18results about How to "Realize dynamic generation" patented technology

The invention discloses a cross-domain parallel strategy generation technology based on constraint guidance and redundancy calculation, and aims to research how a cross-domain distributed calculation framework dynamically generates a cloud edge end cross-domain training parallel strategy. The method comprises the following steps: firstly, researching a runtime parallel plan regeneration mechanism, fitting cloud edge end environment change to reduce a search space by constructing a constraint model and a performance model, and applying an optimized graph matching algorithm to obtain final runtime matching; and secondly, researching a parallel execution mechanism of multiple parallel plans during operation, determining a specific switching plan by constructing a video memory constraint and parallel coverage model, obtaining a new Micro-Batch parallel plan, and completing switching and execution of the related parallel plans. Through the mechanism, the parallel strategy generation is dynamically and quickly generated according to the dynamic change of the environment, the strategy search time is shortened, the switching influence is reduced, and the dynamic nature of the cloud edge end computing environment is effectively dealt with.

The invention belongs to the technical field of artificial intelligence, and particularly relates to a power abnormal data risk grading and handling method and system, and the method comprises the following specific steps: collecting data from a multi-source channel, and carrying out the preprocessing and feature engineering of the collected data; building a classifier based on a plurality of decision tree models, and selecting information gains as evaluation indexes to construct decision trees; finely tuning the locally deployed GLM open source large model; disposing suggested retrieval enhanced questions and answers; according to the method, various information sources and the processing knowledge base are combined in real time, more comprehensive and accurate processing suggestions are provided, and real-time updating of the processing knowledge base and dynamic generation of the processing suggestions are achieved through a retrieval enhancement scheme.

The invention relates to a quick form building method. The method comprises the following steps: defining a quick search custom component based on a JSON (JavaScript Object Notation) format; defining a reusable UI module based on a Vue framework, wherein the UI module receives the attribute value and triggers a custom event; the front-end component introduces a quick search custom component; defining a to-be-searched form object and transmitting the to-be-searched form object as an attribute parameter value to the quick search custom component; configuring attribute parameter values for displaying the searched data; configuring an attribute parameter value for quickly retrieving and displaying the configuration item; custom content is inserted into the rapid search custom component through a custom slot content; and using a quick search custom component in the template of the Vue component, and binding an attribute parameter value and an event monitor. According to the method, the dynamic form search is realized based on the JSON configuration, and the fields, types and verification rules of the form are defined through the unified JSON configuration file, so that the dynamic generation and management of the form interaction interface are realized.

The present application relates to a kind of forest leaf instance segmentation method and system based on deep learning, method includes: obtaining vegetation image, vegetation image is input into leaf instance segmentation model, obtains leaf instance segmentation prediction result;Leaf instance segmentation model is trained using training set;Training set includes: vegetation original image;Feature extraction and enhancement are carried out using backbone module in leaf instance segmentation model, and adaptive spatial fusion mechanism in progressive feature pyramid network is integrated to dynamically adjust feature weight, generate dynamic fusion feature;Through the dynamic asymmetric spatial perception mechanism built-in in dynamic anomaly regression head module, the corresponding multi-source deformation feature layer of dynamic fusion feature is obtained, and the feature fusion strategy of top-down cascaded decoding module is used to optimize multi-scale feature, obtain multi-source fusion feature layer, further using multi-source fusion feature layer, generate leaf instance segmentation prediction result.The present application solves the problems of data scarcity, poor adaptability and low efficiency.

The embodiment of the invention provides a task scheduling method and device, electronic equipment and a storage medium. The method comprises the steps that conventional inspection tasks are formulated according to the growth stage of a target crop to obtain a task queue, and the task queue comprises the conventional inspection tasks; issuing the conventional inspection task to an inspection robot to be executed by the inspection robot; based on the execution result of the conventional inspection task, adjusting the inspection tasks in the task queue, and sorting the inspection tasks in the adjusted task queue according to a target sorting strategy to obtain a task execution list; and issuing the inspection task in the task execution list to the inspection robot so as to be executed by the inspection robot. According to the scheme, the problem that a traditional static scheduling scheme is disjointed with the crop growth stage and the execution result of the inspection task can be solved, dynamic generation and adjustment of the inspection task are achieved through a two-dimensional task driving algorithm based on the growth stage and the execution result, and therefore the intelligent and extensible inspection task management capacity is achieved.

A method for planning an autonomous landing safe trajectory suitable for UAV inspection is proposed. After the UAV takes off, it acquires the current pose of the UAV, the point cloud map information of the surrounding environment, and the spatial orientation of the target point. By establishing the spatial transformation relationship between the camera coordinate system and the body coordinate system, the orientation vector relative to the ground target is calculated, and the target point is generated within the radar-sensible area. Based on the generated target point, an optimization function that minimizes the fourth derivative is constructed based on multiple high-order polynomials to generate a smooth trajectory with continuous position, velocity, and acceleration. The initial trajectory is transformed into a B-spline curve, and local trajectory segments with collision risk are optimized. A segmented penalty mechanism is designed based on the boundary threshold to guide the trajectory away from the boundary area. The time span is adjusted by calculating the trajectory derivative over-limit ratio, and a trajectory similarity cost function is introduced to ensure that the final optimized trajectory maintains the original geometric structure, thereby achieving trajectory control optimization and safe landing of the UAV.

