119 results about "Performance improvement" patented technology

The invention provides a large model cue word design method, system and device in an industrial scene and a medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: constructing a private cue word template library, and forming a domain special knowledge unit set through multi-source industrial data collection, intelligent semantic analysis and classified storage; customizing a four-layer structured template, sequentially establishing a task overview layer, a step disassembly layer, an instruction refining layer and an output specification layer, and constructing a layered mapping model from task definition to output execution; template library management and iterative optimization are carried out, and dynamic updating and performance improvement of a cue word system are realized through centralized system management, multi-source feedback acquisition and data-driven optimization. According to the method, a cue word design and optimization system based on a hierarchical structure is constructed for complex task requirements in an industrial scene, and the task processing performance of a large model in scenes of industrial production, quality detection, equipment operation and maintenance and the like is improved.

The invention relates to the technical field of industrial equipment control, and discloses an industrial equipment intelligent control system and method based on deep reinforcement learning. The system comprises an equipment state monitoring module, a reinforcement learning decision-making module, a control parameter adjustment module, an abnormity intervention module and an equipment performance optimization module. The equipment state monitoring module collects real-time operation data of equipment, analyzes state change by means of a deep reinforcement learning algorithm, and generates a state feature data set; the reinforcement learning decision-making module calculates a control strategy by using a deep reinforcement learning model and generates a control strategy parameter set; the control parameter adjusting module optimizes the operation efficiency according to the adjusted parameters and generates an optimized control parameter set; the abnormity intervention module monitors abnormity, recognizes intervention through deep reinforcement learning, and generates an abnormity intervention adjustment data set; and the equipment performance optimization module optimizes the performance indexes according to the parameters and generates a performance optimization parameter table. The system realizes accurate monitoring of equipment, intelligent optimization of strategies, timely intervention of abnormities and performance improvement, and meets intelligent management and control requirements of the equipment.

The invention discloses an instruction scheduling method, device and system, a product and a medium, and relates to the technical field of processor design. The invention provides an instruction scheduling method aiming at the problem that the dynamic change in the instruction transmitting time cannot be accurately scheduled by the existing out-of-order scheduling strategy. The historical delay information of each instruction is stored through the delay cache, and when the instruction execution delay corresponding to the instruction is dynamic delay, the dynamic delay is predicted through the historical delay information stored in the delay cache, so that instruction scheduling is performed. According to the method, the dynamic delay of which the delay time cannot be predetermined is included in the consideration range of instruction scheduling based on the dynamic delay prediction mechanism. On the basis of the method, for dynamic change scenes in instruction transmission time such as memory access delay and cache miss, the optimal instruction transmission opportunity can be accurately predicted, so that the effect of instruction scheduling on performance improvement of a processor is further improved.

The invention provides a database fault processing method and system based on reinforcement learning, and relates to the technical field of database management, and the method comprises the steps: collecting a performance index, an execution plan and error log data at a fault moment, constructing a causal decision tree to recognize a fault root cause, and generating a propagation link; constructing a reinforcement learning action sequence based on the priority list; dividing a master control action set and an adjustment action set according to the influence degree of the action on the node state; and the execution strategy is dynamically adjusted according to the performance improvement ratio. According to the method, the fault root cause can be accurately positioned, the processing strategy is adaptively optimized, and the fault recovery efficiency and the system reliability are improved.

The invention provides a system and method for managing the lifecycle of machine learning models, from development to deployment and ongoing operation, across various environments including on-premises, cloud, and hybrid infrastructures. The system features a model build platform for data processing, feature generation, model development, training, and hyperparameter tuning. A model analytics engine extracts metadata, performs complexity analysis, and generates configuration files specifying environment settings and resource needs. A secure model repository enables version-controlled storage, while a deployment platform retrieves, validates, and deploys models in containerized environments like OpenShift or Kubernetes. The platform dynamically allocates resources, supports real-time and batch scoring, and monitors model performance with guardrails. Customizable agents provide real-time feedback and automated optimization, and the system can securely decommission models while maintaining detailed lifecycle records. The invention enhances the efficiency, security, and scalability of machine learning operations with continuous performance improvement and compliance automation.

