395 results about "Performance model" patented technology

The invention provides a GPU heterogeneous computing resource allocation method and system based on task scheduling, and the method comprises the steps: carrying out the feature analysis of a to-be-scheduled task, so as to extract the calculation intensity, video memory demands, inter-task data dependence and historical execution data of the to-be-scheduled task, and obtaining a task feature vector; constructing a performance model of heterogeneous GPU resources to predict execution efficiency and resource consumption of the GPU under the target task based on hardware performance parameters and running states of the GPU; and matching the to-be-scheduled task with the GPU resources according to the task feature vector and the performance model of the heterogeneous GPU resources. Through task feature analysis and GPU resource performance modeling, the adaptive relation between tasks and GPU resources can be deeply mined, to-be-scheduled tasks are accurately allocated to the most appropriate GPU, the resource utilization rate is greatly increased, the task execution sequence and resource allocation are dynamically planned, and the task execution time is effectively shortened.

The invention relates to the technical field of chip task allocation, and discloses a task allocation optimization method and device for a multi-core heterogeneous ASIC (Application Specific Integrated Circuit) computing mainboard. Computing core setting information of the computing mainboard is collected, a mainboard computing performance model is generated through digital feedback processing, an executable task set is obtained, and the computing requirement of each task is analyzed; performing simulation analysis on the calculation demand set based on the performance model by using an optimization algorithm, generating a task allocation plan, identifying and generating task execution characteristics of a currently executed task, allocating the task execution characteristics according to the task allocation plan, and scheduling calculation resources to process the current task. The resource performance is accurately simulated through the performance model, the optimization algorithm ensures efficient resource allocation, responds to task changes in real time, maximizes the utilization of heterogeneous computing resources, improves the overall computing efficiency, and solves the problem of execution delay when real-time task allocation is carried out on the multi-core heterogeneous ASIC computing mainboard in the prior art.

The invention relates to a network slice optimization method for a 5G private network, and the technical scheme comprises the steps: constructing a time sequence performance model in a slice operation process, and monitoring tiny changes including end-to-end time delay, throughput and packet loss rate; then extracting a short-period fluctuation mode of the index, and comparing the gradient with a historical stability threshold value; judging whether the index is in a performance critical state or not by identifying an index drifting behavior which continuously jitters but does not default yet; non-equal-ratio responses including jitter improvement and delay decline after each time of resource adjustment are recorded; then trend fitting is carried out on the data, and whether a performance platform period or a critical jump point exists or not is judged; delimiting three intervals of resource investment in the model: an inactive area with invariable investment, a high-sensitivity area with rapid performance change and a rising saturation area with stable performance and continuous resources, wherein the three intervals include an inactive area with invariable investment, a high-sensitivity area with rapid performance change and a rising saturation area with stable performance; and identifying the current response area of the slice through the model, determining that the regulation strategy does not take equal proportion input as the core, and taking the optimal response area as a resource allocation interval to make a choice.

The invention belongs to the field of big data, and relates to a data migration method, which comprises the following steps: obtaining storage cluster pressure test data, and constructing a performance model, the pressure test data being obtained by testing under different loads. And acquiring real-time monitoring data, and evaluating the current pressure state in combination with the performance model. And if the threshold value is exceeded, extracting a historical data change rule, and generating a load prediction model. And calculating a target migration concurrency number based on the real-time performance index and the prediction model. And adjusting the current data migration concurrence number and the bandwidth limiting parameter to target values. And finally, carrying out data migration of the storage cluster according to a preset protocol and the target parameter. The invention further provides a device, equipment and a medium. In addition, the invention also relates to a block chain technology, and the pressure measurement data and the real-time monitoring data can be stored in a block chain. The data migration efficiency can be improved.

