45results about How to "Reduce bottlenecks" patented technology

The invention discloses a multi-warehouse intelligent distribution and scheduling method for raw material inventory. The method comprises the following steps: collecting raw material inventory data of each warehouse, and monitoring the inventory data of the warehouses in real time; based on the real-time inventory data and the historical inventory data, the inventory state of each warehouse is analyzed to determine the inventory balance state of each warehouse, and the warehouses with excess or short inventory are identified; and calculating an optimal inventory distribution scheme by adopting a dynamic adjustment algorithm based on an inventory state analysis result. According to the invention, by collecting the real-time inventory data and the historical inventory data, the inventory state of each warehouse is deeply analyzed, the inventory balance state is determined, and the warehouse with excess or short inventory is identified. Through the real-time data monitoring and analysis, the system can provide accurate inventory early warning, help managers to take measures in time, avoid production interruption and resource waste, and improve the accuracy and predictability of inventory management.

The invention belongs to the technical field of industrial control, and discloses a bottleneck prediction optimization method, device and equipment for a battery positive electrode material production line, and the method comprises the steps: collecting the spatial data and time series data of the production line in real time; synchronizing the spatial data and the time sequence data to a digital twinborn model, wherein the digital twinborn model is pre-constructed based on the production line; converting the spatial data and the time sequence data into spatio-temporal data; inputting the spatio-temporal data into a pre-trained spatio-temporal diagram convolutional network to obtain a prediction result of the bottleneck of the production line; when it is judged that a production line bottleneck exists according to the prediction result, based on the prediction result, an optimization strategy is generated according to a priority rule, the optimization strategy is operated on the digital twin model in a simulation mode, a simulation result is obtained, and if it is judged that the simulation result meets the expectation, the production line is controlled according to the optimization strategy. Efficient and stable operation of a production line can be ensured, and production efficiency and stability are improved.

The invention provides connector assembly automation equipment, a connector comprises a frame and a contact pin, the connector assembly automation equipment comprises a divider turntable rotating around the axis of the divider turntable, and a plurality of stations matched with the frame are arranged on the divider turntable at intervals in the circumferential direction. A frame feeding mechanism, a contact pin feeding mechanism, a contact pin press-fitting mechanism and a product discharging mechanism which correspond to different stations are distributed on the outer side of the divider rotating disc. According to the connector assembly automation equipment, manual operation is replaced by automatic and intelligent equipment, the production efficiency is comprehensively improved, efficient line production can be achieved through the multi-station divider rotating disc and the automatically-controlled feeding and discharging system, the bottleneck in production is reduced, the output speed of products is increased, and the production efficiency is improved. The visual inspection mechanism is introduced to replace naked eyes of operators to observe the assembly quality, and the qualified rate of products is increased.

The invention discloses a network security attack and defense confrontation scene generation system for a novel power system. The scene editor module receives user interaction and constructs or modifies an attack and defense scene; the network simulation module generates real network environment data based on a network space surveying and mapping technology, constructs a simulation network environment and manages simulation data through a simulation database; the AI automatic control module integrates an AI algorithm to simulate an attack behavior, intelligently generates and adjusts a defense strategy, and manages an attack and defense rule through a rule engine; the security isolation module deploys security equipment to realize sensitive data isolation; the data flow direction is controlled through strategy configuration, and information leakage is prevented; the evaluation and feedback module is used for collecting data in the drilling process, carrying out multi-evaluation on the drilling effect and generating reports and suggestions; and the business module adopts a plug-in architecture, dynamically loads or unloads each functional module, constructs a multi-layer defense system, and realizes generation, evaluation and feedback of an intelligent confrontation strategy. According to the invention, the authenticity and effectiveness of attack and defense drilling are improved.

The invention discloses a method and a system for improving data interaction efficiency of a hyper-converged architecture, relates to the technical field of hyper-converged architectures, and aims to solve the problems of network bottleneck and storage delay of data transmission among different virtual machines or containers in the conventional hyper-converged architecture. According to the scheme, a data transmission optimization module needs to be introduced into a hyper-converged architecture, so that the data access speed and the network efficiency are improved through data prefetching and an intelligent routing technology; the distributed cache module is used for relieving the pressure of a single node and ensuring accurate synchronization of data through data locality optimization and cache consistency management; the parallel data processing module is used for improving the data processing efficiency through task parallelization and data fragmentation technologies; and the self-adaptive load balancing module is used for dynamically adjusting resource allocation according to the system load and the resource use condition so as to realize the balance of calculation, storage and network load. The method reduces the delay, improves the throughput, and improves the overall performance and resource utilization rate of the system.

