36results about How to "High concurrency" patented technology

The invention provides an efficient AI model management and safe loading system and method oriented to a Serverless framework, and aims to solve the high concurrency performance bottleneck and potential safety hazards of AI model loading in a Serverless environment. According to the system, an AI model management module is introduced, efficient semantic cutting is carried out on an uploaded AI model, and fine-grained encryption model blocks are generated; a high-performance P2P network is constructed by utilizing a Serverless cluster, and distributed storage and fast routing of model blocks are realized; a function instance loads a model block in a high-concurrency mode through a P2P network, a multi-stage intelligent cache is constructed locally, and dynamic caching and prefetching are achieved. According to the method, the model loading parallelism degree is improved by adopting a heuristic cutting algorithm, lightweight model updating is realized by utilizing incremental updating transactions, and the loading overhead is minimized. According to the method, the AI model loading performance and efficiency in the Serverless scene are effectively improved, meanwhile, the flexibility and safety of model management are considered, and the method has remarkable technical innovation and practical value.

The invention provides a layered artificial intelligence reasoning method and system based on Serverless, and is used for solving the problem of high cold start delay of a large-scale artificial intelligence model in a Serverless scene. The method comprises the following steps: firstly, dividing a to-be-reasoned large artificial intelligence model into a plurality of logic layers and physical layers, and constructing a workflow to coordinate an execution sequence and data transmission of each physical layer; after an inference request is received, a workflow engine creates Serverless function instances matched with resource requirements of all physical layers in parallel according to workflows, and parallel loading of models is achieved; and then coordinating serial execution of each physical layer function instance through a workflow to realize streamlined processing of model reasoning. Optimization mechanisms such as self-adaptive batch processing and cross-layer caching are further introduced, and the throughput and performance of the system are further improved. Compared with the prior art, the method has the advantages that the cold start delay of the large artificial intelligence model in the Serverless scene can be remarkably reduced, the resource utilization rate and the reasoning performance are improved, and the practical value is high.

The invention provides a business process intelligent configuration and security execution method, system and device, and the method comprises the steps: converting a natural language demand into a BPMN 2.0 process model based on a large language model and an enterprise knowledge base; the service rule description is converted into a Grouping script through multi-level security compiling; the process and the script are executed in the Flowable process engine integrated with the security sandbox, and execution data are collected; and calculating a process health score based on the data, generating and implementing an optimization suggestion, and feeding back a result to the knowledge base. The problems that a traditional process system depends on manual coding, the script safety is poor, and continuous optimization is lacked can be solved. The full-link automation and intelligentization of the business process from design, execution to optimization can be realized, the development efficiency and the system security are greatly improved, and the process has the self-evolution capability through closed-loop feedback.

The invention provides an intelligent audio and video data transmission method and device in a weak network environment, and the method comprises the steps: building a connection identifier in a protocol of an audio and video gateway based on a QUIC protocol, and connecting an interface request when a migration gateway parameter changes through the connection identifier; receiving a frame sequence, and maintaining an initial frame after the frame sequence is distributed as an IDR frame through the first frame buffer sequence container; when the read-write speed of the frame sequence is smaller than a frame tearing critical value, a second frame buffer sequence container is newly built, and frame buffer is carried out through the first frame buffer sequence container and the second frame buffer sequence container; the optimal execution action is obtained through optimization of an MDATDA algorithm, and the frame sequence obtained after frame caching is completed is played through the execution action. The method supports the video connection migration of the client, and can improve the transmission efficiency in a mobile network and weak network environment.

The invention relates to the technical field of software testing, and particularly discloses an automatic inspection platform based on a grey box test, comprising a detection console module used for managing and controlling a test process; the grey box security test overall module is used for enabling the data flow generated by the test to penetrate through the network; the grey box security test network module is used for taking a rear-end server as a computing resource and a storage resource, and packaging the rear-end server into a service by a certain management server for external provision; the grey box security test function module is used for detecting newly generated vulnerabilities in the software development process in real time; and the grey box security test tool module is used for detecting security problems in the software. By using the instrumentation technology and the flow agent, real-time vulnerability detection is carried out on the running application and API in the research and development test stage, full life cycle management and multi-dimensional application security management and control of vulnerabilities detected by an application system are supported, and the method has the advantages of high coverage, low false alarm, real-time detection and the like, and can be seamlessly fused with a DevSecOps process.

