32 results about "Adaptive hardware" patented technology

The invention discloses a computing power assessment and resource allocation method for large language model reasoning. The computing power assessment and resource allocation method comprises the steps that S101, a large language model to be assessed is input; step S102, analyzing the large language model, and identifying a virtual operator; step S103, analyzing the operator power demand of the virtual operator; step S104, generating a computing power use judgment table; s105, collecting the hardware capability of the heterogeneous GPU pool in real time to obtain heterogeneous hardware capability parameters; step S106, carrying out standardization processing on the heterogeneous hardware capability parameters; step S107, mathematically matching the computing power demand with the hardware capability; s108, dynamically recommending and selecting an optimal adaptive hardware combination; and S109, starting the large language model according to the recommended optimal adaptive hardware combination. Through a set of dynamic evaluation mechanism, computing power resources required by the model are accurately predicted, and the most suitable hardware combination is intelligently recommended and selected from the heterogeneous GPU resource pool according to the computing power resources.

The invention relates to an encryption transmission method and device, equipment and a storage medium, and relates to the technical field of communication. The method comprises the following steps: acquiring plaintext data to be transmitted, and partitioning the plaintext data to obtain plaintext partitioned data; under the condition of supporting a hardware encryption instruction set, encrypting the plaintext block data by adopting a hardware encryption channel and / or a software encryption channel based on a derived public key in the derived session key pair, and adding an encryption channel identifier in the encrypted ciphertext block data; the encryption channel identifier is used for representing that the adopted encryption channel is a hardware encryption channel or a software encryption channel; and sending the ciphertext block data to a receiving end, so that the receiving end performs decryption by adopting an adaptive hardware decryption mode or software decryption mode based on a derived private key in the derived session key pair to obtain plaintext block data. Encryption is carried out by polling a hardware encryption channel and a software encryption channel, and CPU resources consumed in the data encryption process are effectively relieved.

The invention provides a self-service terminal hardware service driving method, system and device for achieving cross-platform self-adaption and a medium, and belongs to the technical field of hardware service driving. The method comprises the steps that a front-end application is received to generate a hardware service request; the service bus dynamically loads a corresponding hardware service plug-in from a plug-in library according to the target hardware identifier; the interface adaptation layer selects a protocol template according to the interface type, and injects an operation parameter into the selected protocol template to generate an intermediate instruction; the loaded hardware service plug-in converts the intermediate instruction into a platform-independent hardware instruction; the platform adaptation layer detects the type of a current operating system, and calls a corresponding bottom driver library to convert a platform irrelevant hardware instruction into an operating system level hardware instruction; and sending an operating system level hardware instruction to the physical hardware, packaging a return result into a uniform response format, and returning the return result to the front-end application. According to the invention, cross-platform adaptive hardware driving is realized, and compatibility, flexibility and maintenance efficiency are improved.

The invention provides a software transplantation method based on a processor. The software transplantation method comprises the following steps: firstly, carrying out hardware type selection and instruction set compatibility analysis, analyzing a software binary code by utilizing a specific algorithm, counting and comparing instructions, and comprehensively evaluating hardware performance to determine adaptive hardware. Then, software transportability is evaluated, and a software dependency graph and a transplantation feasibility prediction model are constructed; then, an operating system is adapted, the loading sequence and parameters of a kernel module are dynamically adjusted, and setting of a basic input and output system is optimized; in a software transplantation implementation stage, source codes are analyzed and converted based on an abstract syntax tree, open source software adopts incremental compiling cache, binary software analyzes and processes compatibility problems, and meanwhile, a one-key installation program is optimized. And finally, performing server monomer testing and platform testing, monitoring performance, checking functions, analyzing faults and optimizing until the system meets requirements. According to the method, many key problems in the process of transplanting the software to the processor are effectively solved, and the success rate of transplanting and the stability of the system are improved.

The invention discloses a metering detection intelligent glasses system integrating artificial intelligence visual identification and a cloud platform, and relates to the technical field of computer software, the metering detection intelligent glasses system comprises an intelligent glasses driving module and adaptive hardware, and has the functions of dynamic scheduling, multi-thread processing, communication self-adaption and non-lagging AR rendering; the AI visual identification module supports batch identification, fine defect detection, off-line operation and incremental learning by using a lightweight model; the cloud platform software framework module adopts a distributed micro-service framework to realize data partition storage, real-time parallel analysis and the like and elastic capacity expansion and contraction; the interaction control module develops a multi-mode program and supports batch instructions and the like; and the system collaboration module realizes multi-module scheduling based on event driving, and ensures efficient collaboration of the whole detection process. According to the method, the precision, efficiency and environmental adaptability of measurement detection are remarkably improved, the batch processing capacity and the network fluctuation coping capacity are greatly enhanced, and efficient and reliable software support is provided for measurement detection.

