148 results about "Performance requirement" patented technology

The invention discloses a quality detection method for a safety valve spring, and relates to the technical field of spring quality detection, and the method comprises the steps: constructing an appearance detection module, and carrying out the surface defect detection and size verification of a target safety valve spring; performing performance test and performance grade evaluation on the target safety valve spring through a performance detection module; and combining a surface defect detection result, a size verification result and a performance evaluation result to carry out spring quality comprehensive evaluation, and extracting a defective finished product production number according to a quality judgment result to carry out correlation analysis and defective product traceability. The technical problem that an existing safety valve spring quality detection method is low in accuracy and reliability due to the complexity of an application environment and the diversity of performance requirements in the prior art is solved, and the purpose of detecting the quality of the safety valve spring through all-dimensional and multi-angle appearance quality detection and performance attribute detection is achieved. And the accuracy and reliability of quality detection of the safety valve spring are improved.

The invention discloses a nodular cast iron component and process design method based on an active learning strategy, and belongs to the technical field of material process design. In order to improve the prediction precision and design efficiency of material process design, the method comprises the following steps: collecting experimental data including nodular cast iron components, process, microstructure characteristics and performance in literatures; constructing a nodular cast iron component-process-microstructure characteristic model and a nodular cast iron component-process-microstructure characteristic-performance model, setting different weights for different target performances based on an optimization algorithm, and obtaining optimized nodular cast iron components and process parameters with different performance requirements by adopting a weight-based target function; and through an active learning strategy, data with relatively large uncertainty is preferentially selected for experiment and is supplemented to a data set, then model updating is performed, model prediction precision is optimized, and finally nodular cast iron components and process parameters meeting expected performance requirements are designed. The experiment frequency and cost are reduced.

The invention provides an optimal process parameter prediction method, system and device and a storage medium, and the method comprises the steps: collecting semiconductor performance test data which comprises process parameters of a tested piece and performance data of the tested piece; training a machine learning model based on the semiconductor performance test data to obtain a trained process parameter prediction model, the process parameter prediction model being used for predicting performance requirements which can be met by various process parameter combinations, and selecting a process parameter combination which can meet specific performance requirements; setting a target performance requirement, and obtaining an optimal process parameter combination meeting the target performance requirement based on the process parameter prediction model; and verifying whether the optimal process parameter combination meets the target performance requirement or not. According to the method, a traditional trial and error experiment method or a simulation calculation method based on experience is abandoned, the optimal process parameter searching efficiency is improved, and the research and development period is remarkably shortened.

The invention discloses a configuration method, system and device, a program product and a storage medium. The method comprises the steps of obtaining a large language model LLM type and demand information; generating configuration parameters based on the LLM type and the demand information; creating a test service based on the configuration parameters; performing a pressure test on the test service by using a test script to obtain a pressure test result; and adjusting the configuration parameters based on the pressure measurement result to obtain the configuration parameters meeting preset performance requirements. The resource allocation problem caused by lack of professional knowledge is effectively solved, and meanwhile, the working efficiency and the service quality are greatly improved, so that even a non-professional operation and maintenance team can easily complete a complicated LLM deployment task only by inputting the LLM type and the requirement.

The invention relates to a method, a system and equipment for predicting creep stress at a joint of a concrete structure and a medium. The method comprises the following steps: testing a concrete structure material and a concrete structure joint filling material, and constructing a joint material performance database based on test data; constructing a multi-scale mechanical model comprising a macrostructure model and a microcosmic interface model based on the seam material performance database; constructing an interface constitutive relationship based on the seam material performance database and the multi-scale mechanical model; and predicting the creep stress at the joint of the concrete structure based on the material performance database at the joint, the multi-scale mechanical model and the interface constitutive relationship. The method can fully consider the sudden change of the performance of the material at the joint and the interface effect, provides reliable technical support for the safety evaluation and maintenance of the concrete structure, and is particularly suitable for engineering application of large-span and high-rise buildings and the like with higher requirements on the long-term performance of the structure.