The invention discloses a wind power generation system operation optimization method based on intelligent control, and the method comprises the following steps: S1, collecting the real-time data during the operation of a wind power generation system, including the wind speed, the wind direction, the rotating speed of a generator, the output power, and the environment temperature; s2, constructing a performance model of the wind power generation system according to the real-time data; s3, based on the performance model, adjusting the operation parameters of the generator in real time to optimize the power generation efficiency of the system; and S4, when detecting that the system is abnormal, automatically triggering a fault diagnosis module, generating a fault processing suggestion and executing an optimization strategy. Through real-time data acquisition, performance model construction, operation parameter dynamic adjustment and intelligent fault diagnosis, the problems that a traditional wind power generation system is low in efficiency and not timely in fault response are solved, and the system has the advantages of dynamically optimizing power generation efficiency, timely processing system abnormity and improving operation stability.

The invention discloses an energy-saving control method and device for a gas compressor and a medium. The method comprises the steps that historical operation data of a gas compressor are collected, and the historical operation data comprise compressor operation parameters, environment parameters, gas parameters and pipe network operation parameters; training a compressor performance model, a load prediction model and an energy efficiency optimization model according to the preprocessed historical operation data; performing operation data prediction according to the collected real-time operation data by using a compressor performance model, a load prediction model and an energy efficiency optimization model to obtain predicted operation data in a future predetermined time period; calculating the predicted operation data by adopting a non-dominated sorting genetic algorithm, and determining an optimal operation scheme; and the control parameters of the optimal operation scheme are issued to control systems of all the compressors for energy-saving control over the compressors.

The invention relates to the related technical field of integrated service management, in particular to a digital intelligent technology and software and hardware integrated service management method and platform and a storage medium, and the method comprises the steps: connecting an industrial internet, receiving integrated service requirements, deploying software and hardware based on specifications, fitting a service scene, setting a topological architecture conforming to the specifications, and collecting data to establish a performance model. According to the software and hardware matching topology architecture, a communication architecture is adopted, a monitoring mechanism is set, and abnormal response management is performed on each component and each service in the software and hardware matching topology architecture, so that the technical problems that a service scene is fixed and a static deployment mode is difficult to quickly respond to dynamically changing service requirements in a software and hardware integration process are solved; the technical effects that the industrial internet is connected, the integrated service demand information is received in real time, changes of service demands are flexibly coped with in the software and hardware configuration process, dynamic configuration and adjustment are carried out according to different service scenes and demands, service changes are rapidly responded, and the flexibility and adaptability of resource configuration are improved are achieved.

The invention provides a game model penetration prevention method and device, equipment and a storage medium. The method comprises the steps that a moving direction vector of a hand model is calculated according to current frame position data and previous frame position data of the hand model in a virtual environment; predicting the next frame of position data of the hand model according to the moving direction vector and the current frame of position data, and emitting rays based on the next frame of position data and the moving direction vector; and judging whether the ray has a collision event in the virtual environment or not, and performing position adjustment and model rendering on the hand model when the ray has the collision event. According to the method, through vector calculation and ray detection, the calculation complexity is reduced, and the real-time performance is improved. The introduction of model prediction enables the system to cope with the potential mold-penetrating situation in advance, and further improves the efficiency of mold-penetrating prevention. A direct position adjustment strategy is adopted for collision processing, complex physical simulation calculation is avoided, and therefore the problems that in the prior art, calculation complexity is high, and real-time performance is insufficient are solved.

The invention discloses a big data driven dynamic resource scheduling method for an IPv6 edge computing node, and relates to the technical field of edge computing, and the scheduling method comprises the following specific steps: S100, resource data collection and integration: deploying monitoring equipment at the IPv6 edge computing node, collecting computing, network, storage and energy resource data, and transmitting the data to a server; through multi-dimensional resource data acquisition and storage, comprehensive monitoring and quantitative integration of calculation, network, storage and energy resource data are realized, the resource scheduling is more accurate and efficient through the comprehensive data acquisition mode, and through construction of the energy consumption performance model, the performance value can be accurately calculated, and the energy consumption performance is improved. And the model is trained and optimized by using big data, so that the balance of energy consumption and performance in the resource scheduling process is realized, the resource utilization efficiency is improved, the energy consumption cost is reduced, and powerful support is provided for green and sustainable development of edge computing nodes.