The invention relates to a digital factory production management method and system, and the method comprises the steps: predicting a future initial order demand based on a data analysis tool and historical sales data, and adjusting the initial order demand based on a current market trend and a current market strategy corresponding to a target peer, and obtaining a target order demand; obtaining the required production capacity and the equipment utilization rate corresponding to each piece of production equipment based on the target order demand; acquiring real-time operation data corresponding to each piece of production equipment, and predicting maintenance requirements corresponding to the real-time operation data, the required production capacity and the equipment utilization rate based on the trained deep learning model; and making a maintenance plan of each production device in advance based on the maintenance demand and the maintenance priority corresponding to each production device. The production equipment is subjected to preventive maintenance, the reliability and stability of the production equipment are improved, the use of maintenance resources is optimized, and the maintenance efficiency and quality are improved.

According to the macular oedema diagnosis system, method and device based on color fundus photography, the medium and the program product, the bright part enhancement and dark part enhancement of the CFP image are combined, the problem that diagnosis is difficult due to the fact that macular oedema features are not obvious in traditional CFP image diagnosis is solved, and the diagnosis accuracy and efficiency are greatly improved. Experimental results show that the accuracy rate of the method can reach 95% or above on a private data set, and the accuracy rate of multi-stage diagnosis of macular edema on a public macular degeneration and macular edema combined data set IDRID can reach 85%. Besides, a Mama layer in an SSM (Spatial-Sense-Modulation) framework is adopted to perform global feature modeling, so that the capability of capturing the global relationship of the image is further improved, the bottleneck problem encountered in actual clinical application due to the fact that a Transform model depends on computing resources is reduced, the requirement for high-performance computing equipment is reduced, and the method has the advantages of being high in practicability and easy to popularize. And possibility is provided for application in areas with limited resources.

The invention relates to a data processing method of a neural network, the neural network and a chip, the data processing method of the neural network comprises the following steps: a weight data storage step: loading weight data into a PIM; a weight data distribution step: distributing the weight data to the corresponding Bank; and a calculation step: performing Mac calculation based on the data of each Bank. The method comprises the following steps of: loading weight data to a process In Memory (PIM) unit, distributing the weight data to different Banks, and carrying out multiply-add (MAC) operation on the basis of data in each Bank. According to the method, data processing and storage are integrated, so that the transmission bottleneck of data from a memory to a processing unit is reduced, the calculation delay and power consumption are reduced, and particularly, the efficiency can be remarkably improved in large-scale neural network calculation.

The invention relates to the technical field of database data import, and provides a method, system and device for importing time sequence data into an openGauss database, and the method comprises the steps: reading original time sequence data through a worker thread, partitioning the original time sequence data according to a time mark and a label mark, and storing the partitioned original time sequence data into a corresponding skip list in a memory; reading the data in the skip list through a write thread, and writing the read data into a bulkload rows memory structure according to columns; and constructing corresponding metadata information according to the data in the bulkload rows memory structure through a write thread, and writing the data into a column storage file. According to the method, the system and the equipment for importing the time sequence data in the openGauss database, resources such as a CPU (Central Processing Unit) and a memory of an operating system can be fully utilized, the bottleneck of the system resources is reduced, the data importing efficiency is ensured, and the storage characteristic of the time sequence database is not lost.

The invention belongs to the field of production management, and discloses a method and a system for intelligently arranging mixed styles on a sewing assembly line, and the method comprises the steps: combining the working procedures of various styles into operation elements according to the similarity detection and mixed style process analysis of single style process analysis data before the mixed arrangement production of various styles; generating a mixed intelligent process flow diagram containing an operation element table based on the operation elements; calculating order values of the operation elements according to the wiring setting parameters in combination with the mixed intelligent process flow diagram, generating an operation element sequence table, and combining the operation elements according to a preset rule to obtain an optimal arrangement scheme of the operation elements; and calculating the commissioning sequence of the mixed arrangement of the styles, and cyclically commissioning in the mixed assembly line according to the commissioning sequence. According to the method, small and medium-sized clothing factories are not subjected to high equipment investment, and for quick-return small bills with multiple styles, small quantity and tight delivery time, the small bills are changed into large bills, so that continuous and efficient output is realized.

