258results about How to "Reduce resource consumption" patented technology

The invention discloses a prediction control method based on a world model and online fine tuning, a medium and equipment, and belongs to the technical field of control algorithms. The method comprises the following steps: acquiring multi-modal environment observation data, wherein the environment observation data comprises point cloud data, an RGB image and IMU attitude angle information; inputting the environment observation data into a trained variational auto-encoder network, and predicting a potential state; for the predicted potential state and candidate action sequence, predicting a potential state sequence and a decoded occupation grid map of subsequent steps by using a trained cyclic state space model; and for a current action and a plurality of candidate action sequences, taking minimization of a total cost function of each track as an optimization target, and searching an optimal action sequence in an action space. According to the invention, the prediction efficiency and accuracy are improved.

The invention relates to the technical field of computer networks, in particular to a data security access method based on a service side. A primary access request sent by a client is converted into a secondary access request through a public network server, the secondary access request is sent to a message queue module, the message queue module sends the secondary access request to a sandbox module for compliance inspection, and if the compliance inspection is passed, the sandbox module sends the secondary access request to the message queue module; if yes, the sandbox module sends the second-level access request to the mirror image server, the mirror image server generates a feedback response according to the received second-level access request and sends the feedback response to the message queue module, and the message queue module sends the received feedback response to the public network server; the public network server sends the received feedback response back to the client; according to the invention, the problems of poor security, large resource consumption and the like in the prior art can be effectively solved.

The invention relates to the technical field of network acceleration, and discloses an edge-end large language model reasoning acceleration method and accelerator, and the method comprises the following steps: reconstructing a calculation process of a decoding stage, and carrying out the deep fusion of a multi-head attention mechanism and the calculation operation of a feedforward network; the weight and key value data are stored in HBM, and the coefficient and the accumulated attention score are stored in DDR; for linear matrix calculation, a unified matrix calculation unit is used for executing multi-precision matrix operation; for nonlinear function calculation, a mathematical transformation and linear fitting method is adopted, a Softmax function is converted into operation with 2 as the bottom through a bottom conversion formula, and truncation and third-order linear fitting are conducted on a Sigmoid function; and constructing a key value screening algorithm based on the accumulated attention score, dynamically adjusting a key value storage position, maintaining a recent key value cache region and an important key value cache region in a limited cache space, and realizing key value efficient cache in long text reasoning.

The invention discloses a large model reasoning hardware accelerator based on data flow execution, and belongs to the technical field of large model reasoning hardware acceleration. Each node comprises a linear calculation kernel used for sequentially completing query / key / value projection and attention output projection in multi-head attention calculation, dimension raising projection and dimension reduction projection calculation in a feedforward neural network layer, and inverse quantization operation after each projection is output; the multi-head attention calculation kernel is used for completing loading and quantization of query / key / value vectors, multi-head division and cache management, attention score calculation, Softmax normalization, context vector generation and KV cache updating; the RLA kernel is used for completing residual addition, layer normalization, quantization / inverse quantization and nonlinear activation calculation of each layer; the three cores in the same node are located in the same SLR area, and data transmission is carried out through a hardware queue and an HLS blocking mechanism. According to the invention, high hardware utilization rate and low-delay reasoning can be realized.

The invention discloses a session management method and device, computer equipment and a storage medium. The method comprises the following steps executed by a participating node: receiving a first connection establishment request sent by a coordinating node when the coordinating node is switched from a fault state to a normal state, and obtaining a current state of the participating node; when the current state of the participating node is a normal state, inquiring a participating node session pool by adopting a global unique session identifier, and judging whether a historical local session corresponding to the participating node exists or not; when the historical local session corresponding to the participating node exists, verifying the historical local session based on the version information, and determining and executing a current local session corresponding to the participating node; and when the historical local session corresponding to the participating node does not exist, creating and executing a current local session corresponding to the participating node based on the session parameter information. According to the method, the dependence degree of the participating nodes on the coordination nodes can be reduced, the fault processing complexity and the system resource consumption are reduced, and the fault processing efficiency is improved.