A system and method for an adaptive hardware system. The adaptive hardware includes rail segments and reconfigurable anchor assemblies. The anchor assemblies are secured onto different portions of the rail segments at anchor installation ports. The anchor assemblies are adaptable for accepting common fastening elements (e.g., bolts), such that interior monuments and fixtures may be accommodated without interior aircraft design changes.

An adaptive hardware safety disconnect system for a motorized traffic vehicle may utilize configurable hardware logic circuitry to selectively disable a traffic vehicle system, such as a high voltage contactor, an airbag, or a low voltage power supply, in a deterministic manner. The first disconnect mechanism may disable the first set of systems, while the second disconnect mechanism disables the additional system to fully power down the traffic vehicle. The system may be configured for different use cases, such as emergency response, service, and development testing, while using common hardware components.

The invention provides a flash memory data packet writing method and device, and relates to the field of flash memory data packet writing and garbage collection optimization. The method comprises the steps of obtaining file features, user behavior features and system metadata features of to-be-written data, and predicting a life cycle frequency band of the data based on multi-dimensional features; according to the predicted life cycle frequency band, aggregating data with similar failure characteristics into data groups, and matching the data groups to corresponding physical storage blocks; writing the data in the data group into the same physical storage block adaptive to the hardware specification so as to realize high synchronization of data failure time; and when the valid data volume in the data group is lower than a preset threshold value, triggering a garbage collection mechanism, and directly executing a whole block erasing operation on the physical storage block to avoid valid data migration. According to the method provided by the embodiment of the invention, write amplification and system delay can be effectively reduced, the service life of the flash memory storage system is prolonged, and the storage performance and the resource utilization rate are improved.

The invention discloses an open circuit determination method and device for a compensation resistor of a broadband oxygen sensor and a medium, and relates to the technical field of electronic fault diagnos.The method comprises the steps that under the condition that it is determined that an obtained diagnosis enabling mark is effective, connection between a target chip and the broadband oxygen sensor is controlled to be disconnected, and a first current value of the target chip at the first moment is obtained; after a first preset duration, the target chip is controlled to be connected with the broadband oxygen sensor, and a second current value of the target chip at a second moment is obtained; and calculating a current difference value between the first current value and the second current value, and when the current difference value is determined to be smaller than or equal to a preset fault threshold value, determining that the broadband oxygen sensor compensation resistor has an open-circuit fault. According to the application, a series of problems occurring when an adaptive hardware circuit is adopted to solve the problem of a chip without a compensation resistor diagnosis working condition in the prior art are solved, and the open-circuit diagnosis of the compensation resistor of the broadband oxygen sensor is efficiently and accurately realized by utilizing a software algorithm.

An adaptive hardware safety disconnect system for electric vehicles can use configurable hardware logic circuits to selectively and deterministically disable vehicle systems such as high-voltage contactors, airbags, or low-voltage current. A primary disconnect mechanism can disable an initial set of systems, while a secondary disconnect mechanism disables additional systems to completely shut down the vehicle. The system can be configured for various use cases, including emergency response, maintenance, and development testing, using common hardware components.

The application provides an intelligent detection system of an RGB camera combined with a deep learning algorithm, comprising a defect feature adaptive engine, a hardware component, an algorithm component and a feedback component; the defect feature adaptive engine serves as a core control unit, links the hardware component, the algorithm component and the feedback component to realize full-link closed-loop detection; the defect feature adaptive engine performs the following operations: identifying scene working conditions and core working condition pain points, extracting the subdivided features of target defects in the scene and matching the feature label library, dynamically calling the adaptive hardware configuration, algorithm combination and preprocessing strategy, optimizing the detection parameters in real time, allocating the computing power resources according to the defect risk level and triggering the corresponding early warning, realizing the intelligent identification, positioning, quantification and grading early warning of defects of equipment such as pressure pipelines and pressure vessels, and through the implementation of the above system, the problems of poor adaptability, insufficient precision and unreasonable resource allocation of the existing system are solved, and the industrial detection is realized in a scene-based, refined and efficient manner.