The invention discloses a data disaster recovery method supporting multi-backup path switching, which comprises the following steps: generating a dynamic granularity data fragment based on the sensitivity of input data and the characteristics of a storage medium, and generating a metadata label containing a data fragment identifier, storage preference and recovery priority; constructing a heterogeneous transmission path performance model through metadata labels, generating a path quality map, distributing an optimal transmission path for each data fragment, and monitoring path performance; detecting anomalies through the standardized deviation of accumulated transmission delay and migrating fragments; carrying out parallel verification by adopting incremental integrity verification and Merkle proof stored in the block chain, and preferentially retransmitting fragments which fail in verification; according to the invention, through the cooperation of dynamic data fragmentation, backup path optimization and recovery verification, the reliability and stability of data disaster recovery are ensured so as to meet the actual business requirements.

The invention provides a helicopter flight parameter prediction method, system and device and a storage medium, and belongs to the field of helicopter performance prediction.The method comprises the steps that a helicopter and engine integrated performance model set is established; acquiring historical data including flight parameters of a helicopter, operation parameters of an engine and environmental parameters under different flight conditions and states; fitting undetermined coefficients of the helicopter and engine integrated performance model group by combining historical data with specific design data of a helicopter and an engine to obtain model coefficients conforming to specific helicopter and engine models; representing a to-be-predicted flight mission as a flight mission profile containing input parameters, substituting the input parameters of the flight mission profile into the helicopter and engine integrated performance model group with the determined coefficients, and solving the model to obtain predicted flight parameters. According to the method, reliable prediction of flight parameters of the helicopter in different flight environments can be realized, and a basis is provided for engine design, modification and use optimization.

The invention discloses a regulation and control strategy for a water-wind-light-storage integrated multi-energy complementary system in an extreme high-temperature scene, and relates to the technical field of optimal scheduling and toughness evaluation of the multi-energy complementary system. The method comprises the following steps: acquiring meteorological and equipment parameters, and constructing a performance model of photovoltaic, wind power, thermal power and power transmission facilities under an extreme high temperature condition; an objective function is established based on the minimum operation cost of the multi-energy complementary system, and constraint conditions are established according to supply and demand balance, equipment operation, peak regulation margin and load reduction; performing optimization scheduling on the system through an optimization tool in combination with a demand response strategy to obtain a multi-energy collaborative operation scheme; a toughness evaluation model based on a comprehensive hierarchy-entropy weight method is adopted, the toughness of the system is quantitatively analyzed from the dimensions of economy, flexibility, safety and the like, and toughness scores and improvement suggestions are output. The economic operation efficiency and the anti-risk capability of the multi-energy complementary system under the extremely high temperature condition are effectively improved, and scientific support is provided for safe and stable operation of the multi-energy complementary system.

The invention discloses a marine ranching early warning method and system based on a knowledge graph, and relates to the technical field of marine ranching management, and the method comprises the knowledge graph of large-class keywords of water quality management, equipment management and biological resource management and progressive-level subclass keywords of the large-class keywords, terminal subclass association preset acquisition parameter groups and state description; generating a marine ranching plane graph; equipment states and detection parameters are collected to form a normal state parameter group, and a normal planar graph representation model is constructed; generating an exception model based on a preset exception adjustment normal parameter group, and dynamically associating the exception model with small-class keywords in the knowledge graph; and collecting data in real time to construct a real-time model, comparing the real-time model with an abnormal model to trigger an alarm, and marking an abnormal root in the knowledge graph. According to the method, the limitation of traditional single-dimensional monitoring is broken through, multi-source data are integrated through knowledge graph semantization, dynamic threshold and spatial correlation analysis are combined, abnormal accurate positioning and cross-dimensional risk early warning are achieved, and the marine ranch management efficiency and safety are remarkably improved.

The invention provides a digital integrated construction method and system for a multi-source heterogeneous model. The method comprises the steps that standardized packaging of the multi-source heterogeneous model is stipulated; carrying out composite data model design; building association mapping of the multi-source heterogeneous model and the composite data based on a macro command; a task engine drives the multi-source heterogeneous model; and performing integrated joint simulation based on the multi-source heterogeneous model. According to the method, a multi-source performance model integration environment is constructed based on a digital model integration construction key technology, a basic environment is provided for digital model integration, testing, scheduling and application, and credible support data is provided for auxiliary decision making and application packaging.