This invention discloses a construction method for cast-in-place box girder spanning an existing overpass, comprising the following steps: Step 1: Construction preparation stage, including foundation treatment, formwork processing, and steel reinforcement processing; Step 2: Erection and reinforcement of full-span scaffolding. First, the scaffolding is erected. After the scaffolding is erected, the bottom formwork and outer side plates of the box girder are installed. Then, the scaffolding is pre-stressed. After unloading, the elevation of the bottom formwork is adjusted as needed. Finally, the inner side formwork and the bottom formwork of the top plate of the box girder are installed; Step 3: Casting of the box girder. After the entire cross-section of the box girder is cast, it is cured. Tensioning is performed when the concrete strength reaches 85% of the design strength, and finally, grouting is carried out; Step 4: Dismantling of the scaffolding. In this invention, disc-lock full-span scaffolding is used to reinforce existing highway bridges. Disc-lock full-span scaffolding is then erected on top of the existing highway bridge as cast-in-place scaffolding. Installation and dismantling are convenient and can save construction time.

This disclosure relates to a data query method, apparatus, electronic device, and storage medium, comprising: acquiring a query request, the query request including a first query condition and a query target, wherein the first query condition includes a first vehicle speed and a first torque; determining a second query request based on the query target, the second query request including a second query condition, wherein the second query condition includes a second vehicle speed and a second torque, the second query request being a query request from a previous moment corresponding to the query request; and determining a query result based on the second query condition and the first query condition in the query request, thereby improving the efficiency of data query.

The application discloses a kind of multilayer superimposed cylinder suction type vortex hydroelectric generation device systems, belong to hydroelectric generation equipment technical field, it is related to new clean energy innovation technology.The system uses multistage multilayer superimposed lever mechanics structure, the layer of power generation unit, interval distance and water turbine impeller diameter are within the scope of the patent protection;Based on fluid mechanics and Bernoulli principle, construct by high to low by wide to narrow "tapered flow guide-vortex acceleration-suction gain "tornado type "vortex flow technical scheme, three-stage potential energy conversion system, by continuity equation, vortex dynamics etc. Theory support, realize the efficient progressive conversion of water flow potential energy to kinetic energy.Single power generation unit is composed of vortex generator, mixed-flow water turbine impeller and vertical (vacuum) siphon pipe, through large-diameter water turbine multistage multilayer superimposed lever amplification effect, realize single water potential energy after superposition times conversion and multiple reuse, adapt to multi-field hydroelectric generation scene.The application deeply integrates the core principles of multi-disciplinary and new energy innovation highlights, solves the problem of low efficiency, low power expansion, low energy utilization rate of traditional water turbine single-layer device, and has significant energy-saving and emission-reducing benefits, social value and economic benefits.

The invention discloses a wild barley HbAUX1 gene and application, belongs to the technical field of plant genetic engineering, the nucleotide sequence of the HbAUX1 gene is shown as a sequence 1, the invention discloses application of the HbAUX1 gene in plant dwarfing and / or plant seed yield increasing, and contributes to cultivation of excellent high-yield dwarfed plants.

The present disclosure relates to a hierarchical calibration lightweight crystal frequency offset estimation and adaptive compensation method, aiming to solve the problems of precision limitation, slow convergence of frequency offset estimation, lack of long-term stability guarantee and high computational complexity of pure software scheme in existing technology when dealing with crystal frequency offset. The scheme designs a "short time + long time" hierarchical calibration lightweight crystal frequency offset estimation and adaptive compensation scheme with the help of the system clock and hardware interrupt mechanism of the chip, and considers the frequency offset estimation of the offline situation. Based on CH585 chip, the scheme can control the equivalent frequency offset to below 0.02ppm within 10 seconds, and the 10-second cumulative error is less than 200ns, which can meet the requirements of nanosecond-level long-term synchronization of distributed wireless system. Without additional DPLL circuit or high-precision temperature compensation crystal oscillator, the hardware cost can be reduced by more than 80%, the power consumption can be reduced by more than 50%, and it is especially suitable for large-scale sensor networks and device deployment with limited resources.