The invention discloses a numerical control system feed-forward compensation method based on an edge controller, and the method comprises the steps: enabling the edge controller to obtain a machine tool instruction and a machine tool reality of a numerical control system in real time, calculating a tracking error, calculating a feed-forward compensation amount through an algorithm, transmitting the feed-forward compensation amount to the numerical control system, and installing a PLC function module in the numerical control system. And transmitting the received feed-forward compensation quantity to a servo feed-forward channel through a functional block, thereby realizing the feed-forward compensation function of the numerical control system. Through cooperation of edge calculation and a numerical control system, a feed-forward compensation algorithm is decoupled from a main controller, and high-precision real-time compensation is achieved through the parallel calculation capacity of an edge module. The design not only solves the contradiction between computing power and delay of a traditional system, but also supports multi-source data fusion and adaptive algorithm deployment through an open architecture, and provides a brand new path for performance improvement of a numerical control system.

The invention belongs to the technical field of industrial equipment control, and discloses a dynamic control method and system for feeding of a card breaking machine, and the method comprises the steps: obtaining the multi-dimensional state characteristics of the card breaking machine, and generating a recommended feeding control action through an automatic feeding control strategy generation logic; monitoring an actual feeding control action executed by the operator through the tile breaking machine, and comparing the actual feeding control action with the recommended feeding control action to identify a correction behavior of the operator; when the correction behavior of the operator is identified, monitoring equipment operation data of the mahjong machine in an observation time period after the actual feeding control action is executed, so as to evaluate multi-dimensional equipment performance improvement information; according to the multi-dimensional state features, the recommended feeding control action, the actual feeding control action and the multi-dimensional equipment performance improvement information, an automatic feeding control strategy generation logic is adjusted so as to perform feeding control on the tile breaking machine; therefore, the experience of an operator can be learned, and the self-adaptive capability and performance of the tile breaking machine under complex working conditions are improved.

The invention relates to the technical field of switch testing, and provides a switch testing system data interaction method and device, and the method comprises the steps: obtaining the operation state information and data packet flow information of each port, carrying out the extraction of the data packet flow information, obtaining a data packet feature parameter matrix, and carrying out the dynamic flow scheduling of a switch testing system, obtaining a port flow regulation scheme, and generating test result data based on the port flow regulation scheme and the operation state information; and performing comprehensive evaluation on the performance indexes of the switch based on the test result data, generating corresponding optimal configuration parameters based on an evaluation result, and applying the optimal configuration parameters to the switch test system. By acquiring the running state information and the data packet flow information of each port, the optimal configuration parameters are generated, efficient optimization and performance improvement of the switch test system are realized, and the problems of low response speed and insufficient flexibility in facing a dynamically changing network environment are improved.

The invention discloses a hydroelectric generating set bolt service stress nephogram inversion optimization algorithm, and relates to the technical field of hydroelectric generating set bolt stress measurement, and the method comprises the following steps: S1, finite element simulation model construction: S2, carrying out data cleaning on physical field simulation data and field measurement data obtained in S1, and obtaining a physical field data set; s3, constructing a stress nephogram inversion model; and S4, searching for optimal parameters. According to the method, the defects in the aspects of actual environment modeling and structure optimization in the prior art are overcome, and a new technical approach is provided for maintenance and performance improvement of the bolts of the hydroelectric generating set. And high-precision simulation in an actual operation environment is realized, and the defects in the aspect of digital modeling in the prior art are overcome. Based on theoretical physical constraints, a reinforcement learning algorithm can be constructed to invert stress distribution of a bolt in long-term service, comprehensive information is provided for structural health monitoring, and certain limitation exists in the aspect in the prior art.

A method for providing automated agent performance improvement includes obtaining a ranking for each agent of a plurality of agents; obtaining customer-agent interactions for each agent; determining a task label for each of the customer-agent interactions; selecting a first set of customer-agent interactions from the customer-agent interactions corresponding to a first task label; receiving application event streams corresponding to the first set of customer-agent interactions and comprising agent application usage data; analyzing the application event streams for statistically relevant differences for one or more activities performed by a first group of agents compared to a second group of agents; determining at least one activity of the one or more activities that is performed by the first group of agents differently than the second group of agents; generating guidance corresponding to the least one activity for the second group of agents; and deploying the guidance to the second group of agents.

The invention relates to a dynamic power control method and system for a storage and charging all-in-one machine, and the method comprises the following steps: data collection and monitoring: collecting and analyzing key parameters in real time through multi-source data fusion; dynamic safety threshold calculation: dynamically calculating a safety operation threshold of each subsystem based on real-time data, and ensuring that the system operates in an optimal state; multi-objective optimization control is carried out, an intelligent algorithm is adopted, and an optimal balance point is found among battery safety, power grid stability and charging efficiency; according to a fault processing and redundancy mechanism, a multi-layer fault response system is established, and the safety and continuity of the system under the abnormal working condition are ensured; according to the energy efficiency management and strategy optimization, long-term performance improvement and economical efficiency optimization are achieved based on a data closed loop, and the problem that potential safety hazards are caused due to the fact that a traditional storage and charging all-in-one machine system neglects the common influence of multiple factors such as SOC, temperature and power grid loads is solved.