The invention relates to the technical field of bridge engineering detection and repair, in particular to a bridge crack repair effect prediction method and system based on performance simulation, and the method comprises the following steps: obtaining basic data and environmental load data of a target bridge, and matching at least two candidate repair schemes from a preset repair scheme library; constructing an initial performance model before repair, and calculating an initial bearing capacity loss rate and a crack propagation rate; quantifying the interaction influence of the multi-modal data through a coupling factor matrix, and iteratively calculating the repair condition of a set time period after repair to obtain an initial simulation result; and collecting real-time monitoring data in a set time period in the repairing process and after repairing, and inputting the real-time monitoring data into a preset LSTM adaptive correction model to obtain a corrected simulation result. According to the method, the long-term repair condition after repair can be accurately calculated, the accuracy of a prediction result and the environmental adaptability are remarkably improved, and a scientific basis is provided for long-term service performance pre-judgment.

The invention discloses a collaborative resource scheduling method and system based on a programmable network. The method comprises the following steps: performing performance modeling on heterogeneous computing resource data in all computing resource data to obtain a uniform resource performance index; defining target computing resource data required by multi-dimensional resource demand services, and determining matching relationship information between each multi-dimensional resource demand service and the current computer cluster; in the deployment process, according to all the matching relation information, the correspondingly selected multi-dimensional resource demand service is deployed to each computer cluster; determining a shortest deployment path between the routing nodes in each computer cluster based on a Floyd algorithm; and in the actual scheduling process, executing the computing power arrangement scheme generated in the deployment process. According to the collaborative resource scheduling method and system based on the programmable network provided by the embodiment of the invention, the utilization rate of computing power resources in a wide area network can be improved, and a computing power scheduling strategy is dynamically generated.

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 discloses a training method and system for an abnormal transaction recognition model of an e-commerce platform, and particularly relates to the field of e-commerce. According to the method, multi-source transaction data are integrated, the accuracy of labeling is judged through cross validation, data labeling is divided into accurate labeling and inaccurate labeling, and after the inaccurate labeling is recognized, the similarity between prediction and an actual result is calculated and the performance stability of a model is evaluated by analyzing the sample proportion fluctuation condition of the model influenced by labeling errors; according to an evaluation result, the models are divided into a stable performance type and an unstable performance type, and different training strategies are adopted: for the stable performance model, the labeled data set is continuously used for training; and for the model with unstable performance, training is carried out in stages, pre-training is carried out by using accurate annotation data, then inaccurate annotation data is introduced, and optimization adjustment is carried out through an adaptive learning algorithm, so that the negative influence of annotation deviation on the model is effectively reduced, and the accuracy of the model in practical application is improved.

Techniques are disclosed in which a computer system receives, from a plurality of user computing devices, a plurality of device-trained models and obfuscated sets of user data stored at the plurality of user computing devices, where the device-trained models are trained at respective ones of the plurality of user computing devices using respective sets of user data prior to obfuscation. In some embodiments, the server computer system determines similarity scores for the plurality of device-trained models, wherein the similarity scores are determined based on a performance of the device-trained models. In some embodiments, the server computer system identifies, based on the similarity scores, at least one of the plurality of device-trained models as a low-performance model. In some embodiments, the server computer system transmits, to the user computing device corresponding to the low-performance model, an updated model.

The invention discloses a heat exchange station circulating pump characteristic adaptive optimization control method and system, and relates to the technical field of circulating pump control, and the method comprises the steps: building a building terminal heat transfer model and a building energy balance model based on the temperature distribution of the surface of a radiator, and then building a building load prediction model; performing numerical modeling on the performance curve of the heat exchange station circulating pump, and establishing a circulating pump performance model; based on a depth deterministic strategy gradient algorithm, establishing a neural network control model for self-adaptive control of the circulating pump of the heat exchange station; establishing a state, an action and a reward function of an agent of the neural network control model; and training the neural network control model to obtain a trained control model, and performing real-time control on the heat exchange station circulating pump based on the trained control model. The technical problems that a traditional control method of the circulating pump is low in operation efficiency and serious in energy waste are solved.