The invention discloses a design method of a noise reduction fairing, which comprises the following steps: determining a performance demand parameter of the noise reduction fairing, and constructing an initial noise reduction model based on a noise spectrum of a target application scene and an air duct structure defect; structural parameters are designed according to the performance requirement parameters; establishing a fluid-solid-sound coupled parameterized model of the layered composite structure and the flow guide assembly, and performing iterative optimization to generate an optimized configuration of the noise reduction fairing; a fairing prototype is manufactured based on the optimized configuration, actual measurement data are collected, simulation errors are corrected through an empirical regression model, and boundary conditions of a parameterized model are updated; the corrected noise reduction fairing is disassembled into modular assemblies, and an assembly process based on modularization is designed; in the assembling process of the modular assembly, a digital positioning system is adopted to be matched with an assembly interface; through the collaborative design of the layered composite structure and the diversion assembly and the module assembly design of the manufacturing link, the balance of broadband noise suppression and heat dissipation efficiency is realized, and the overall performance is improved.

The invention particularly relates to a method for generating Prompt based on large language model low-code programming. According to the method for generating the Prompt based on the large language model low-code programming, a Prompt overall framework is constructed according to the function requirement and the performance requirement of a project, and the Prompt overall framework is subdivided into four parts including a role, a requirement, a standard and a result; complementing and optimizing the content of the Prompt framework, and adding more detailed information capability description; and optimizing the generated Prompt, and continuously adjusting and perfecting the content according to the output result feedback of the large language model until the quality evaluation index and the effect verification index both reach self-defined threshold values so as to meet the project requirements. According to the method for generating Prompt through low-code programming based on the large language model, intelligentization and high efficiency of programming and document generation are achieved, flexibility is high, the programming output dimension is greatly enriched, the overall period of a project from demand analysis to test release is shortened, and the method has extremely high practical value.

The embodiments of the present application provide a method and device for testing the performance of a solid-state hard disk, a storage medium, and an electronic device, which relate to the field of computers, including: determining a target file, wherein the target file has the performance of a virtual function VF of a solid-state hard disk when i queues are allocated, i is an integer greater than or equal to 1 and less than or equal to N, and N is the maximum number of queues allowed to be allocated by the VF; allocating queues to each of the multiple VFs of the solid-state hard disk according to the target file, and determining the performance of each of the multiple VFs; determining whether the solid-state hard disk meets the performance requirements according to the performance of each of the multiple VFs. The above scheme solves the problem that the solid-state hard disk SSD cannot be tested well after SRIOV resources are allocated to the solid-state hard disk SSD.

The invention relates to an adaptive energy consumption control method, system and device based on a computing power chip and a medium. The method comprises the following steps: constructing an optimal prediction regression model based on multi-channel hardware state indexes of the computing power chip at different frequencies to predict performance data of the computing power chip at different frequencies; acquiring to-be-predicted data of the computing power chip under the current working load, predicting the to-be-predicted data based on the optimal prediction regression model, and acquiring a target optimal frequency meeting a preset performance index condition, the target optimal frequency corresponding to optimal energy consumption; and performing adaptive adjustment on the working frequency of the computing power chip under the current working load based on the target optimal frequency. According to the method and the device, the target optimal frequency which meets the performance requirement and is lowest in chip energy consumption can be accurately selected, the energy consumption is reduced while the chip performance is ensured, and the problems of low adjustment precision and high energy consumption of frequency adjustment of a computing chip in different working load performances are solved.