Embodiments of the present application provide a kind of based on graph's alarm event determination method, device, electronic equipment and storage medium, method includes: the security log is converted into the target graph information corresponding to the security log;Based on the target graph information and the graph rule configuration information obtained in advance, generate first graph query statement;The graph rule configuration information at least includes: the rule for generating the first graph query statement;The graph information matching operation is carried out between the first graph query statement and the graph information in graph database;The graph database is used to store at least one kind of graph information;According to the graph information matched to the first graph query statement, generate alarm event.The embodiments of the present application realize the association of security log before and after and the association between multi-source security log by graph database, can effectively reduce false alarm rate;In addition, based on the high efficient query ability of the association relationship of graph database itself, the low time and low resource consumption of the association analysis of multiple security logs can be realized.

The invention provides an agent communication method, a readable storage medium and electronic equipment, and is applied to the technical field of software. The method comprises the following steps: when a target message sent by a source agent in a plurality of agents is received, caching the target message into a message queue, extracting the target message from the message queue, determining a target agent for processing the target message from other agents based on the extracted target message, and sending the target message to the target agent. In the method, each agent communicates through a message structure and a communication interface, so that each agent can indirectly communicate through a communication framework without protocol conversion, protocol conversion in a communication process can be avoided, consumption of hardware resources and time is reduced, and the communication efficiency is improved. Therefore, the running efficiency of a software system can be improved.

The application discloses a deep foundation pit support dynamic decision method and system based on reinforcement learning, and relates to the technical field of reinforcement learning; the technical points are as follows: building information model data of a deep foundation pit project is acquired; a space-time observation feature matrix corresponding to a current construction stage is constructed; the comprehensive uncertainty scores of each space-time node are determined, and a weighted space-time observation feature matrix with enhanced credibility is obtained; the weighted space-time observation feature matrix with enhanced credibility is input into a decision network based on a deep deterministic policy gradient which is pre-trained, and a continuous initial support adjustment action vector is output by an executor network in the decision network; a reference yield strength limit matrix of a field support component is acquired, and a target support structure internal force adjustment vector is obtained; based on the target support structure internal force adjustment vector, corresponding field support parameter adjustment instructions are generated; and the application improves the efficiency of deep foundation pit support regulation and control under complex working conditions.

The invention provides a magnetic resonance compatible multispectral fluorescence mesoscopic imaging system, which comprises a fixing device located in a magnetic resonance cavity of magnetic resonance equipment and used for fixing an experimental object to be observed; the optical equipment comprises a light emitter, a light collector and a light conduction assembly, the light emitter is used for emitting light rays of at least two different wave bands, the light rays emitted by the light emitter are used for exciting fluorescence in an experiment object, and the light emitter and the light collector are both located outside the magnetic resonance cavity; and the computer is used for producing images. The system can realize synchronous acquisition of magnetic resonance imaging and multispectral mesoscopic fluorescence imaging when an experimental subject is in a waking state.

The application relates to a method and system for realizing 90-degree integer multiple rotation of an image, wherein the method comprises the following steps: an FPGA receives original image data output by an image sensor and image processing parameters issued by an ARM, the image processing parameters at least including width, height and a rotation angle, wherein the rotation angle is an integer multiple of 90 degrees; the original image data is preprocessed and demosaiced to obtain a full-color image in RGB or YUV format; according to the width, height and rotation angle of the image, a coordinate transformation formula for 90-degree integer multiple rotation is used to determine the position mapping relationship between each source pixel in the full-color image and a target pixel after rotation; the full-color image is written into a BRAM cache in a ping-pong mode, and each pixel is read according to the order of the position mapping relationship and written into a target address in a DDR, wherein the target address in the DDR is calculated according to the position relationship.