The invention provides a memory and an allocation method of flash memory blocks thereof, and the memory comprises the flash memory blocks which are divided into healthy blocks and risk blocks; the main controller is used for dividing the data invisible to the outside in the memory into unimportant data which does not need to be updated, important data which does not need to be updated and data which needs to be updated according to types; wherein the main controller is used for writing the data which does not need to be updated and is not important into a certain risk block; the main controller is also used for writing the important data which does not need to be updated into a certain health block; and the main controller is also used for writing the data needing to be updated into another health block. According to the memory and the allocation method of the flash memory blocks, the flash memory blocks in the memory with small capacity can be fully utilized.

The invention relates to a sketch general configuration method and system based on a data processing unit, and belongs to the technical field of data processing. The method comprises the following steps: (1) identifying resource parameters; (2) selecting a process model; (3) resource constraint input; (4) optimizing target configuration; and (5) parameter configuration recommendation. The invention aims to optimize the communication efficiency of the DPU and the host, reasonably allocate hardware resources, and realize flexible process model selection and efficient state operation configuration, thereby achieving the goals of high accuracy and high throughput of network measurement under limited resources.

The application discloses a data read-write method and device based on cooperation of DDR and SPI interfaces, and the method comprises the following steps: acquiring state data sets of the interfaces, data transmission priority requirements, real-time bus access states and system timing coordination requirements; performing load evaluation according to the state data sets to obtain a classification result, combining the data transmission priority requirements to calculate the emergency degree of bus access, and obtaining a priority sequence; performing time slice allocation according to the priority sequence and the real-time bus access states to obtain an access right allocation scheme; executing the access right allocation scheme to generate a resource conflict resolution strategy; performing task re-allocation according to the resource conflict resolution strategy and the system timing coordination requirements to obtain a read-write task re-allocation result; monitoring the read-write task re-allocation result to obtain interface state change information, and performing optimization according to the read-write task re-allocation result and the interface state change information to obtain a multi-interface cooperative working configuration.

The invention provides an election method and device used between switches, relates to the technical field of network communication, and aims to improve network stability and communication efficiency in a ring network system. The method comprises the following steps: after a node is initialized, a ring network port of a switch initiates an election message regularly, and the election message comprises an MAC address of the node, a node ID, an election port, a priority, a node consistency coefficient and a network load adjustment parameter; the node with the highest priority is defaulted as a main node, and other nodes stop election and forward election messages; and after the main node finishes election, the port state is dynamically adjusted based on the received message. The device comprises a processor, a memory and a bus, and the processor executes instructions in the memory to implement the election method. According to the method, indexes such as the node consistency coefficient, the network load adjusting parameter and the link health degree are introduced, the election process is dynamically optimized, the stability and the election efficiency of the network are improved, and the problems of resource exhaustion and network instability in the prior art are solved.

The invention discloses a hardware-aware early-quit neural network optimization method, which aims to solve the problems existing in research, performs hardware-aware early-quit neural network optimization research by combining related parameters of a data stream and a hardware accelerator, and performs joint optimization on related settings of an early-quit branch and hardware parameters through an NSGA-II algorithm. According to the method, the acceleration of the inference process and the reduction of the energy consumption of the early-quit network are realized, so that the calculation task is optimized, a hardware accelerator is better supported to execute the calculation task of a specific type, the experiment is carried out based on CIFAR-10, CIFAR-100 and TinyImageNet data sets, software MAESTRO popular in current accelerator simulation is used for calculating the energy efficiency of a convolutional layer and a full-connection layer in the convolutional neural network, and the energy efficiency of the convolutional layer and the full-connection layer in the convolutional neural network is improved. And under the same accuracy score level, the time delay is reduced while the energy consumption is reduced.