Systems, methods and computer-readable storage media are utilized to model a multi-dimensional performance score. A processing circuit including memory and one or more processors is configured to authenticate a first exchanger and a second exchanger, receive a selection of at least one actionable element corresponding to an activity of the first exchanger and the second exchanger, identify a performance model based on the activity, collect at least one performance indicator associated with the first exchanger or the second exchanger, collect first activity data of the first exchanger and second activity data of the second exchanger, model the at least one performance indicator, the first activity data, and the second activity data to generate a MDP score, and provide the MDP score.

The invention discloses a normalization measurement method for a heterogeneous computing resource multi-dimensional index system. The method comprises the following steps: constructing the multi-dimensional index system including performance, energy efficiency and utilization rate; constructing a performance model HCRN based on the multi-dimensional index system; collecting multi-dimensional index system data in real time through a hardware monitoring tool and a performance counter, and inputting the data into a performance model HCRN after data preprocessing; constructing a reference task library to evaluate the performance of the system performance under different field applications; dynamically adjusting the weight of the performance model HCRN by adopting a reinforcement learning algorithm to realize the self-adaption of the model to task load and hardware state change; and performing performance evaluation verification and iterative optimization on the adjusted model. According to the method, a scientific and quantitative decision basis is provided for resource scheduling, model selection optimization and service deployment, the utilization efficiency of heterogeneous computing power resources is improved, the overall performance of the system is improved, and it is ensured that the system can operate efficiently and stably in different environments.

The invention relates to the technical field of elevator component performance analysis, in particular to a construction method and system for an elevator component performance model and a storage medium. The method comprises the following steps that elevator use records are collected, flow frequency spectrum modeling is carried out to generate start-stop frequency spectrum features and elevator use mode features, then part response analysis is carried out on the use mode features to obtain dynamic characteristic indexes, the independent performance of parts is calculated, and the independent performance of the parts is calculated. And an elevator interaction network is reconstructed according to the independence performance and the use mode characteristics, energy consumption distribution of the elevator is calculated, overall parameters of the elevator are obtained, framework simulation is conducted, framework movement performance loss is calculated, part position projection is conducted on a simulation framework based on part independence performance, interaction performance is evaluated, and therefore an elevator part performance model is constructed. According to the elevator part performance model construction method, elevator start-stop frequency spectrum characteristics and energy consumption distribution can be accurately described, and the response characteristics and interaction performance of the parts can be dynamically evaluated.

The invention provides a section-operation-power system joint optimization design method for an electric vertical take-off and landing aircraft. The method comprises the following steps: defining a power system design variable, a section design variable and an operation optimization variable of the electric vertical take-off and landing aircraft; a power system quality model, an aircraft profile demand analysis model, a variable pitch propeller operation performance model and a motor operation performance model are established; power system design and operation control plan making of the electric vertical take-off and landing aircraft are combined, section-operation-power system joint optimization design of the electric vertical take-off and landing aircraft is achieved, and the characteristic of great difference between the wide-speed-range operation characteristic of a variable-pitch propeller and the hovering and cruising flight stage of the vertical take-off and landing aircraft is considered to the maximum extent. And compared with the design of firstly carrying out physical system design and then carrying out operation control design, the optimal design of the overall performance of the aircraft is realized.

The invention relates to a cache architecture optimization method and device for a multi-chip integration system and a storage medium, and the method comprises the steps: obtaining an architecture topological graph of the multi-chip integration system; the architecture topological graph comprises a storage hierarchy graph and a network topological graph; the storage hierarchy graph is a hierarchy connection graph of a cache subsystem of the multi-chip integration system; the network topological graph is a core particle connection graph of the multi-core particle integrated system; constructing a system performance model based on the architecture topological graph; and solving the system performance model to obtain a target cache parameter, and generating a target cache architecture graph based on the target cache parameter and the storage hierarchy graph. According to the method and the device, the problem that the architecture of the cache subsystem in the multi-chip particle integrated system is difficult to achieve optimal optimization is solved.

