21results about How to "Avoid over-occupancy" patented technology

The invention discloses a large model cloud computing dynamic updating method based on elastic adaptive incremental learning, which comprises the following steps: S1, in a cloud computing platform, acquiring a current version of a large model and a new data stream to be processed; s2, local parameter updating is conducted on the large model according to the new data, and parameters or model layers needing to be updated are dynamically selected according to the characteristics of the data in the updating process; s3, introducing an adaptive learning strategy, and monitoring the change condition of data distribution in real time; s4, in a large model updating process, splitting a large model updating task to a plurality of computing nodes for parallel execution through a multi-node collaborative scheduling mechanism of the cloud; s5, after the large model is updated, the system verifies the performance of the large model; and S6, in the process of updating and deploying the large model, the system dynamically recycles computing resources which are not used any more. According to the method, idle computing resources can be fully utilized in a large-scale cloud computing environment, so that computing tasks are distributed in a balanced manner, and dynamic adjustment of task loads is ensured.

The invention relates to the technical field of vehicles, and discloses a vehicle-mounted process resource management method and system, electronic equipment, a storage medium and a vehicle, and the method comprises the steps: creating at least one configuration file which is used for defining a resource limitation parameter of each vehicle-mounted application process; calling a Cgroup interface, creating a corresponding Cgroup for each process, and setting a corresponding resource limit; in response to initialization of the vehicle-mounted application process, dynamically creating and managing a thread pool according to the resource limiting parameters; the configuration thread attribute setting interface is used for dynamically configuring the scheduling strategy, the priority and the affinity of the thread by the vehicle-mounted application process according to task requirements; configuring a thread monitoring mechanism for monitoring the execution time and state of each thread and recovering the corresponding thread; and configuring a cross-platform API (Application Program Interface) for uniformly calling interfaces related to process resource management and thread management. According to the method, excessive occupation and waste of resources are avoided, the thread execution efficiency and the response speed are improved, and the development difficulty and the maintenance cost are reduced.

The invention discloses a database AI question answering system and method based on a large language model and retrieval enhancement, and relates to the technical field of database retrieval. The system comprises a front-end interaction module, a distributed cache management module, a task scheduling center, a large language model processing sub-module, a retrieval enhancement sub-module, an answer optimization sub-module and a database management sub-module. All the sub-modules are mutually independent and work cooperatively, through an efficient task scheduling mechanism, the system can monitor the load and resource states of all the sub-modules in real time, accurate allocation of tasks is conducted according to an intelligent algorithm, in a large-scale concurrent request scene, single-point faults and resource waste are effectively avoided, stable operation of the system is ensured, and meanwhile, the system can be widely applied to large-scale concurrent requests. The distributed cache management module further optimizes the storage and access of the data, reduces repeated data query and transmission, and improves the overall performance.

The invention discloses an efficient webpage dynamic rendering method, system and device and a medium. The method comprises the following steps that interactive behaviors such as rolling and clicking when a user browses a webpage are monitored; determining a current user visible area according to the interaction behavior of the user; dynamically rendering the webpage elements in the user visible area; in the dynamic rendering process, the change of the user interaction behavior is detected in real time, and a rendering area is adjusted according to the change; the rendering efficiency and performance can be improved by dynamically rendering the elements of the user visible area part in the webpage, meanwhile, the page loading fluency and the user operation response speed are guaranteed, the rendering efficiency and performance are effectively improved, excessive occupation of performance resources is avoided, and the user experience is improved. The event operation of the user can be immediately responded in the page rendering process, and the use experience of the user and the smoothness of page loading are improved.

The invention relates to the technical field of classroom intelligent terminals, in particular to a classroom intelligent terminal integrating a micro host, network access and audio acquisition. According to the technical scheme, the classroom intelligent terminal integrating a micro host, network access and audio acquisition is provided. Comprising a main control unit, an acoustic array device, a network interface device, a deep learning audio processing unit, a multi-mode interaction device, an adaptive network technology component, an edge computing unit, a virtual reality and augmented reality device, a block chain module and a voice synthesis and emotion recognition module. The system provides more advanced, intelligent and interactive rich education experience, is expected to improve the learning effect and teaching efficiency, has potential revolutionary influence in the education field, and provides a more flexible, personalized and efficient learning environment.