The invention relates to a vertical structure LED wafer thickness detection method and system and computer equipment. The method comprises the steps that a sample wafer is divided into a plurality of areas; calculating the initial wafer thickness corresponding to each region based on the motor movement data, obtaining the real wafer thickness of each region, training the initial thickness error optimization model based on the initial wafer thickness and the real wafer thickness, and obtaining a target thickness error optimization model; obtaining the final wafer thickness of each area of the to-be-measured wafer through the target thickness error optimization model; and calculating a thickness change value of the to-be-detected wafer based on each final wafer thickness, comparing the thickness change value and each final wafer thickness with a corresponding preset threshold value, and judging whether the to-be-detected wafer meets performance requirements or not based on a comparison result. According to the invention, the abnormal problems of scratch, fragment and pollution of products caused by manual measurement can be avoided, and the product yield can be improved.

The invention discloses a thermodynamic system parameter distribution method and device, and the method comprises the steps: S1, analyzing a block definition diagram and an internal block diagram from a SysML model to obtain a module structure description, a module value attribute, an interface specification of an equipment unit and a thermodynamic parameter description, and generating an analysis result, the analysis result comprises equipment unit port and connector information. And S2, based on the analysis result, mapping the equipment module to a simulation software element through name matching, and generating a steady-state thermal simulation model. And S3, associating the module value attributes with element attributes of the steady-state thermal simulation model through a parameter mapping mechanism, and generating attribute mapping data. And S4, generating a complementation parameter list according to the external design value and the heat balance demand. Therefore, the efficiency and accuracy of parameter allocation are improved, manual intervention is reduced, and it is ensured that the allocation result meets the requirements of system performance.

The invention discloses a request batch processing method and electronic equipment, and relates to the technical field of computers. When a batch processing object meets a first batch processing optimization condition, the current comprehensive error of the batch processing object is determined according to the performance requirement of the batch processing object in the current application scene and the current actual performance; when a first batch processing optimization condition is satisfied, carrying out first adjustment on the size of the current batch to obtain a first target batch size, and when a second batch processing optimization condition is satisfied, carrying out second adjustment on the size of the current batch according to the length change information of the request waiting queue to obtain a second target batch size; and finally, performing batch processing on the to-be-processed request according to the first target batch size or the second target batch size. The size of the current batch is adjusted according to the performance requirement of the application scene and the current actual performance expression or the length change information of the request waiting queue, so that the size of the target batch can adapt to the current application scene and the dynamically changing load, and the waste of hardware resources and the increase of user request response delay are avoided.

The invention discloses a performance testing method, device and equipment of a graphic processing unit and a storage medium, and relates to the technical field of servers. The configuration file comprises category weights of a plurality of test categories corresponding to the preset scene, test item weights of a plurality of test items and a reference running duration corresponding to each test item; running the test case corresponding to each test item on the GPU to be tested, and determining the actual running time length corresponding to each test item; according to the test item weight of each test item, the category weight of the test category to which the test item belongs, the reference operation duration and the actual operation duration, determining a comprehensive performance index of the to-be-tested graphic processing unit; and determining whether the performance of the to-be-tested graphic processing unit meets the performance requirement of the preset scene or not according to the comprehensive performance index. According to the method, the weight of each test index for evaluating the performance of the to-be-tested GPU can be adjusted according to the preset scene, and the performance of the to-be-tested GPU can be accurately evaluated.

The invention relates to a code generation method and device, electronic equipment and a storage medium. Comprising the steps of obtaining a programming problem data set and a code generated by a target model, and obtaining a data set with rewards; based on an MSE regression training strategy and a comparison loss training strategy, training a reward model by using the data set with the reward to obtain a code reward model; and obtaining programming problem training data, sorting the programming problem training data by using the code reward model to obtain sorted training data, and obtaining a final reward according to the sorted training data, a reward model score output by the code reward model and a preset unit test score, and reinforcement learning training is carried out on a preset language model to obtain a code generation model, so that codes meeting preset performance requirements are generated through the code generation model. Therefore, the problems that in the prior art, the model reasoning cost is high, and it is difficult to capture the intermediate state and partial correctness in the code execution process to evaluate the code quality in a fine-grained mode are solved.

