual-drive mechanism, and when the classification confidence is lower than a threshold, calling associated weight correction and outputting a response coefficient; dynamically updating the association weight based on the correction result in a sliding weighting manner; according to the optimization effect reconstruction sample rule, generating a training set and outputting an efficiency index; the fusion weight is optimized based on the training set, and if the efficiency does not reach the standard, a bypass mechanism is triggered to backtrack and adjust the associated weight; and updating the reference library to the evolution strategy library by adopting reinforcement learning, and dynamically selecting an optimal self-calibration path to obtain a target knowledge graph. According to the method, closed-loop self-evolution of knowledge graph construction and reasoning is realized, and the commodity classification precision and the anomaly recognition capability are remarkably improved.

The present invention relates to a shared electric vehicle operation optimization method and system based on the fusion of column generation and ant colony algorithm, and the method includes: S1: obtaining shared electric vehicle reservation orders and forecast information; S2: determining the objectives and constraints of shared electric vehicle fleet configuration and path optimization according to the shared electric vehicle reservation orders and forecast information; S3: establishing a mixed integer linear programming model according to the objectives and constraints to maximize the total profit of the shared electric vehicle system; S4: using D-W decomposition to decompose and reconstruct the mixed integer linear programming model to obtain a reconstructed planning model; S5: using a hybrid algorithm composed of column generation and ant colony algorithm to solve the reconstructed planning model, and calculate the fleet composition and path planning scheme. The present invention solves the operation optimization problem in the shared electric vehicle system, improves the profitability of the operating enterprise, and has high decision reliability and strong calculation stability.

The present invention discloses a text image repair method and device thereof, and belongs to the technical field of image text recognition. The method comprises a distribution difference parameter acquisition step, a reconstruction loss parameter acquisition step, and an OCR loss parameter acquisition step; the distribution difference parameter acquisition step, the reconstruction loss parameter acquisition step, and the OCR loss parameter acquisition step are repeated until the total loss function constructed by the distribution difference parameter, the reconstruction loss parameter, and the OCR loss parameter converges; the text image to be repaired is input into the image repair model to reconstruct the text image, and a repaired text image is obtained. The present invention improves the recognition accuracy of OCR by repairing the image. In addition, the repaired image can be recognized by humans, and the result has stronger visualization and interpretability.

The present disclosure relates to the technical field of computer, and particularly relates to a query request response method and device, equipment and storage medium, which are applied to query response of enterprise internal data resources, comprising: obtaining a query request input by a user, performing intent understanding and context analysis on the query request, and generating an intent analysis result containing query semantics, task type and preliminary sub-task disassembly result; generating an agent scheduling instruction based on the intent analysis result; the agent scheduling instruction is used to call at least one sub-agent in a data query agent, a root cause analysis agent or an intelligent decision agent; according to the agent scheduling instruction, a corresponding data query, root cause analysis or decision suggestion generation operation is performed by the scheduled sub-agent to obtain a sub-agent execution result; whether the sub-agent needs to be continuously scheduled is evaluated according to the sub-agent execution result, at least one agent in the data query agent, the root cause analysis agent or the intelligent decision agent is called in a loop when the sub-agent needs to be continuously scheduled, until a task termination condition is met, and an intelligent analysis report is obtained; and the intelligent analysis report is displayed.

The invention relates to the technical field of map slicing, and discloses a freehand map hierarchical slicing method for agriculture tourism science and technology service, which comprises the following steps: acquiring a freehand map of a target area, determining a minimum enclosing rectangle covering the freehand map, and acquiring geographic coordinates of vertexes of the minimum enclosing rectangle; performing feature point detection on the hand-drawn map to obtain a feature point distribution map; determining a scaling hierarchy, slicing the hand-drawn map, and determining the row and column numbers of each tile; calculating a first coordinate of each tile, and calculating a second coordinate based on the geographic coordinate; judging whether the range covered by the first coordinate is within the range covered by the second coordinate, if so, calculating the pixel coordinate of the tile corresponding to the first coordinate in the hand-drawn map, and otherwise, setting the tile corresponding to the first coordinate as blank; querying feature points covered by the pixel coordinates in the feature distribution map, and resampling tiles corresponding to the pixel coordinates based on the feature points to obtain new tiles corresponding to each tile; and storing all the new tiles according to the scaling hierarchy and the row and column number corresponding to each tile. According to the invention, the quality of the hand-drawn map slice can be effectively improved; and meanwhile, the data storage pressure, the network transmission pressure and the front-end rendering pressure are reduced for the tile service application.