The invention relates to the field of storage, and discloses a storage access method and a storage system. The method comprises the following steps: acquiring a plurality of operation requests sent by a plurality of hosts, transmitting the operation requests to a storage array through multi-level nodes, and executing the operation requests; wherein for one sub-node, the scheduling sequence of the current sub-node when the current sub-node sends, receives and / or responds to the multiple operation requests is determined according to the priority of each operation request; and re-determining the priority of the operation request based on the current child node. According to the multi-level scheduling method and device, multi-level scheduling is carried out on the multi-level transmitted operation requests at different nodes according to different level dimensions, the multiple multi-level transmitted operation requests can be effectively processed, excessive occupation of resources is avoided, the priority of each operation request can be transmitted among different nodes, and the efficiency of the multi-level transmitted operation requests is improved. The scheduling efficiency of the scheduling operation request of each node is improved, and the scheduling consumed time is reduced.

The present application relates to a virtual scene rendering method, device, electronic device and storage medium. The method counts the meshes of the models in all bounding volumes of a model cluster, deduplicates the primitives referenced by the meshes based on the material data corresponding to the meshes, and finally generates a unified material data and texture data for all bounding volumes in the model cluster. Since a large number of duplicate primitives are removed from the texture data, the number of primitives contained in the final generated texture data is reduced, thereby avoiding excessive occupation of memory resources and saving memory.

The embodiment of the present invention provides a method, device, electronic device and readable medium for adjusting the switch state of a network card function. The method includes: by determining the network connection speed and the number of data packets received by the network card, the peak number of data packets that the network card can receive can be determined according to the network connection speed, and the switch state of the network card receiving data packet function and / or the network card interruption function can be further adjusted according to the number of data packets received by the network card and the peak number. The peak number of data packets that the network card can receive is dynamically adjusted according to the network status, and then the number of data packets received by the network card and the peak number are used to dynamically adjust the switch state of the network card receiving data packet function and the network card interruption function, so as to control the reception amount of broadcast packets and multicast packets and the number of interruption reports, avoid excessive occupation of CPU resources, enable the CPU to process normal business, and ensure CPU work efficiency and stability.

The utility model relates to the technical field of electrical detection, in particular to an electrical circuit detection device with a temperature and electromagnetic detection function, which comprises a main controller, an ASK signal receiving interface, a serial peripheral interface, an ADC signal receiving interface and a USART serial port are arranged in the main controller, and the ASK signal receiving interface receives door magnetic signals of a wireless door magnetic detector. The serial peripheral interface receives a residual current magnitude signal of the residual current sensor and a single-phase current effective value signal of the current mutual inductance sensor. The ADC signal receiving interface receives a crystal oscillator oscillation frequency signal of the temperature sensor. And the control signal is input into the NB-IoT communication module through a USART serial port. And the NB-IoT communication module uploads the packaged data of the signal and the control signal to the Internet of Things cloud platform. The electric power system is accurately monitored, early warning is given out in time, and higher performance is achieved.

The invention relates to the technical field of cloud computing resource management, in particular to a model training resource isolation and management system and method in a multi-tenant environment, and the system comprises a data collection module, a resource adjustment module, a trend prediction module, an evaluation adjustment module and an execution module. According to the method, the resource use condition of each named isolation tenant in the model training process is monitored in real time, the preset evaluation standard and trend prediction are combined, the resource quota is dynamically adjusted, reasonable allocation and efficient utilization of the resources are ensured, and through comprehensive evaluation of the priority and the synchronization degree, the service life of the named isolation tenants is prolonged. The system can correct the resource allocation strategy in time when resource use fluctuates, resource waste or excessive occupation is avoided, the resource management efficiency of the whole system is improved, and the dynamic allocation mechanism not only improves the resource utilization rate, but also can adapt to the demand changes of different tenants, and improves the user experience. The problems of low resource utilization efficiency and unstable system performance caused by dependence on static configuration are effectively solved.

The invention provides a data development debugging method and device, equipment, a storage medium and a program product. Relates to the technical field of data development. The method comprises the following steps: in response to a received debugging request initiated by a user through a data development platform, allocating computing resources based on the debugging request and starting an interactive debugging environment instance; the debugging request comprises a Python operator code to be debugged; executing the Python operator code to be debugged in the interactive debugging environment instance, and displaying a code execution result and a variable value; and backfilling the debugged Python operator code in the interactive debugging environment instance into a static code editor of the data development platform. According to the method provided by the invention, the code execution state can be checked in real time, the code writing and debugging process is deeply integrated, environment switching and manual operation are avoided, and human errors are reduced; through a reasonable resource management and scheduling mechanism, it is ensured that computing resources can be fully utilized in the starting and running process of the interactive debugging environment instance, and resource waste is avoided.