This application is directed to resource management in a storage system that includes a non-volatile memory and a collection of resources having one or more processing cores. The storage system allocates a first subset of resources to process queues of I / O access operations requested by a host device. The first subset of resources includes a storage controller corresponding to a subset of processing cores. The storage system obtains a first request for adjusting resource allocation of the storage device, and the first request includes a target performance requirement for processing the queues of I / O access operations. The storage system determines that the target performance requirement can be satisfied by allocation of at least a target subset of resources. In response to the first request and based on the target subset of resources, a second subset of resources is allocated for processing the one or more queues of I / O access operations.

Methods of operating a storage device include receiving a first performance requirement from a first host, allocating a 0-th striping group of the nonvolatile memory devices to the first host, based on the first performance requirement, writing first data of the first host in the 0-th striping group, receiving a second performance requirement from a second host, reallocating the nonvolatile memory devices such that a first striping group and a second striping group of the nonvolatile memory devices are allocated to the first host and the second host, based on the first and second performance requirements, and performing a garbage collection operation for the first host. The second striping group includes some of nonvolatile memory devices included in the 0-th striping group, and a first part of the nonvolatile memory devices included in the second striping group stores the first data and is not selected for the garbage collection operation.

The invention provides a composite material formula process optimization method based on a performance demand, and relates to the technical field of data processing, the method comprises the following steps: obtaining matrix material information and a performance demand, and determining a missing performance demand and a missing degree thereof; determining a reinforcing material configuration relation list; configuring formula test variables; constructing a test space, obtaining a test data set, analyzing and processing the test data set, and constructing a target data set; and optimizing the target data set to obtain a formula process optimization result. The technical problem that the performance of the composite material cannot meet the required performance due to the lack of performance fusion analysis of the reinforcing material in the prior art can be solved, after the missing performance analysis is performed on the base material, the formula processing technology of the reinforcing material is optimized and analyzed, and the performance of the composite material is improved. The technical effects that the formula process accuracy of the composite material is improved, and then the adaptation degree of the performance of the composite material and the required performance is guaranteed are achieved.

The embodiment of the invention provides a design parameter generation method and device for drill bit cutting teeth, a medium and a program product. The method for generating the design parameters of the drill bit cutting teeth comprises the steps that parameters of different types of drill bit cutting teeth and the mapping relation between various performance indexes of the drill bit cutting teeth are obtained, and the parameters of the drill bit cutting teeth comprise the structure parameters and the material parameters of the drill bit cutting teeth; based on the type of the target drill bit cutting tooth, the expected performance index of the target drill bit cutting tooth, the expected performance type of the target drill bit cutting tooth, and a mapping relation between the parameters of different types of drill bit cutting teeth and various performance indexes of the drill bit cutting teeth, generating design parameters of the target drill bit cutting tooth; according to the method, multiple performance indexes are comprehensively considered, the generated parameters are more practical and targeted, and therefore it is ensured that the designed drill bit cutting teeth can meet the performance requirements of different geological conditions.

The invention provides a method for measuring the performance of a multi-stage semiconductor cooler. The method comprises the following steps: determining a calculation equation of the performance of the multi-stage semiconductor cooler; when the geometrical parameters of the multistage TEC and the material characteristic parameters of the used gradient bismuth telluride are known, the electric performance parameters of the multistage semiconductor cooler are obtained through the equation; and when the electrical performance parameters of the multistage TEC are known, the geometric parameters of the multistage semiconductor cooler and the material characteristic parameters of the used gradient bismuth telluride are obtained through reverse inference by using the equation. The method has the beneficial effects that according to the calculated electrical performance parameters of the multistage TEC, an optimized multistage TEC use scheme for the load needing temperature control can be formed; the most appropriate bismuth telluride material is selected through guidance of a calculation result, so that the purposes of effectively reducing the material cost and meeting the high-standard temperature control requirement are achieved; a rapid design method is provided, through simulation calculation and a cyclic exhaustion method, a design scheme meeting performance requirements can be found in a short time, and trial and error time and cost in traditional design are reduced.