A system for electric aircraft fleet management for at least an electric aircraft is provided. The system includes a computing device communicatively connected to at least an electric aircraft, wherein the computing device is configured to receive measured data from the at least an electric aircraft, generate the aircraft performance model output as a function of the measured data, transmit the aircraft performance model output to a user, receive a flight plan from the user; and manage the at least an electric aircraft using the flight plan.

The invention relates to a method, equipment and medium for predicting the compressive strength of recycled coarse aggregate concrete based on an optimized machine learning algorithm, and the method comprises the steps: collecting factors which affect the performance of the recycled coarse aggregate concrete from literature data, taking the compressive strength as an output variable, and preprocessing the data; and combining at least one machine learning model and at least one optimization algorithm into a plurality of prediction model combinations. The preprocessed data is divided into a training set and a test set, and an optimal performance model is determined through comprehensive evaluation after the model is trained. And performing reverse prediction on the key factors by using a prediction model with optimal performance to obtain the influence trend of the change of the key factors on the compressive strength of the recycled coarse aggregate concrete. Compared with the prior art, the recycled coarse aggregate concrete compressive strength prediction model based on the optimized machine learning algorithm solves the problems of insufficient precision and poor generalization ability in a traditional prediction method.

The invention relates to the technical field of solar cell panel control, and discloses an operation control method and system for a solar cell panel, and the method comprises the steps: obtaining the specification information of the solar cell panel, setting the positioning information, carrying out the digital simulation according to the specification and positioning information, and generating a light receiving performance model and a sunlight environment model; generating a current-day illumination simulation model by combining seasonal and climatic factors by utilizing a historical database, analyzing the illumination simulation model and a light receiving performance model to obtain initial light receiving characteristics and a preset light receiving track, continuously recording illumination conversion electric energy of the cell panel to correct the preset light receiving track, and adjusting an inclination angle of the cell panel to obtain a current-day illumination simulation model. The energy conversion efficiency of the solar cell panel is improved, sunlight resources are fully utilized, different seasons and climate changes are adapted, the performance of the solar cell panel in a changing environment is enhanced, and the problem that in the prior art, the efficiency of collecting illumination data in real time for angle adjustment is low is solved.

The invention discloses a local reasoning acceleration system and method for a large-scale MoE model under limited resources, and the method comprises the steps: carrying out the MoE model performance modeling on large language model reasoning equipment, and obtaining a module calculation performance model; executing a sliding window type greedy prediction-scheduling strategy: performing expert module routing prediction by utilizing a gating network in the MoE model to obtain activated experts of each layer and an input token of each expert, and dynamically determining the number of layers of the MoE model contained in a single scheduling window by utilizing the module calculation performance model in combination with a minimum sliding window algorithm; and according to the predicted expert module route, performing optimal execution strategy search and MoE model layer scheduling on the activated expert module in the window with the single scheduling window length. According to the method, efficient and stable local large model reasoning deployment is realized through an expert activation prediction mechanism without training cost.

The invention discloses a cross-domain parallel strategy generation technology based on constraint guidance and redundancy calculation, and aims to research how a cross-domain distributed calculation framework dynamically generates a cloud edge end cross-domain training parallel strategy. The method comprises the following steps: firstly, researching a runtime parallel plan regeneration mechanism, fitting cloud edge end environment change to reduce a search space by constructing a constraint model and a performance model, and applying an optimized graph matching algorithm to obtain final runtime matching; and secondly, researching a parallel execution mechanism of multiple parallel plans during operation, determining a specific switching plan by constructing a video memory constraint and parallel coverage model, obtaining a new Micro-Batch parallel plan, and completing switching and execution of the related parallel plans. Through the mechanism, the parallel strategy generation is dynamically and quickly generated according to the dynamic change of the environment, the strategy search time is shortened, the switching influence is reduced, and the dynamic nature of the cloud edge end computing environment is effectively dealt with.