The embodiments of the present disclosure disclose a distributed computing communication method and device, a distributed computing method and system. The distributed computing communication method includes: receiving an operation instruction; in response to the operation instruction, identifying the type of the processing unit of the target device, and loading the target communication library to the target device according to the type of the processing unit of the target device; based on the type of the processing unit of the target device and the target communication library, selecting the first control instruction of the target device, and realizing local communication between the processing units of the target device by executing the first control instruction; controlling the processing unit of the target device to transmit the execution result to the first network switching device; controlling the communication between the first network switching device and the second network switching device, wherein the second network switching device is a network switching device that connects the target device to the heterogeneous device. This method can solve the problem of communication difficulties between each heterogeneous device, help multiple types of devices to work together, and improve the efficiency of distributed computing.

This invention relates to the field of artificial intelligence technology and discloses a method for constructing a rewriting model based on a mixed-precision strategy. The method includes instantiating computation nodes in a computation flow graph into multiple intelligent computing units. Each intelligent computing unit maintains a local state vector containing its current computational precision and encapsulates a rule engine. The rule engine performs local inference and generates impact messages containing adjustment proposals. When impact messages are transmitted between intelligent computing units, distributed negotiation based on these messages is triggered, and the computational precision of the corresponding intelligent computing units is collaboratively adjusted according to the negotiation results. A central policy controller collects global state information from the computation flow graph and dynamically optimizes the inference rules used by the rule engine based on this global state information. Through a dual-layer driving approach of local inference and global layer optimization, the mixed-precision strategy and the computation flow graph are dynamically matched and optimized, balancing the allocation of computational precision and hardware resources.

This invention discloses a front-end hardware acceleration system, a program proof hardware acceleration system, and a method. Addressing the problem that existing zero-knowledge virtual machine front-ends still heavily rely on software implementation, this invention proposes a dedicated heterogeneous hardware architecture for front-end execution, main trajectory generation, and auxiliary trajectory generation. This significantly reduces front-end bottlenecks. Furthermore, the main trajectory unit, trajectory buffer unit, and auxiliary trajectory unit work collaboratively, forming a unified on-chip processing link for program execution, main trajectory generation, and auxiliary trajectory generation. On-chip collaborative processing of the main and auxiliary trajectories involves direct on-chip data transfer through the trajectory buffer unit, avoiding the need to read all main trajectories back from main memory and reducing the round-trip overhead of "main trajectory in main memory—software read back—regeneration of auxiliary trajectories." Based on the front-end hardware acceleration system, further through units such as trajectory shaping, polynomial processing, constraint combination, commitment generation, low-degree testing, query initiation, and proof encapsulation, the front-end output is directly extended to the back-end proof construction, reducing stage boundary overhead and forming an integrated end-to-end hardware processing architecture from "program execution - trajectory generation - commitment construction - low-degree testing - query initiation - proof encapsulation". Moreover, through the trajectory shaping unit, the front-end output can be directly connected to the back-end proof construction, reducing additional data format conversion overhead.

The present invention discloses a multi-UAV collaborative path planning method, system, device and storage medium, which relates to the technical field of UAV path planning, and includes the steps of: obtaining the flight trajectory of multiple UAVs represented by cylindrical coordinates; obtaining a detection function using the increment of the objective function of each path point in the current global optimal path, and obtaining the bottleneck path point of the global optimal path through the probability of each path point in the detection function; obtaining a candidate target vector according to the probability, and adding disturbance to the randomly sampled path, obtaining the candidate path through the current disturbance operator, and generating a new target path point of the global optimal solution; planning the multi-UAV collaborative path through the target path point. The present invention corrects the detected bottleneck waypoints through a two-stage update strategy constructed by operations such as disturbance, and the generated path greatly reduces the existing bottlenecks, generating a high-quality global optimal path with a high success rate.

The invention provides a data reconstruction method, a distributed storage system and a computer readable storage medium, and is applied to the technical field of distributed storage. The data reconstruction method is applied to a proxy repair service, the proxy repair service is deployed at a persistence layer client of a distributed storage cluster, the distributed storage cluster comprises a plurality of nodes, and each node is configured with the persistence layer client and an object storage server. According to the method, the proxy repair service is applied in the distributed storage cluster, local loopback communication between the object storage server and the persistence layer client is realized, and cross-node network data transmission is remarkably reduced by reading local source fragmented data and executing erasure code reconstruction, so that the data reconstruction network bottleneck in a large-scale EC scene is effectively relieved, and the data reconstruction efficiency is improved. And the data reconstruction efficiency and the overall availability of the cluster are improved.