The invention relates to a PLC data acquisition control method and device for an AI exhibition hall, equipment and a medium, and the method comprises the steps: executing an acquisition control code, carrying out the statistics of a data volume index of PLC data at a current moment based on a preset first acquisition frequency, and constructing an observation data volume at the current moment according to the data volume index; according to the observation data volume at the current moment, the historical smooth data volume at the previous moment, the maximum response coefficient and the minimum smooth coefficient, calculating and determining a dynamic smooth coefficient of the data volume index; calculating and determining the smooth data volume at the current moment, calculating and determining the collection frequency adjustment amount of the first collection frequency by adopting a preset PID control algorithm, updating the first collection frequency to determine a second collection frequency, and continuing to collect the PLC data of the target interaction object by the central control system of the AI exhibition hall according to the second collection frequency; and repeating the steps until the PLC data of each target interaction object in the AI exhibition hall is acquired. According to the invention, the problems of blocking and poor real-time performance of a traditional PLC data acquisition system can be solved.

The invention relates to a data sinking method and device, equipment, a storage medium and a program product, and the method comprises the steps: determining a target continuous key range and a target layer according to a first key value of a first character string file in a first layer and a second key value of a second character string file in the second layer in a process of sinking data from the first layer to the second layer; all the first data blocks in the target continuous key range on the target layer are copied to a third character string file, and all the second data blocks outside the target continuous key range are written into the third character string file. According to the method, the target continuous key range is recognized, and the direct copy operation is executed on the first data block in the range, so that calculation-intensive operation such as analysis, comparison and recoding on each key value pair in the traditional process is effectively avoided, resource consumption in the merging process is remarkably reduced, and the merging efficiency is improved. And the data sinking efficiency is greatly improved.

The invention provides a network node processing method and device, electronic equipment and a computer program product, belongs to the field of network security, and aims to solve the problem that the existing method is difficult to quickly sense and respond to network security threats. The method comprises the following steps: preprocessing each node in a network, and determining candidate network nodes; identifying and evaluating network nodes which are not preprocessed in the network, and determining supplementary network nodes; taking the candidate network node and the supplementary network node as target network nodes; and obtaining target data of the target network node, analyzing the target data, and determining a network security risk according to an analysis result.

The application provides a cache-restricted LT code degree distribution optimization method based on reinforcement learning, comprising the following steps: S1, generating a source symbol sequence with a length of k; S2, constructing a reinforcement learning agent for determining a degree distribution of a transmitting end, the reinforcement learning agent is composed of a 2-node input layer, a 2k-node hidden layer and a k-node output layer; S3, updating a current degree value through the reinforcement learning agent, and performing LT coding on the degree value and the source symbol by an encoder to obtain a coded symbol and send the coded symbol to a receiving end decoder; S4, judging the degree value of the received coded symbol according to the currently recovered source symbol by the decoder, and performing a cache or decoding according to the degree value; S5, training the reinforcement learning agent by using the reinforcement learning environment in step S3, and finally obtaining a degree distribution that makes the reward tend to be maximum through iterative training. The application can obtain a degree distribution with smaller redundancy overhead required for full recovery under the condition of cache restriction.

This application discloses a device management method and an electronic device, relating to the field of computer technology. The device management method includes enabling the configuration parameters to be applied to the hardware parameter layer without requiring a configuration file layer or device restart when the storage protection function is switched to an enabled state. This reduces the need to restart the electronic device when enabling the storage protection function. Furthermore, it retains default parameters from the configuration file, so that when the storage protection function is disabled and the electronic device is restarted, the hardware parameter layer configuration can be restored using these default parameters, further reducing the restart frequency of the electronic device. This solves the technical problem of requiring the restart of the electronic device during storage management, achieving the technical effect of reducing the restart frequency of the electronic device to improve the business sustainability of the electronic device, thereby improving the performance of the electronic device.