The present application relates to the technical fields of intelligent wearable device and health monitoring, in particular to a health detection control method of intelligent wearable watch, comprising: multi-source data acquisition step: synchronously acquiring physiological indexes and situation indexes, and acquiring apparent physiological data and situation characteristic data in real time through a sensor network; situation semantic analysis step: based on situation characteristic data analysis, calculating situation metabolic equivalent expected value through a motion dynamics model; compensation evaluation solving step: extracting heart rate variability and recovery rate characteristics, calculating the difference between actual metabolic level and situation metabolic equivalent expected value to obtain compensation deviation degree, and comprehensively generating compensation exhaustion index; adaptive hardware control step: dynamically generating control instruction sequence based on the numerical interval of compensation exhaustion index, and adjusting the working state of the sensor network; the present application significantly reduces the overall power consumption of the system, eliminates invalid false alarms, and improves the monitoring specificity and robustness in complex application environments.

The application discloses a kind of based on radio frequency and communication circuit calibration system, the present application relates to wireless communication transmission technical field, calibration system includes following module: environment perception module, intelligent algorithm module, adaptive hardware module and interactive verification module, the advantages of the present application are: reinforcement learning agent and cross-domain transfer learning model collaborative optimization calibration parameter, combine multi-modal environment perception data dynamic adjustment radio frequency circuit center frequency and impedance matching value, programmable hypersurface array is based on electromagnetic property reconstruction and realizes millisecond level beamforming, FPGA real-time loading optimization parameter, calibration cycle is greatly shortened, AR interactive interface is simplified artificial operation by high-precision gesture mapping and high-frequency calibration technology, overall calibration efficiency is improved by more than 40%, bit error rate is reduced to extremely low level, meet the real-time demand of 5G / 6G high-frequency communication, break through the efficiency bottleneck of traditional artificial debugging.

The embodiment of the invention provides a self-adaptive hardware abstraction layer configuration method and system based on hardware feature fingerprints, and belongs to the technical field of computers. The method comprises the following steps: collecting multi-dimensional feature information of hardware equipment; constructing a hardware feature fingerprint based on the multi-dimensional feature information; determining an HAL strategy according to the hardware feature fingerprint; determining interface configuration of each hardware device according to the HAL strategy; and packaging the hardware equipment according to the interface configuration. According to the method and the system, hardware feature fingerprints are captured from multi-dimensional feature information of hardware equipment, then unified interface configuration of the hardware equipment is determined according to a strategy matching mode, and finally packaging is performed based on the interface configuration, so that unified interface standardized operation of different hardware equipment is realized, and the operation efficiency is improved. Therefore, the compatibility of hardware is improved.

The invention provides a system and a method for direct connection interaction between a browser front end and hardware, relates to the technical field of Web front end and hardware communication, is based on a special browser developed by CEF, and realizes direct connection interaction between the browser front end and target hardware by utilizing a three-layer architecture deployed on the special browser, and comprises the following steps: initiating call requests adapted to different hardware; according to the hardware type, an adaptive hardware plug-in is loaded, and the calling request is converted into hardware protocol format data; and sending the hardware protocol format data to target hardware, returning result data to a hardware interaction layer after the target hardware executes specific operation, and analyzing to obtain structured data. According to the method, the browser function is expanded through the CEF, the three-layer architecture of the front-end adaptation layer-the core control layer-the hardware interaction layer is constructed, direct communication between the front end and hardware is achieved, meanwhile, safety verification, plug-in management and an automatic parameter conversion mechanism are integrated, and the method has the advantages of being high in stability, high in adaptability and good in safety.

The invention aims to provide a data storage method based on a life cycle strategy and a log auditing system applying the method. According to different attributes of index nodes to which the data belong, respectively configuring adaptive hardware and life cycle management strategies, and carrying out distinguishing processing on data reading, writing and checking; and automatic data archiving, deletion, backup and recovery strategies are configured, so that the data processing efficiency can be effectively improved, system resources are saved, and the method has positive significance in improving log data storage and auditing.

The invention provides a programmable logic controller based on a drag node. The method mainly comprises the following steps that S1, node dragging and node line connection can be conducted through a visual interface, parameters are set for corresponding nodes to finally complete writing of the whole control process, and the control process can be exported and stored as a file and imported to be regenerated; s2, process nodes with adaptive hardware and a visual interface are used for compiling man-machine interaction and an actual control process; s3, a set of complete downloading and uploading programs matched with the interface end and the hardware end can download the edited control flow into the logic controller hardware, and can upload the currently running control flow to an editing interface from the logic controller hardware; and S4, the logic controller hardware realizes the operation of the edited process, and realizes IO digital quantity control. The controller is provided with two 485 ports to realize a Modbus RTU slave station function, and can realize a Modbus RTU master station function by dragging node setting parameters; and S5, logic controller hardware has a 5-24V wide voltage power supply range to adapt to most of household and industrial power supply scenes, and a special external power supply and communication expansion module is designed for the serial port screen.