The invention relates to the technical field of Internet of Things data processing, and discloses a large-scale Internet of Things data processing method based on an object model and a rule engine, and the method comprises the following steps: equipment access and data collection: firstly, enabling a large number of Internet of Things equipment and sensors to access an edge gateway, and enabling the edge gateway to be responsible for the preprocessing of equipment data, equipment data formats are unified, then Internet of Things data are collected through Internet of Things message middleware, and large-scale data analysis is carried out subsequently based on an object model. According to the method, the object model is adopted as a unified data analysis standard, unstructured data generated by the Internet of Things is converted into meaningful structured data, semantic information is given to the data, the data understanding difficulty of a user is lowered, the data analysis speed is greatly increased, the data analysis difficulty and the equipment access threshold are lowered, and the user experience is improved. Therefore, the problems that the collection standards of heterogeneous devices of the Internet of Things are different, and collected data are difficult to analyze and standardize are solved.

The application relates to an intelligent voice communication system and method for real-time interaction between a device end and a large language model cloud end. It solves the problems of high interaction delay, lack of true full-duplex voice stream support, separation of audio stream and instruction execution, and poor interaction efficiency in the prior art. It comprises a device end and a cloud end. The device end comprises a microcontroller running a FreeRTOS real-time operating system, the microcontroller is connected with a signal acquisition and processing unit, and the microcontroller is connected with an NFC controller for device end interaction and communicates data with the cloud end through a wireless communication module. The cloud end comprises a full-duplex voice gateway, a GPU array, an AI acceleration card and a device perception scheduler. The application has the advantages that natural continuous conversation of voice in and out is realized; user instructions are recognized and executed in real time during the conversation process, voice feedback is returned synchronously, and low-delay, low-power-consumption, high-concurrency full-duplex voice interaction is realized.

The invention discloses a cloud server cipher machine-oriented virtual cipher machine fault drift method and system. The method comprises the following steps of: managing a VSM resource pool and issuing a platform drift key pair to each virtual cipher machine VSM; continuously monitoring the running state of each VSM, and when a fault is detected, packaging a migration packet of core data carried by the fault VSM; generating a symmetric key, packaging the migration packet into a digital envelope through a platform drift key, and generating a digital signature to obtain a key data drift packet; a target VSM is dynamically selected, a key data drift packet is received, a digital envelope is decrypted to obtain a migration packet, a digital signature is verified, the migration packet is transmitted to the target VSM, key assets and configuration information are thermally loaded, and data recovery is completed; unbinding the IP address of the source fault VSM and configuring the IP address to the target VSM in real time; performing secure erasure and initialization operation on the faulted VSM, and re-incorporating the faulted VSM into the resource pool; the system comprises a management center, an equipment management service and a virtual cipher machine.

The invention discloses a time efficiency label generation method and device and a storage medium, and relates to the technical field of logistics time efficiency, and the time efficiency label generation method comprises the steps: responding to a time efficiency label obtaining request, and determining whether the client time corresponding to the time efficiency label obtaining request is before the preset time; and if the client time is before the preset time, constructing query input based on a commodity identifier corresponding to the timeliness label acquisition request, and querying the white list data in the memory cache. When the subitems in the white list data are hit, the shipping warehouse information and the receiving place information corresponding to the timeliness label acquisition request are assembled into a key, and the precomputation state of the target route is inquired in the Redis cache based on the key assembled by the shipping warehouse information and the receiving place information. And if the pre-calculation state is normal, the next day arrival tag is used as a response of the timeliness tag acquisition request and is sent to the client. According to the method, the technical effect of high-robustness and high-certainty time-efficiency label calculation capability of the million-level QPS can be realized.

The invention discloses a computing power resource dynamic slicing and isolation method for an artificial intelligence target range, and the method comprises the following modules: S1, business feature perception and dynamic resource slicing: carrying out the analysis and modeling of an artificial intelligence attack and defense task through a mechanism, and mapping a task demand into a corresponding computing power slicing specification; s2, a driving-level video memory firewall, which monitors the video memory usage behavior of the virtual instance in real time on a driving layer of an execution node, and prevents attack behaviors such as abnormal occupation of the video memory; and S3, carrying out atomized rollback based on the model state of the shared memory. According to the mechanism, rapid recovery of the environment is realized through page table resetting. The method is oriented to an artificial intelligence target range, and the problems of static low efficiency, insufficient safety and isolation, lack of a quick recovery mechanism and the like of resource scheduling are solved through the mechanisms.