The present application relates to the technical field of computers. Provided are a frequency adjustment method and an electronic device, which make the frequency of a GPU meet GPU performance requirements or functional requirements, thereby ensuring the usage experience of a user. In the method, a load that matches a scenario corresponding to the currently rendered image can be determined from among a predicted load and a historical load, and the frequency of the GPU is then determined on the basis of the load, wherein the predicted load can be obtained on the basis of rendering information of the currently rendered image frame. A scenario of an image frame can be determined on the basis of an application state or a system state. The greater the load matching the scenario, the higher the determined GPU frequency, and thus a GPU can process more loads, thereby improving performance, reducing stuttering during image frame display, and improving the usage experience of a user; and the smaller the load matching the scenario, the lower the determined GPU frequency, and thus loads to be processed by the GPU are minimized, thereby reducing the power consumption of the GPU.

The invention belongs to the technical field of material science, and discloses a shape memory alloy component prediction method and a related device, and the method comprises the steps: processing a target stress-strain curve of a shape memory alloy through a pre-trained WCGAN-GP network, and obtaining a plurality of groups of phase field parameters; processing the multiple groups of phase field parameters through a trained LSTM model to obtain a predicted stress-strain curve corresponding to each group of phase field parameters; calculating the similarity between the target stress-strain curve and each predicted stress-strain curve, screening the predicted stress-strain curve of which the similarity with the target stress-strain curve meets the requirement, and taking the predicted stress-strain curve of which the similarity meets the requirement as the optimal predicted stress-strain curve. And taking a phase field parameter corresponding to the optimal prediction stress-strain curve as a final component of the shape memory alloy. According to the method, the components of the shape memory alloy can be predicted according to the performance requirements of the shape memory alloy, and the reverse design capability urgently needed for designing a new material is achieved.

The invention discloses a PID (Proportion Integration Differentiation) controller parameter setting method and electronic equipment. The method comprises the following steps: acquiring historical operation data of a controlled object; determining a target model of the controlled object and target model parameters of the target model according to the historical operation data; determining an initial value of a PID controller parameter based on the target model parameter; calculating a performance index of the PID controller based on the historical operation data, the target model and the current value of the PID controller parameter; determining an objective function of the PID controller according to the performance index; and based on the initial value of the PID controller parameter and the target function, carrying out multiple times of iterative optimization on the PID controller parameter to obtain an optimized target PID controller parameter. According to the method, parameter setting can be automatically realized without depending on expert experience, performance-oriented parameter setting is realized, the setting precision and efficiency are improved, different performance requirements can be self-adapted, and parameter setting optimized as required is realized.

The encoding device (100) includes a circuit (a1) and a memory (a2) connected to the circuit (a1). During operation, for a layer set including at least one output layer, even if the number of layers included in the layer set is 1, the circuit (a1) encodes performance requirement information representing the performance requirements of the decoding device (200) into a header shared by the layers included in the layer set and generates a bitstream. The bitstream includes the header shared by the layers included in the layer set and data obtained by encoding at least one image included in the output layer.

The system power-saving operation device includes a design unit that designs a configuration of a DAG (disaggregated) node included in a DAG system, based on the amount of resources required for service operation on the DAG system, so that a service can be operated at a level that meets a required performance requirement and power consumption is minimized, and a control unit that controls the configuration of the DAG node included in the DAG system based on the designed configuration.

The invention relates to training-free error compensation for compressed large language models. A Large Language Model (LLM) learns, understands and generates human-like text through machine learning, and thus is very powerful when used for various language-based tasks (e.g., text summarization, translation, and content generation). However, in order to provide superior performance, LLM typically has a considerable model scale and requires higher inference costs. In order to reduce the size and execution costs of LLMs, a method has been developed for specialized compression of LLMs. However, compared with an uncompressed model, most of the existing methods have the advantages that the accuracy is obviously reduced, or the training time is longer, and the adaptability is generally limited by a limited compression format range supported by hardware. According to the method, error compensation is provided for the compressed LLM in a training-free manner, so that flexibility is provided for various performance requirements.