The invention discloses a post-quantum digital signature key generation method based on AVX2 acceleration and matrix multiplication optimization, which comprises the following contents: on the basis of original coding, parallel optimization of signature compression and decompression processes is realized by using an AVX2 instruction set, the coding and decoding efficiency is remarkably improved, and the method is particularly suitable for a large-scale data processing scene. A matrix multiplication optimization method for an algorithm is provided, and recursion operation is avoided by only performing matrix partitioning once. And in combination with a register optimization technology, the memory access is effectively reduced while the calculation accuracy is ensured, and the calculation efficiency in the signature generation process is remarkably improved. A register optimization strategy is introduced in matrix multiplication and addition operation, and a calculation result is stored in a register as much as possible by reducing memory access operation, so that the number of times of writing back to a memory is reduced, the execution efficiency of matrix operation is further improved, and the method has important significance on efficient implementation of a post-quantum digital signature scheme.

The present invention relates to the technical field of electronic digital data processing, and specifically discloses a parallel tensor filling method based on a GPU. The method comprises: initial matrix generation, tensor division accuracy determination, sub-tensor parallel processing and parallel tensor filling, wherein the initial matrix is ​​generated according to a matrix method, sub-tensors are obtained by a preset tensor division method based on the data amount of a data set to be filled, and the division accuracy is verified to optimize the division method; sub-tensor binary coordinates are obtained by a predefined coordinate positioning method, a parallel sub-tensor sequence is obtained by combining a parallel tensor search method, and the sub-tensor sequence is transmitted to a GPU memory space for parallel processing, the GPU parallel processing performance is evaluated, the weight of each sub-tensor sequence is obtained by parallel processing, the initial matrix is ​​updated to an update matrix, and finally the missing data is filled in a predefined method, the parallel tensor filling of the data is completed, and the accuracy and integrity of the data are improved.

This invention discloses a method and system for reusing discarded convolutional information in small-sample fine-grained image classification, belonging to the field of computer vision. The method includes the following steps: acquiring images of a support set and a query set, and extracting the backbone feature descriptors of the images using a convolutional neural network; after at least one convolutional block in the convolutional neural network, retrieving the discarded feature information of the convolutional block through a feature retrieval path to generate a retrieved feature descriptor; fusing the backbone feature descriptor with the retrieved feature descriptor to obtain a high-dimensional feature descriptor in a unified format; calculating the similarity between the images of the query set and the support set based on the high-dimensional feature descriptor, and performing small-sample fine-grained image classification based on the similarity, thereby realizing the reuse of discarded convolutional information. This invention can effectively retrieve discarded feature information in convolutional neural networks, improving the performance of small-sample fine-grained image classification.

The utility model discloses an adhesive tape manufacturing device which comprises a concentric-square-shaped machine frame and a cavity arranged at the top end of the concentric-square-shaped machine frame, a supporting plate is fixed at the top end of the concentric-square-shaped machine frame on one side of the cavity, a supporting column is installed on the outer wall of the side, close to the cavity, of the supporting plate in a rotating mode, and a blocking disc used for limiting an adhesive tape column is fixed at one end of the surface of the supporting column. The top end of the concentric-square-shaped rack is provided with a pneumatic lower swing type cutting assembly used for cutting the adhesive tape column. The automatic blanking device has the advantages that the active blanking speed is higher than that of manual blanking, more adhesive tapes can be processed within the same time, the bottleneck of a production line can be relieved, the production can be carried out more smoothly, the high-yield requirement is met, meanwhile, the time of operators near operating mechanical equipment can be shortened by adopting a mechanical structure to carry out active blanking, and the production efficiency is improved. The operation safety is improved.

The invention relates to generating a mapping of a physical space from point cloud data. In particular, this specification describes systems and methods for generating a mapping of a physical space from point cloud data of the physical space. The method may include: receiving the point cloud data of the physical space; filtering the point cloud data to at least remove sparse points from the point cloud data; aligning the point cloud data along x, y, and z dimensions corresponding to an orientation of the physical space; and classifying the points in the point cloud data as corresponding to one or more types of physical surfaces. The method may also include identifying a particular physical structure in the physical space based at least in part on a classification of the points in the point cloud data; and generating the mapping of the physical space to identify the particular physical structure and a corresponding profile of the particular physical structure within the orientation of the physical space.