The invention discloses a rocket launching simulation interactive display system, and the system comprises a break angle self-adaptive hardware module which is used for building an interactive hardware foundation and forming a hardware carrier adaptive to multiple visual angles; the model selection-knowledge linkage touch control module is used for realizing cooperation of operation and science popularization, constructing a three-layer interface on a touch screen, calculating combination rationality through an algorithm after all model selection is completed, and generating an interaction instruction; the working condition adaptive naked-eye 3D module is used for generating a customized 3D scene and generating naked-eye 3D display content adaptive to the model selection based on an LED screen bevel structure and a digital rendering technology; and the double-end scene collaboration module is used for coordinating double-end content synchronization, associating the interface state of the type selection-knowledge linkage touch control module with the display content of the working condition adaptive naked eye 3D module, and generating a double-end collaboration interaction effect. According to the embodiment of the invention, rocket launching simulation interactive display with multi-view adaptation, operation and science popularization collaboration, scene dynamic adaptation and double-end synchronization can be realized.

According to the dynamic coverage area-CPU load balancing scheduling method oriented to the low-orbit satellite system, multi-dimensional load evaluation and dynamic family migration are adopted, the CPU load state can be sensed in real time, continuous coverage areas can be accurately divided, the overload CPU load is rapidly transferred to idle CPUs, overload of a computing power unit is effectively avoided, and the resource utilization rate is increased; adjacent coverage areas are preferentially aggregated to the same CPU, so that the cross-CPU interference coordination overhead is reduced, spectrum multiplexing is supported, the system interference complexity is reduced, and the spectrum utilization rate is improved; the user context is kept complete in the migration process, and a cooling mechanism is set, so that service interruption and frequent migration jitter are avoided, the QoS stability is guaranteed, and the user communication experience is improved; the method supports periodic global load rebalancing and preferentially recovers computing power resources with the lowest load, so that idle CPUs can be dynamically closed, hardware architecture and track characteristics can be self-adapted, energy efficiency optimization is achieved, and the method has high expansibility and compatibility.

The invention provides an operator lightweight design method, device and equipment for a humanoid robot and a medium, and relates to the technical field of robots. According to the method, through scene modeling and quantitative analysis, the minimum necessary set of operator functions is extracted from the interaction behavior data, and function redundancy is avoided. And according to the operator demand document, function cohesion analysis and independent function analysis are performed on the operator, so that the system complexity is reduced. Under the constraint of the modular architecture framework, algorithm redundancy reduction is carried out on each operator function module, and algorithm redundancy is reduced. Data interaction parameters are optimized and adjusted, and data transmission delay and redundancy are reduced. Physical parameter and functional performance verification is carried out in an adaptive hardware platform, and the performance and stability of system operation are ensured. Technical document data and configuration information are packaged into a reusable operator lightweight design scheme, so that reuse and iteration of the scheme are facilitated, and the reliability and sustainability of the humanoid robot in a medical care application scene are further improved.

The invention relates to an AI inference acceleration architecture and a security isolation method, and relates to the technical field of industrial real-time control, and the architecture ensures the real-time performance of a key task through hardware-level time window allocation, dynamically reconstructs the computational logic of a field programmable gate array based on load characteristics, and improves the security of the field programmable gate array. Physical-level security isolation of the model weight is realized in a trusted execution environment constructed by a kylin operating system and a boron hardware platform; the security isolation method comprises the following steps: dividing an encrypted memory area in a kernel layer of an operating system, implementing real-time data protection by a hardware encryption engine, and blocking unauthorized access in combination with memory access control and a physical protection mechanism; according to the invention, collaborative breakthrough of deterministic real-time response, adaptive hardware optimization and multi-level safety protection of artificial intelligence reasoning in an industrial control system is realized, and autonomous and controllable technical support is provided for intelligent upgrading of high-end equipment.

An adaptive hardware safety disconnect system for electrified vehicles may utilize configurable hardware logic circuits to selectively disable vehicle systems, such as high voltage contactors, airbags, or low voltage power in a deterministic manner. A first disconnect mechanism may disable a first set of systems, while a second disconnect mechanism disables additional systems to fully power down the vehicle. The system can be configured for different use cases such as emergency response, servicing, and development testing while using common hardware components.