This application discloses a knowledge- and data-driven elastic positioning method, system, and storage medium. The method includes: generating a knowledge graph based on a knowledge base in the positioning domain; the knowledge base includes multiple pieces of first knowledge information, including entity pairs in the positioning domain, relationships between entity pairs, and a first confidence level of the relationships; nodes in the knowledge graph represent entities, edges in the knowledge graph represent relationships between different entities, and the attributes of the edges include the first confidence level of the relationship represented by the edge; inferring a positioning model applicable to the target object based on currently available information sources, observation data of the available information sources for the target object, and the type of observation data; the positioning model includes positioning parameters, observation equations, and state transition equations; and locating the target object based on the observation data and the positioning model to obtain the positioning result of the target object.

The present invention relates to the field of brush cleaning control technology, and in particular to a flexible and controllable brush system and method for treating the surface of a spacecraft. The method comprises: obtaining spacecraft surface data, constructing point cloud data, and generating three-dimensional point cloud data; preprocessing the three-dimensional point cloud data, converting the preprocessed three-dimensional point cloud data into a three-dimensional model, identifying the cleaning area and extracting regional feature information; based on the three-dimensional model and regional feature information, using a dynamic geometric feedback iterative path planning algorithm to calculate the cleaning path with the highest coverage rate, generating a preliminary cleaning path, using a nonlinear multi-feedback flexible brush optimization algorithm to dynamically adjust and optimize the cleaning configuration parameters of the brush, and obtaining the optimized cleaning configuration parameters; and treating the surface of the spacecraft according to the preliminary cleaning path and the optimized configuration parameters. The method solves the technical problems of not being able to accurately adapt to and fully cover the surface of the spacecraft and having poor dynamic control capabilities when cleaning the surface of the spacecraft.

The invention discloses an automatic operation and maintenance method and device, equipment, a storage medium and a program product, and relates to the technical field of data processing.The automatic operation and maintenance method comprises the steps that when an operation and maintenance trigger event is detected, data is obtained from multiple data sources, and multi-dimensional data is obtained; generating a strategy decision context object according to the multi-dimensional data; determining a resource condition rule matched with the current resource state according to the current resource state and a current resource condition rule set in a rule base; determining a test condition rule matched with the first test information of the current code according to the first test information of the current code and a current test condition rule set in a rule base; determining a target operation and maintenance strategy according to a resource condition rule matched with the current resource state and a test condition rule matched with the first test information of the current code; and executing the target operation and maintenance strategy. According to the method, the accuracy of the target operation and maintenance strategy is improved, and then the software delivery efficiency and the stability of a service system are improved.

The invention provides a display device and method and interaction equipment, which can be applied to the technical fields of display, human-computer interaction and computer vision processing. The display device comprises a processor configured to determine a writing point or a watermark point based on a plurality of writing signals collected at the same time, the plurality of writing signals are collected by a plurality of signal collectors of an input device in the process that a target object controls the input device to execute writing operation on a non-touch-sensitive surface, and the watermark point is a watermark point or a watermark point. The plurality of signal collectors have at least two types; and a display panel configured to display the writing point or the watermark point from the processor.

The invention belongs to the technical field of software upgrading, and discloses a software as a service (SaaS) application component upgrading method and system. The method comprises the following steps: in an SaaS platform, constructing a data analysis model, a resource scheduling model and an upgrading strategy generation model; collecting real-time monitoring data, and performing standardization processing; performing data analysis by using a data analysis model according to standard real-time monitoring data; performing resource scheduling by using a resource scheduling model according to a real-time data analysis result; generating an upgrading strategy by using an upgrading strategy generation model according to the real-time data analysis result and the real-time resource scheduling scheme; and performing offline upgrading on the corresponding SaaS application component according to the real-time upgrading strategy. According to the method, the problems of high cost, low practicability, lack of flexibility and adaptivity, low resource utilization rate and poor strategy dynamism in the prior art are solved.

The invention relates to the technical field of retail inspection, and particularly discloses a retail store inspection method and system based on a configurable questionnaire, and the method comprises the steps: collecting the multi-dimensional state data of a target store in real time, and enabling the multi-dimensional state data to comprise static attribute data, dynamic business data flow and environment context data; performing standardization processing on the multi-dimensional state data to generate a target store state vector; based on a pre-trained optimal strategy curved surface model, the target store state vector is mapped to an inspection strategy vector space, a recommended inspection strategy vector is obtained through calculation, and the optimal strategy curved surface model represents a nonlinear mapping relation between the store state vector space and the inspection strategy vector space; according to the recommended inspection strategy vector, dynamically assembling and generating an inspection questionnaire from the modular questionnaire component library, and generating a corresponding inspection task execution scheme; and dynamic strategy optimization based on state perception for retail store inspection is realized.