The present invention discloses an offline teaching learning system for a long-range continuous control task of an intelligent body, comprising: a long-range candidate path planner module, which is responsible for generating multiple candidate paths containing multiple sub-goal states according to the initial states of the manipulated object and the intelligent body in the task environment; an optimal path selection module, which is responsible for evaluating the value of the generated multiple candidate paths and selecting the path with the highest value as the optimal path; an optimal action selection module under sub-goal constraints, which is responsible for using the optimal path as a guidance path, using a single sub-goal state on the guidance path as a constraint, generating the action of the intelligent body according to the current manipulated object and intelligent body state to reach a new state, and after multiple state transitions, replacing the sub-goal until all sub-goal states on the guidance path are tracked, and completing the long-range continuous control task of the manipulated object. The present invention is conducive to the intelligent body to use human teaching data for offline learning in a real environment.

The invention discloses a traffic detection method, and particularly relates to the technical field of traffic detection, which comprises the following steps of: screening unencrypted network traffic in all network traffic in equipment, preferentially detecting, extracting encrypted traffic after the unencrypted network traffic is detected, and acquiring the total number of encryption types corresponding to the encrypted traffic; obtaining a traffic quantity coefficient, a transmission frequency coefficient and a transmission instability coefficient of the corresponding encrypted traffic of each encryption type, and obtaining a detection priority index in combination with a preset decryption difficulty value corresponding to an own encryption mode; according to the method, the detection priority indexes are sequenced from large to small, and the encryption traffic corresponding to the encryption type with the maximum detection priority index is preferentially detected, so that decryption detection can be reasonably performed on the encryption traffic corresponding to different encryption modes, the bottleneck of a detection system is reduced, and the detection efficiency is improved. The traffic with a relatively large abnormal degree can be identified in time, and the overall network security and stability are further improved.

The invention provides a multi-device polling control method and system for a PLC electric power communication automatic control cabinet. Relates to the technical field of PLC power communication. The method comprises the following steps: acquiring historical state acquisition data of a plurality of slave devices, performing state fluctuation collaborative clustering on the plurality of slave devices to obtain a plurality of state collaborative device sets, and generating global fault probability parameters and fault collaboration parameters of the state collaborative device sets according to historical fault data of the slave devices; performing host polling distribution analysis on the plurality of state coordination equipment sets to determine the number of polling control hosts of the state coordination equipment sets; and segmenting the state collaborative device set to obtain a plurality of target collaborative device sets, determining a target polling priority parameter and a global monitoring sorting reference list of each target collaborative device set, generating a polling control strategy about the plurality of slave devices, and performing polling control on the plurality of slave devices based on the polling control strategy. According to the invention, the comprehensiveness of multi-device polling data acquisition is improved.

The invention relates to the technical field of signal transmission, and discloses a field virtual bus signal transmission method and system for power transmission tower grounding measurement, and the method comprises the steps: collecting the state data of each first object, and building a distribution architecture; monitoring the state of the first object through the distribution architecture; the status of the first object is transmitted to the second object. And efficient and real-time monitoring of the grounding signal of the wide-area power transmission tower is realized through a distributed architecture. The bottleneck of a centralized architecture is greatly reduced by distributed data acquisition and processing, and the response speed of the system is improved. Meanwhile, the encryption transmission module ensures the safety of the data in the transmission process, so that the monitoring center can stably and timely receive the state information of each tower, especially under the abnormal condition, the early warning information can be quickly transmitted, and the safety and reliability of a power grid are improved.

The application provides a UWB positioning method, system, electronic equipment and computer readable storage medium, aiming at improving the problems of channel congestion, limited capacity and insufficient scalability of the existing UWB system in large-scale deployment and multi-anchor cooperative ranging. The method part comprises: based on the local tag lookup table, a ranging request is sent to each to-be-ranged tag in the local tag lookup table in turn to obtain ranging information corresponding to each to-be-ranged tag; after the ranging request is sent to the local tag lookup table, a networking request is broadcasted, and the address of the tag sending the networking reply is added to the local tag lookup table based on the received networking reply to update the local tag lookup table; the obtained ranging information is reported to the positioning server, and the next target anchor point in the network is determined according to the local anchor lookup table, and the ranging right is handed over to the next target anchor point.