The invention discloses a network card hardware function configuration method based on software definition, which belongs to the technical field of network security, and comprises the following steps: monitoring logic layer and physical layer characteristics generated when a network card processes a data stream, comparing the logic layer and physical layer characteristics with a normal behavior baseline, and triggering defense reconstruction when deviation occurs; an evolution reference sample set is generated according to the abnormal traffic, and initial hardware logic configuration is retrieved and combined from a hardware module library according to the threat type; iteratively optimizing and testing the configuration in an evolution sandbox isolated in the network card to generate a superior hardware logic configuration; extracting a dynamic session state from the current main path hardware and transferring the dynamic session state to a superior configuration; and during the operation period of the superior configuration, the idle programmable logic unit is driven to execute randomization calculation to generate coverage physical noise, so that adaptive hardware-level precise defense for unknown threats can be realized, the service continuity and the line speed processing performance are guaranteed, and meanwhile, the protection capability for bypass channel attacks is enhanced.

The invention discloses a self-adaptive hardware BFD (Bidirectional Forwarding Detection) link detection method and a self-adaptive hardware BFD link detection system, according to the scheme, millisecond-level link state detection is realized through a special BFD processing engine integrated by a switching chip, the transceiving processing of a BFD message and the maintenance of a state machine are completely completed by hardware, and the processing delay of a CPU (Central Processing Unit) is avoided; the transmission delay of historical messages is counted in real time by adopting a sliding window in the BFD link detection process, and a fault judgment threshold value is dynamically calculated and updated based on a statistical result; when packet loss is detected but the time delay is smaller than a dynamic threshold value, the BFD message sending priority is automatically improved, and when continuous packet loss occurs and the time delay exceeds the dynamic threshold value, a standby link switching mechanism is immediately triggered, so that the problem of frequent switching of a main link and a standby link caused by network fluctuation is avoided, the accuracy and reliability of network fault detection are remarkably improved, and the service life of the main link and the standby link is prolonged. And stable operation of the key service network is ensured.

The application discloses a cross-patient seizure detection method based on a graph-enhanced pulse state space model, first, the brain electrical signal data of a patient is acquired and an electrode space topology graph is constructed; then, a pulse state space model is used to extract the time sequence features of physiological channels, and a pulse graph convolution network is used to perform message aggregation between adjacent nodes in a discrete pulse domain across time steps; finally, feature aggregation is performed through a global average pooling operation at a graph level, and a two-class prediction probability is output through a fully connected classifier. The application follows the design logic of "first time and then space", the pulse state space model models the potential brain state dynamics of the time sequence change, and realizes stable long context feature processing through the cyclic updating mode of the adaptive hardware; the pulse graph neural network completes the information transmission and fusion between channels based on the electrode correlation graph, can capture the spatial dependence relationship which is crucial for the representation of seizures, reduces the calculation energy consumption, and improves the overall performance.

This invention discloses an adaptive, hardware-friendly hybrid quantization method based on deep neural networks. By analyzing the sensitivity of different network layers to quantization errors, it adaptively selects the optimal quantization strategy for each layer, thereby reducing the model's storage and computational resource consumption while maximizing model accuracy. The method includes: obtaining a full-precision deep neural network model to be quantized; extracting the weight parameters and activation parameters of each network layer; performing statistical analysis on the weight parameters and activation parameters of each network layer to construct an evaluation index for measuring quantization error; performing quantization sensitivity analysis on different network layers based on the evaluation index; adaptively determining the quantization method corresponding to each network layer based on the quantization sensitivity analysis results; performing hybrid quantization processing on the weight parameters and activation parameters according to the quantization method to obtain a hybrid precision quantization model; and completing the deep neural network inference computation based on the quantized weight parameters and activation parameters.

The invention discloses a self-adaptive hardware data prefetcher switch control method and device, and the method comprises the steps: obtaining memory access information from a high-speed cache based on an index collector, calculating a prefetch validity index of each prefetcher in stages, and transmitting the prefetch validity index to a control center, the prefetch validity index is constructed based on three parts of an access delay reduction value brought by a useful prefetch request, an access delay increase value caused by cache pollution and a conventional memory access delay increase value caused by the prefetch request; and judging according to the prefetching validity index of each prefetching device based on the control center so as to adaptively control the switch of each prefetching device. According to the invention, the on-off state of the prefetcher can be adaptively adjusted according to the prefetch validity index, and hardware data prefetch on-off control with high performance and low storage overhead is realized.