The embodiment of the invention discloses an equipment control method and device and an ad hoc network system, and the method comprises the steps: setting a plurality of semantic tags for node equipment of an ad hoc network from a plurality of different dimensions, embedding a target node equipment tag group in equipment control information when scene control or group control is carried out, and carrying out the scene control or group control. And positioning the target node equipment needing to be controlled by using the target node equipment label group, and sending the equipment control information in the ad hoc network in a broadcast mode. And meanwhile, each node device matches and judges whether a target node device exists or not based on a semantic tag of the node device and a target node device tag group by broadcasting the received device control information, and executes an action instruction in the device control information when judging that the node device is the target node device. Thus, scene control or group control can be realized in a low-latency manner without controlling a command or address translation of the gateway device.

The invention discloses a coal mine robot unified dispatching command system based on MQTT. The coal mine robot unified dispatching command system comprises a robot, an EMQX server, an application server, a hard disk video recorder and a computer client. The robot is respectively connected with the EMQX server, the application server and the hard disk video recorder; the EMQX server is in communication connection with the application server, and the application server is respectively in communication connection with the hard disk video recorder and the computer client; according to the system, the robot pushes a real-time state to the EMQX server and pushes a video to the hard disk video recorder, the real-time state and the video are uniformly received by the dispatching command program, and a signal for controlling the robot is sent to the EMQX server through the dispatching command program and forwarded to the robot for execution. And the client views and remotely controls the robot by accessing the dispatching command program. The system has the capabilities of low delay and high quality under the conditions of high concurrency and high load, and ensures accuracy and timeliness under the conditions of unstable network environment and strong interference.

The application relates to a connection management method and system, the method is applied to a signaling server, the signaling server is independent of a cluster server, the method comprises the following steps: in response to a registration instruction, sending a registration signal to the cluster server; receiving registration information returned by the cluster server in response to the registration signal; the registration information comprises first address information, the first address information of different signaling servers is different; receiving a connection signal initiated by a client to the corresponding signaling server based on the first address information; and establishing a signaling channel connected with the client based on the connection signal. Through the foregoing method, the interference of the business logic of the cluster server on the signaling connection can be reduced, the cluster server resources can be avoided from being contended, low delay and high concurrency can be ensured, service jitter caused by the rapid change of the business logic of the cluster server can be reduced, the stability of the signaling connection can be improved, the business branches can be simplified, the delay can be reduced, and the connection efficiency and stability can be improved.

The present invention provides a marketing assistance system and method suitable for multiple personalized services, including the following modules: a system table structure module: storing the marketing list of the service; a system service module: providing services according to the stored marketing list of the service. The present invention can efficiently manage target customers, and has the capabilities of high reliability, high concurrency and high performance; and can fully support service marketing, which can meet the needs of personalized service marketing, as well as the mixed marketing capabilities of different services and the ability to quickly respond to any newly added service.

This invention relates to an optimization method for high-speed memory read and write operations. The method involves dividing the memory address space into multiple address units according to a preset address granularity; receiving write and read requests for each address unit; when an address unit receives a write request, if no corresponding write record to be committed exists in the local commit area, the write record to be committed is created and the write data is recorded; if the address unit has the write record to be committed and is not interrupted by the read request, the latest write data in the write record to be committed is updated for subsequent write requests, so that multiple writes to the address unit are merged into a final write result before physical commit; when the read request hits the write record to be committed, the write record to be committed outputs the read result; when the commit condition is met, the final write result corresponding to the write record to be committed is written to physical memory, and the write record to be committed is released.

The present application relates to the field of communication technology, and more particularly to a kind of northbound interface command processing method and system, method includes: through user service and upper system docking, receive the northbound command issued by upper system, parse northbound command and convert into uniform northbound message, then according to the mapping relationship table of equipment network element and service, uniform northbound message is sent to the service corresponding to equipment network element;Through service, uniform northbound message is processed, the conversion of service data is carried out and is issued to equipment network element, and the decomposition and reorganization of service flow are carried out and are sent to equipment network element and execute.The user side and service side function of northbound service are separated in the present application, different types of northbound interface service reuse same service, reduce the cost of opening maintenance and deployment, can make network management system in high concurrency and high throughput, with higher reliability, can reduce the node congestion phenomenon caused by single user or equipment failure.

The invention discloses a bank-based distributed collection service system. The collection service system comprises gateway service, a registration center assembly, a configuration center assembly, a monitoring assembly, a log management assembly and an intelligent collection management service assembly. And meanwhile, k8s deployment is carried out through the cloud platform, so that distributed development is realized, the pod number can be automatically and flexibly expanded, and the requirements of high concurrency, high performance and high availability are met.