This invention relates to a load-balancing-based constellation communication-navigation coupled task scheduling method, belonging to the field of aerospace mission planning technology. It primarily improves resource allocation efficiency under equipment-level constraints and addresses the challenge of ensuring communication task completion while meeting the geometric accuracy requirements of navigation services. The invention uses satellite ephemeris data and mission requirement parameters as input to construct a multi-objective mixed integer linear programming (MILP) model. The optimal binary allocation variables are then obtained through a MILP solver, and the service satellites and service time slots for each mission are extracted to generate a scheduling table. This invention avoids the problem of navigation mission failure caused by communication tasks occupying multi-satellite convergence windows in separate planning. Simulations show that the overall mission completion rate is improved by an average of 12.3 percentage points, and the load variation coefficient is reduced by 66%–88%.

The utility model provides a spliced door butt jointer, which comprises a first guide rail installed on the top of a shutter leaf, a second guide rail fixed on the top of a door frame, a movable butt joint seat installed at the end of the first guide rail, a fixed butt joint seat installed at the end of the second guide rail, the movable butt joint seat and the fixed butt joint seat being mutually butted when the shutter leaf is closed, so that the first guide rail and the second guide rail are in communication, a locking block arranged in the movable butt joint seat, the locking block being linked to a linkage seat arranged on the door frame through a linkage mechanism, the linkage seat driving the linkage mechanism to make the locking block disengage from the first guide rail when the shutter leaf is closed, the linkage seat being separated from the linkage mechanism when the shutter leaf is opened, the locking block being inserted into the first guide rail to limit the sliding of a pulley set. The utility model completely solves the problem that the traditional flat door can only be opened by half, limits the entry and exit of articles, and provides a large-size operation space and an unobstructed view for the user.

The invention relates to the field of data processing, in particular to an adaptive computing power segmentation scheduling method for large model reasoning tasks, which comprises the following steps of: synchronously acquiring characteristics of common dimensions of a plurality of tasks to be reasoned and computing power cluster equipment; clustering the tasks, setting a plurality of batch size schemes according to the tasks, and dividing task batches; for each scheme, the tasks are matched with equipment according to the batch sequence, the matching degree of batches is calculated, and the weights are determined by the ratio of the feature weights to the same-dimension features of the equipment tasks; after matching is completed, counting the equipment computing power, the memory utilization rate and the balance degree, and obtaining a scheme evaluation index in combination with the matching degree; and comparing the scheme evaluation indexes under all the processing modes, and selecting the highest scheme and the matching result thereof as a final scheme. According to the method, the multi-dimensional features and multiple processing modes are comprehensively considered, the optimal scheme is selected for task and equipment matching, the task requirement can be better met, the processing time of the task on the equipment is shortened, and therefore service delay is reduced.

The application discloses a kind of mobile multi-view picture data acquisition methods based on graph neural network, comprising the following steps: S1, collect multi-view picture data and carry out repair preprocessing, construct preliminary image dataset;S2, the image is divided into several image blocks, and input in the improved ViT model and carry out preliminary processing, according to the processing weight of each image block dynamic allocation according to task priority;S3, the arrangement order of image block is optimized in combination with bird swarm algorithm, maximizes labeling accuracy and minimizes the consumption of computing resources;S4, the optimized image block arrangement order is input into the improved ViT model again, and the labeled image block is obtained;S5, the relationship between the labeled image block is modeled using graph neural network, and the image block relationship graph is generated;S6, according to the labeled image block and image block relationship graph, generate final image dataset.The application combines graph neural network, improved ViT model and bird swarm algorithm, etc., and optimizes multi-view image data acquisition and labeling.

The application provides a server, a display device and a live source import method. First, a mapping table based on geographical position and network operators is used for accurate query to filter available live sources in line with the local network environment of a user, avoiding misjudgment across regions. For unknown or unplayable live sources, the server generates a distributed detection task and distributes it to a second display device in an idle state in the same region to detect availability through a lightweight HTTP request, which not only avoids the IP blocking risk of centralized scanning, but also improves detection accuracy by using the local network environment. The available live sources detected are mapped to unique identification codes after semantic normalization processing and are dynamically updated to the mapping table to form a continuously optimized data closed loop. Through resource optimization strategies, the entire process minimizes interference with user devices, and finally provides the user with an accurate and de-duplicated list of available sources, improving determination efficiency and user experience.