The invention provides a method and system for monitoring the health state of server equipment, and relates to the technical field of server monitoring. The system comprises a data acquisition module, a message processing module, an event distribution module, a log conversion module, a mapping table module, a health state management module and an alarm module. According to the method, data and SEL logs are collected in parallel through multiple threads, events are converted into uniform IDL logs, the uniform IDL logs are associated to specific hardware equipment through a mapping table, the sub-health state, the overall health state and the type summarized health state of the equipment are updated based on the alarm level, and accurate alarms are generated when the equipment is abnormal. According to the invention, through the design of the multi-level health state architecture and the mapping table, the accurate positioning of the fault is realized; the monitoring real-time performance is guaranteed through multiple threads and a priority message queue; the resource consumption is reduced through the differentiation strategy, the method is suitable for large-scale server cluster deployment, and the problems that an existing monitoring mode is poor in real-time performance, inaccurate in positioning and the like are effectively solved.

This invention discloses an efficient QR matrix decomposition method based on FPGA for implementing systolic matrices, belonging to the field of software-defined radio signal transmission technology for airborne networks. It is based on the design of a multi-channel digital front-end implemented on an FPGA chip. The method includes: designing a general handshake coordination mechanism, designing a timing flow for systolic matrix multiplication, performing fast QR matrix decomposition single-loop operations by calculating the Householder mirror transformation matrix, designing a top-level loop pipeline for QR matrix decomposition, and finally verifying the hardware scheme of the fast QR decomposition algorithm. This invention converts resource consumption into timing consumption by designing the overall pipeline logic and loop operations, enabling fast processing and accurate decomposition of digital information in a multi-channel digital front-end on an FPGA, effectively saving on-chip resources. In addition, this invention has excellent scalability, supporting 64 to 128-bit high-precision fixed-point number formats, and the data matrix order can reach up to 128.

The application belongs to the technical field of application software development test, and discloses an application software development test system with real-time vulnerability detection, which comprises a code real-time collection module, a multi-dimensional vulnerability detection module, a vulnerability accurate positioning module, a vulnerability risk quantitative evaluation module, a dynamic repair guidance module, a data storage module, a visual interaction module and an iterative optimization module, and each module cooperates to form a whole-process closed-loop vulnerability detection and management and control system. The application software development test system with real-time vulnerability detection is adopted, real-time capture, accurate positioning, risk evaluation and dynamic repair suggestion output of the vulnerability are realized, and the software development test efficiency and application program security are improved.

This invention discloses an AI-based physical education exam scoring method, belonging to the field of exam scoring methods, including the following steps: (1) activating the cameras deployed in the exam room to collect motion images and facial expression images of the examinee during pull-ups; (2) analyzing the collected facial expression images using AI image recognition technology to divide the exam process into three stages; (3) for each stage, analyzing the examinee's motion images using AI technology to determine whether each pull-up motion is a qualified motion; (4) counting the number of qualified motions in each stage and scoring them separately, adding the scores of the three stages to obtain the examinee's final exam score. This invention can divide the exam into three stages based on facial expression recognition, perform differentiated analysis on the motion characteristics of each stage, better fit the examinee's physical fitness change patterns, and make the evaluation results more targeted.

The present application discloses a kind of hibiscus compound extract, its preparation method and application.The composite extract includes hibiscus exosome, hibiscus traditional active ingredient, snowbank flower extract and microalgae extracellular polysaccharide, wherein the hibiscus exosome and hibiscus traditional active ingredient are simultaneously extracted from hibiscus raw materials by integrated extraction process.The preparation method includes raw material pretreatment, integrated extraction, separation and purification, synergistic compounding and post-processing, and two types of active substances are efficiently and synchronously obtained by optimizing solvent, temperature, time and purification parameters.The composite extract exhibits excellent anti-aging, anti-inflammatory and moisturizing effects in cosmetics, and has excellent effects of regulating intestinal flora and enhancing immunity in functional foods.The process produces significant synergistic effect by combining specific synergistic components, and the effect is much better than single component or simple mixture, realizing the maximization of plant resources and the synergistic promotion of efficacy.