An encoder includes circuitry and memory coupled to the circuitry. In operation, the circuitry determines whether a determined relationship between a total number of one or more layer groups and a total number of one or more parameters is satisfied. The one or more layer groups each include at least one output layer, and the one or more parameters each indicate a performance requirement for a decoder. When the determined relationship is satisfied, the circuitry determines, using a determined method, a correspondence relationship for associating any of the one or more parameters with each of the one or more layer groups, and omits storing information indicating the correspondence relationship in a common header shared between layers in the one or more layer groups.

The invention discloses a balancing method for an electric tool. The balancing method comprises the steps that 1, an eccentric block is decomposed into a plurality of modules capable of quantifying structural parameters; step 2, according to geometric characteristics of the modules, constructing a mass function and / or a rotational inertia function of the corresponding module and coordinate parameters of the corresponding mass center relative to the rotation center; 3, defining boundary conditions of module structure parameters based on the minimum size threshold value of each module and the correlation size constraint between the modules; 4, constructing a dynamic balance equation based on the mass function and / or the rotational inertia function and the coordinate parameters; 5, constructing an eccentric block total mass equation based on the mass function, and / or constructing an eccentric block total inertia equation based on the rotational inertia function; and step 6, seeking a specified optimal solution meeting the dynamic performance requirement. According to the method, accurate balancing of the electric tool and light weight of the eccentric block are achieved, the problems that a traditional experience trial-and-error method is low in efficiency, poor in precision and difficult in quality optimization are solved, and the design efficiency and the equipment performance are improved.

A task scheduler device (100, 100A) is provided with: a dedicated command execution detection unit (110) that detects execution of a dedicated command; a dedicated command execution influence core specifying unit (120) that specifies a logical core that operates on the same physical core as the dedicated command execution core due to the execution of the dedicated command and that has a reduced operation frequency; a low-frequency permitted process allocation determination unit (140) that performs a determination for specifying a process that satisfies a prescribed performance requirement even when operating at a low frequency with respect to the logical core in which the operation frequency specified by the dedicated command execution influence core specification unit (120) is reduced; and a process core allocation unit (150) that allocates the process specified by the low-frequency permitted process allocation determination unit (140) to a logic core having a reduced operation frequency.

This disclosure provides a method for determining the frequency of a storage device. The method includes: determining a target performance requirement for the storage device of an electronic device, wherein the storage device performance includes a first performance and a second performance, and the target performance includes at least one of the first and second performances; determining a target frequency scheme for the storage device corresponding to the determined target performance, the target frequency scheme including a first frequency scheme corresponding to the first performance and a second frequency scheme corresponding to the second performance, wherein the first performance and the frequency of the first frequency scheme are linearly correlated, and the second performance and the frequency of the second frequency scheme are linearly correlated; and determining a frequency corresponding to the target performance requirement from the target frequency scheme based on the target frequency scheme and the current operating state of the electronic device.

The embodiment of the invention discloses a dynamic erasing method and a storage system. The method provided by the embodiment of the invention comprises the steps of determining performance demand data and a residual writable space of a target storage system, determining estimated erasing time according to the performance demand data and the residual writable space, and determining an erasing strategy corresponding to the target storage system according to the estimated erasing time, the erasure strategy comprises each block to be erased in the target storage system and a corresponding erasure mode, configuring erasure parameters corresponding to the blocks to be erased according to the erasure mode, and performing erasure operation on the corresponding blocks to be erased according to the erasure parameters. And according to the estimated erasing time, the erasing parameters corresponding to the to-be-erased blocks are adjusted, so that the reliability of the to-be-erased blocks and the system is improved on the premise of ensuring the performance of the storage system, the dynamic balance between the performance and the reliability of the system is realized, and the adaptive capacity and the overall efficiency of the system in a complex scene are enhanced.