The invention discloses a distributed power real-time data access system and method. The system comprises a data ID unified coding module, a unified data access interface module, a fault-tolerant device module and a regional service node module. The data ID unified coding module adopts a structured global unique coding rule, and a code comprises a region code used for directly positioning a data physical storage position; the unified data access interface module receives a client request, groups transmission data, dynamically allocates the request to an optimal regional service agent through a load balancing strategy, and performs data compression, decompression and result aggregation; the fault-tolerant device module realizes overtime processing, failure retry, fault recovery and heartbeat detection functions; the regional service node module is responsible for storage, query and processing; according to the invention, the efficiency and reliability of distributed power real-time data access can be improved.

The application relates to a task scheduling method, device and computer equipment. The method comprises the following steps: obtaining task information of each to-be-scheduled task and resource states of each computing node in a computing cluster; wherein the task information of each to-be-scheduled task at least comprises data volume and data distribution information of processing data required by the to-be-scheduled task; determining execution time lengths of each computing node for executing each to-be-scheduled task according to the task information of each to-be-scheduled task and the resource states of each computing node; determining load imbalance degrees corresponding to each candidate scheduling strategy according to the execution time lengths of each computing node for executing each to-be-scheduled task; the candidate scheduling strategy comprises node allocation of each to-be-scheduled task; selecting a target scheduling strategy from each candidate scheduling strategy according to the load imbalance degrees corresponding to each candidate scheduling strategy; and scheduling each to-be-scheduled task based on the target scheduling strategy. The method can improve resource utilization and task processing efficiency.

The present disclosure relates to a database-based data processing method, device, medium and electronic equipment, the method comprising: receiving a query request task to be executed, and determining a plurality of coroutine tasks corresponding to the query request task, the coroutine tasks being initially added to a global task queue shared by a plurality of threads, and then being distributed from the global task queue to a local task queue of a corresponding thread; in each thread, determining a target coroutine task to be executed according to time information of each coroutine task in the local task queue of the thread; interrupting the target coroutine task and adding the target coroutine task to the global task queue when the target coroutine task is not executed and the target coroutine task is after the current time slice of the execution of the thread; determining a new target coroutine task of the thread according to the global task queue and the local task queue of the thread, and executing the new target coroutine task in the next time slice.

The invention discloses a message platform publishing method and system. The method comprises the following steps: acquiring initial information; key words in the initial information are extracted, and the value corresponding to the initial information is determined; when the value corresponding to the initial information is greater than a preset first threshold value, generating brief information corresponding to the initial information according to keywords in the initial information; determining an information type corresponding to the brief information according to a keyword in the initial information; obtaining a subscribed user side according to the information type of the corresponding brief information; and based on a notification mode preset by the subscribed user side, sending the corresponding brief information to the corresponding subscribed user side. According to the invention, intercommunication and integration among different types of information are realized, complexity is changed into simplification, unified management and distribution are realized, and high-concurrency, high-quality and high-stability message service is provided.

The present invention provides a document generation method, device and electronic device, wherein a preset development tool obtains template configuration data of each original file from a cloud server based on RDMA technology; generates a document template corresponding to each original file based on the template configuration data; obtains the first key content data corresponding to each original file, and generates a file conversion task corresponding to each original file based on the original file type and converted file type corresponding to each original file and the obtained first key content data and document template; and a distributed platform executes the obtained file conversion tasks in parallel to convert each original file into a corresponding document with a converted file type. The present invention can alleviate the problems of low efficiency and poor precision of file format conversion in related technologies.

The invention discloses an item data distribution method and system based on message queues and interfaces, belongs to the technical field of data distribution, and aims to solve the technical problem of how to realize efficient, stable and complete distribution of data. When the matter takes effect and is released or the matter cancels logic change, the data producer pushes key information of matter change logic to a message queue and provides a matter detail interface used for obtaining matter detail information; and the data consumer obtains key information of the item change logic from the message queue, calls an item detail interface to obtain item detail information based on the key information, and performs item analysis and logic storage according to the item change logic and the item detail information.

The embodiment of the invention provides a resource access method of a distributed storage system, the distributed storage system comprises a server and a client, and after the server establishes connection with the client, a plurality of task queues and corresponding threads are created; receiving a plurality of messages sent by the client in the process of accessing the target resource of the server; generating corresponding tasks according to the plurality of messages; task information of the tasks is determined, and the tasks are added to corresponding task queues according to the task information; and processing the task in the task queue through a thread corresponding to the task queue so as to complete the access of the client to the target resource of the server. By dividing the tasks according to types and distributing the tasks to different threads for processing, mutual interference among the tasks can be reduced to the maximum extent, and the overall performance of the system is improved.