The invention provides a patient body and mind multi-dimensional intelligent monitoring system and method based on multi-modal sensing, and relates to the technical field of medical monitoring. A health data acquisition module of the system acquires a limb movement track, a body pressure distribution time sequence diagram and basic physiological parameters of a patient in real time, and when a pressure center continuously moves outwards and the acceleration exceeds a dynamic threshold value, a high-frequency sampling mode is activated. And the health risk analysis module quantifies the corresponding relationship between the body position change and the physiological parameters through multi-modal correlation strength calculation, and outputs a bed falling risk grade signal. And the interactive health assessment module triggers voice interaction when the physiological parameters are abnormal, and combines subjective and objective data to generate a composite health index. And the clinical decision support module generates graded nursing suggestions according to the risk levels and the health indexes, and intelligently deploys the response of the medical care terminal. The physical and mental state of the patient is comprehensively monitored, and the medical monitoring early warning accuracy and clinical response efficiency of the patient are remarkably improved.

The invention provides a household broadband installation and maintenance service work order quality inspection method based on an intelligent agent. In the method, firstly, a quality inspection request is obtained, a quality inspection agent is matched from an agent pool according to the quality inspection request, the quality inspection agent is obtained based on a quality inspection rule and an agent meta-model, and the quality inspection agent comprises quality inspection capability; secondly, the sound recording file is translated into text data, and the sound recording file is used for recording a service call corresponding to the home broadband installation and maintenance service work order; and then, matching a cue word template according to the quality inspection capability of the quality inspection agent, and generating a cue word based on the cue word template and a quality inspection rule corresponding to the quality inspection agent. And finally, reasoning the text data and the cue words through the large model to obtain a quality inspection result. Through the method, the change of the quality inspection service can be quickly responded, and the quality inspection accuracy and the service fitting degree are improved.

The invention discloses a disasters evacuation route planning method and system for vulnerable groups. The route planning method comprises the following steps: acquiring behavior characteristics and movement tracks of the vulnerable groups in a complex dynamic evacuation scene; constructing a behavior influence factor knowledge graph based on the behavior characteristics and the moving track; on the basis of the knowledge graph, predicting the behaviors of the vulnerable population to obtain a prediction result; and combining the prediction result with a macroscopic evacuation strategy to construct an optimal evacuation route. The system integrates the functions of behavior analysis, route planning, simulation evaluation and the like, provides decision support for rescue workers through man-machine interaction and visualization technologies, and effectively improves the safety and efficiency of disaster evacuation of vulnerable people.

The invention provides an intelligent script generation and touring guide system based on urban travel characteristics, and the system comprises a database building module which is used for building a database containing scenic spot data; the data generation module is used for generating world view and principal angle setting; the scenic spot script generation module is used for generating a scenic spot script; the fluency evaluation module is used for evaluating the fluency of the script; the path selection module is used for path selection based on an NSGA-II algorithm; and the script deduction module is used for performing LLM script deduction. The method is not only based on the geographic distance, but also combines the cultural background and tourist interests to realize multi-objective path optimization. The tourists can visit the most relevant scenic spots with the most cultural value in the shortest time.

The invention discloses a data interaction method and device based on a form page, and relates to the technical field of computers. A specific embodiment of the method comprises the following steps: in response to a received form page access request, obtaining a form page identifier from the access request; at least one form element associated with the form page is obtained according to the form page identifier, and the form elements are obtained by abstracting form data input items and have element types; for each form element, obtaining a page rendering template corresponding to the form element according to the element type of the form element; and rendering each form element according to the page rendering template corresponding to each form element to generate a form page, and performing data interaction with a user through the generated form page. According to the embodiment, the form page is dynamically generated and updated, the problem that the form page cannot be dynamically modified is solved, meanwhile, the stability and continuity of database service are guaranteed, and the data security is improved.

The invention relates to the technical field of data processing, in particular to a header generation method and system. According to the method, the header knowledge graph is successfully constructed by defining the header entities, the relationship between the header entities and the attributes of the header entities, the total row number and the total column number of the header are calculated by utilizing the constructed header knowledge graph, the header array is generated, each header entity in the header entity knowledge graph is traversed, and the header entity knowledge graph is constructed. Calculating a row number and a column number of an initial cell of a content coverage area of each header entity, generating the content, the position and the attribute of the header entity into a JSON format in a key-value pair form, storing the content, the position and the attribute in a position corresponding to the header array, traversing non-empty elements in the header array storing the content and the attribute information of the header entity, and storing the content, the position and the attribute information of the header entity; generating a header based on the content, position and attribute of the header entity; according to the method, dynamic generation and flexible management of the report header are realized by utilizing the characteristics of flexibility, multi-level and expandability of the knowledge graph and efficient information integration, association and retrieval capabilities.