The application provides an accounting data management method, system, device and medium based on edge computing, which extracts a structured feature vector of newly added accounting voucher data; extracts a similar voucher set similar to the newly added accounting voucher data from a high-frequency subject area and a sparse subject area in historical accounting voucher data according to a data partition granularity; determines the distance between the structured feature vector of the newly added accounting voucher data and the feature vector of each accounting voucher in the similar voucher set, and then generates a risk confidence value of the newly added accounting voucher data based on the occurrence frequency of similar accounting vouchers and the feature vector distance; and when the risk confidence value exceeds a set threshold, updates the compliance verification result and the structured feature vector corresponding to the newly added accounting voucher data to an accounting voucher database. The technical scheme provided by the application can overcome the interference caused by the natural long-tail distribution characteristics of accounting subjects on similar voucher retrieval in accounting data management.

The invention provides a smart park resource computing power scheduling optimization method and device, equipment and a medium, and belongs to the technical field of resource scheduling. Determining a target computing power node group from the plurality of computing power node groups based on the computing power demand and the task priority of the computing power request task and the predicted load data of each computing power node; screening an optimal execution node based on the predicted load data, the task adaptation degree and the node health degree of each computing power node in the target computing power node group, and distributing the computing power request task to the optimal execution node for execution; and in response to detection that the energy consumption of the optimal execution node exceeds a preset energy consumption threshold value of the node type corresponding to the optimal execution node, or the task execution delay of the computing power request task is greater than a preset delay duration, migrating the computing power request task based on the task priority. The resource utilization rate can be effectively improved, and idle or excessive occupation of resources is avoided.

This invention belongs to the field of rail transit technology, and particularly relates to a train turnaround control method. It includes: after a first train obtains turnaround resources according to its travel path, it controls the switch to the reverse position; after controlling the switch to the reverse position, it begins to turn onto the route, clearing the turnaround resources in the area its tail has passed through, gradually releasing the cleared turnaround resources until the first train has completely cleared the movable area in front of the switch; after the first train has completely cleared the movable area in front of the switch, it controls the switch to the fixed position. This invention, by further refining the switch area, fully utilizes and controls the switch resources available to the train, thereby improving train turnaround efficiency.

The invention relates to a video memory management and control method and system based on a virtual video card. The method is applied to a host machine in a Hypervisor environment, and comprises the following steps: receiving a video memory allocation request of each process sent by a client, and determining the previous video memory use total amount of each process; distributing a corresponding video memory for each process according to a preset process priority; for each process, according to the total usage amount of the first type of video memories used by the virtual graphics card, the total usage amount of the second type of video memories and the video memory allocation request of the process, allocating the corresponding usage amount of the first type of video memories or the usage amount of the second type of video memories to the process, and generating a corresponding allocation result; for each process, determining the current total video memory usage amount of the process according to the distribution result and the previous total video memory usage amount of the process; and for each process, sending the allocation result and the current total use amount of the video memory to the corresponding client. According to the method, the system stability of the Hypervisor virtualization scheme is improved.

The invention discloses a hidden type single-rod clothes airing machine convenient to maintain. The hidden type single-rod clothes airing machine comprises a rack shell, a main machine assembly and an airing rod. The rack shell comprises a shell top plate and two shell side walls; the upper ends of the two shell side walls are connected to the two sides of the shell top plate. The main machine assembly is installed between the two shell side walls. The main machine assembly comprises a machine side part, a machine top wall, a power assembly, a control box and two steel wire rope bogies; a groove for accommodating the airing rod is formed by the machine side part and the machine top wall; the power assembly, the control box and the steel wire rope bogie are installed on the back face of the machine top wall. A plurality of hoisting brackets dispersed in the length direction are mounted in a cavity between the shell top plate and the machine top wall; each hoisting bracket comprises an upper connecting plate, a vertical connecting plate and a lower connecting plate; the upper connecting plate is connected with the shell top plate through a plurality of first screws, and a plurality of second screws penetrate through the machine top wall from bottom to top and then are connected with the lower connecting plate; when the second screw is disassembled, the main machine assembly can be separated from the rack shell, and disassembly, assembly and maintenance are convenient.