The application relates to the field of cloud computing storage, and proposes a logical volume access method, device, equipment and medium. A logical volume access method is applied to an application layer, and the method is characterized in that the method comprises the following steps: calling a target logical volume activation interface of a daemon process; after the target logical volume activation interface is returned, executing a target logical volume access task; and calling a target logical volume reverse activation interface of the daemon process. The application encapsulates the activation and reverse activation operations of the logical volume through the daemon process, solves the safe access of the shared logical volume between multiple tasks in the concurrent multitask scenario of the application layer on the host, and makes the logical volume information used between the concurrent tasks no longer report errors and abnormities, so that the concurrent operation tasks of the upper multitasks can be ensured to be successfully performed.

The invention discloses an FPGA (Field Programmable Gate Array) calling method based on a window sliding method, which comprises the following steps: a CPU (Central Processing Unit) monitors TCP / UDP (Transmission Control Protocol / User Datagram Protocol) data of a port, a plurality of new threads are created according to an intercommunication object, and each thread independently records interaction information with different equipment; when the multiple threads receive the service data at the same time, the CPU dynamically adjusts the sequence of the service data sent to the FPGA according to the current operation condition of the FPGA and the performance of processing different services; the method comprises the following steps of: continuously sending service data to be processed to an FPGA (Field Programmable Gate Array) by using a window sliding method according to the operational capability of the FPGA, and dynamically adjusting the size of each service sub-window in a window according to the operation condition of different services which are currently carried out in the FPGA and the processing condition of each different service. The invention further discloses an FPGA calling device based on the window sliding method, corresponding equipment and a storage medium. According to the invention, the high-concurrency processing capability of the equipment when a large number of different services are processed at the same time is improved, and the equipment is ensured to meet the use requirements of high reliability, high concurrency and low time delay.

The invention discloses an insurance industry distributed data processing method and device, and the method can effectively solve the problems of performance bottleneck and poor expansibility of a traditional centralized database in a mass data scene in the insurance industry. Stream batch integrated processing of high-concurrency transaction processing and batch analysis tasks is realized, and the system processing efficiency and availability are improved.

The invention discloses an Actor model-based industrial data acquisition task dynamic scheduling method, which comprises the following steps of: creating a scheduler Actor, a connection management Actor and a protocol processing Actor in an Actor operation environment, and initializing a layered time wheel timer; dynamically creating an independent connection Actor instance for each data acquisition connection request through the connection management Actor; sending a data acquisition command message to a corresponding connection Actor instance through the scheduler Actor, so that the connection Actor instance performs data acquisition to obtain a data acquisition result message of a data acquisition result; standardizing a data acquisition result in the data acquisition result message through a protocol adaptation layer managed by a protocol processing Actor; and publishing the standardized data to a downstream system. Through the isolation characteristic of the Actor model and an asynchronous message driving mechanism, high-concurrency and low-delay industrial data acquisition is realized, the reliability, the expandability and the resource utilization efficiency of the system are remarkably improved, and the data acquisition requirement in a large-scale industrial Internet of Things scene can be effectively met.

The invention discloses a FastFlood detection method and device based on an Flink flow processing engine, and the method comprises the steps: collecting flow data from various data sources in real time through the Flink flow processing engine, and distributing the flow data to a corresponding sliding time window; sliding the time window to allow the window to slide on the time axis, and capturing the flow change by setting the window size and the sliding step length; carrying out statistical analysis on the flow data in each sliding-out window; the statistical analysis result is compared with a preset threshold value, and whether typical features of the pulse type DDoS attack are presented or not is judged; and if the pulse type DDoS attack is identified, triggering a corresponding protection mechanism. According to the method and the device, the limitation of an operator-level network scale can be spanned, seamless coverage and efficient data acquisition can be realized, attack characteristics can be accurately extracted in a complex and changeable network environment, and a high-sensitivity real-time detection model can be constructed.

The application discloses a kind of intersection traffic safety analysis method and system based on digital twinning, it is related to traffic safety technical field.The application is by dividing intersection physical space into multiple subarea and each establishing independent digital twin body instance, high concurrency, low delay monitoring to local traffic conflict is realized, the real-time and delicacy of safety analysis are significantly improved;Based on the dynamic binding and prediction collision time judgment of vehicle-pedestrian feature pair, individual combination with interactive risk can be accurately identified, avoid the missed detection of traditional macroscopic statistical method to instantaneous dangerous event, and provide complete and traceable original sensor data package for accident traceability, responsibility determination and twin body model iteration.