The invention relates to the field of sheet pile construction, in particular to a tensioning system based on the piling construction process of a U-shaped sheet pile. The state judgment module is used for determining the sheet pile offset state according to the offset balance degree and the dynamic offset degree and determining the interval number setting mode according to the sheet pile offset state; the period analysis module is used for determining the number of tensioning intervals according to the comprehensive deviation degree; the attitude analysis module is used for determining an attitude change state according to the inclination angle change rate and the track convergence degree of the U-shaped sheet pile; the tensioning control module is used for determining a tensioning strategy according to the posture change state; the deviation correction judgment module is used for determining fixed deviation correction force according to the pile sinking resistance variation value; the tensioning device effectively improves the tensioning effect.

The application discloses an identity authentication and key distribution method and system based on a white-box cryptographic algorithm, and the method comprises the following steps: a communication initiator generates an identity authentication token by using randomly selected white-box code and sends the token to a key management center; a session key distribution token sent by the key management center is received, the session key distribution token is generated by the key management center after the identity authentication token is verified to be correct, and the session key distribution token comprises a target round function generated based on a randomly selected session key; the communication initiator refreshes the reference code stored locally by using the target round function and the white-box code stored locally, and generates a communication session white-box cryptographic algorithm; the communication initiator sends the communication session white-box cryptographic algorithm and the session key distribution token to a communication responder for encrypted communication; and the application reduces the resource consumption of white-box code transmission and switching in the whole identity authentication and key distribution process while ensuring security.

The invention discloses a data processing method and device based on an alliance chain, equipment and a readable storage medium. The method comprises the steps of obtaining a target consensus type corresponding to a target alliance party to which a first consensus node belongs; performing consensus voting processing on at least one consensus node contained in the target alliance party according to a consensus voting rule indicated by the target consensus type; the consensus voting processing is used for electing voting representative nodes of the target alliance party from at least one consensus node contained in the target alliance party; the voting representative node of the target alliance party is used for representing the target alliance party to carry out voting consensus processing on the proposal block in the alliance chain. By adopting the method and the device, the decentralization and the fairness of the alliance chain can be improved.

The invention discloses a communication fault processing method and device for a vehicle-mounted terminal. The method comprises the following steps: periodically detecting a network registration state of the vehicle-mounted terminal; and if it is determined that the network is successfully registered, identifying connection states of the plurality of preset network access channels and obtaining a result. And based on the result, selecting a first target network access channel which does not establish effective connection within the first set duration from the channels, and performing progressive connection retry between the first target network access channel and the first target network access channel. And meanwhile, selecting a second target network access channel which does not establish effective connection within a second set time length (greater than the first set time length) from the channels, accumulating connection failure time of the second target network access channel, and sequentially executing different levels of recovery operations according to different preset connection threshold values. Besides, when the network registration state satisfies a preset abnormal condition, network injection abnormity is determined, network injection abnormity time is accumulated, and recovery operations of different levels are executed in sequence according to different preset network injection thresholds so as to guarantee stable communication of the vehicle-mounted terminal.

The application discloses a method and device for quickly evaluating a ranking model, and belongs to the technical field of computers, in particular to the field of machine learning and data mining. The method comprises the following steps: obtaining a first score list of all items, wherein the score is obtained by a ranking model to be evaluated; obtaining a second score list of related items, wherein the score can be obtained from the first score list; dividing the first score list by taking the second score list as a demarcation value, and obtaining the ranking of the related items in the first score list; and calculating a ranking evaluation index and evaluating the ranking model according to the ranking of the related items. The application solves the problems of large calculation complexity and high calculation cost in the existing method for evaluating the ranking model, and improves the evaluation efficiency. By using the method, the ranking model can be quickly and accurately evaluated on the order of 100 million items by using a small amount of calculation resources, and the cost, resources and time are saved.