The invention provides an employee performance prediction method based on multi-source heterogeneous data, and relates to the technical field of enterprise management. Structured data and time sequence data of employees are collected, the two types of data are preprocessed and then associated and fused with unique identifiers of the employees as cores, and a panoramic data wide table is constructed; then static and dynamic time sequence features are extracted from the wide table, the features are input into an artificial intelligence evaluation model, the model calculates basic performance indicators through a rule engine, and performance prediction scores are obtained through comprehensive evaluation in combination with a random forest regression model; the method is further provided with a potential prediction model based on an LSTM architecture integrated attention mechanism, and performance improvement potential scores and confidence intervals in the next three months can be output. According to the method, the problems of multi-source heterogeneous data fusion, model self-adaption, single evaluation dimension and the like are solved, and the employee performance can be predicted and evaluated comprehensively, objectively and prospectively.

The application provides a reliability analysis method for positioning accuracy of a shield tool changing robot, realizes sensitivity analysis of error parameters by constructing a kinematic model and an error proxy model, and accurately evaluates the failure probability of positioning accuracy. The assembly and machining errors are comprehensively considered, the error parameters are treated as random variables, the accuracy and efficiency of the analysis are improved. The mean and standard deviation of the random variable are used for sensitivity calculation, and a failure calculation method of positioning accuracy sensitivity is obtained, which provides an important basis for optimization design. The application not only helps to reduce the tool changing time, improve the construction efficiency and reduce the safety risk, but also provides an accurate and reliable tool changing robot manufacturing scheme for designers. The method has universality and flexibility, and can be widely applied to performance improvement and fault prevention of the shield tool changing robot. The implementation of the application will promote the development of shield construction technology and provide strong support for industrial automation and efficient production.

The invention provides a pressure sensing system compensation method and system, and the method comprises the steps: building a mathematical fitting model of an input and output pressure difference value and time under different conditions through testing and processing the data of a system in a starting time period under different input pressure and working temperature conditions, obtaining an initial environment condition when the system is started, and carrying out the calculation of the initial environment condition after the initial environment condition is obtained; and a corresponding mathematical fitting model is substituted to compensate the output pressure, so that the accurate pressure value in the system starting time period can be quickly calculated, and the performance improvement in the whole time period after the system is started is realized. According to the technical scheme, the technical problem that in the prior art, the pressure measurement precision deviation is large due to the fact that a thermal field is unbalanced in the starting period of a pressure sensing system is solved.

A business management system with integrated business intelligence and IT-supported optimization, consisting of: a computer device comprising at least one processor and a memory arrangement for storing executable instructions; a data acquisition and integration unit that is operationally connected to the at least one processor and is configured to receive heterogeneous enterprise data from a variety of source systems, including financial transaction systems, operational control systems, human resource information systems, supply chain systems, customer interaction systems, and IT infrastructure monitoring systems, wherein the data acquisition and integration unit is further configured to normalize syntactic formats, resolve temporal inconsistencies, and map source-specific attributes into a unified enterprise data representation stored in the storage system; a business intelligence processing unit that is operationally coupled with the data acquisition and integration unit and is configured to calculate business performance indicators through multidimensional aggregation, correlation analysis, and trend extraction over the unified business data representation, with the calculated business performance indicators being time-stamped and permanently stored; an optimization unit that is operationally coupled with the business intelligence processing unit and is configured to compare the calculated business performance indicators with stored business objectives and constraints and to determine optimized business control parameters that correspond to resource allocation, operational planning, cost efficiency or performance improvement; a control and orchestration unit that is operationally linked to the data acquisition and integration unit, the business intelligence processing unit, and the optimization unit, wherein the control and orchestration unit is configured to coordinate the execution sequence, forward the results of the business intelligence analysis to the optimization unit, and manage feedback by incorporating the effects of applied optimized business control parameters into subsequent data acquisition cycles; and a communication interface that is connected to at least one processor and configured to exchange data and control signals with external enterprise systems and user terminals.

The invention belongs to the technical field of artificial intelligence, and relates to a dynamic controllability management method and system for long-term evolution of a large model, and the method comprises the steps: S1, obtaining the current state information of the large model; s2, calculating a controllability score of the large model based on the output distribution entropy, the distribution variable quantity and the logits drift of the large model, judging whether the current state of the large model deviates from a controllability critical zone or not based on the controllability score, if the current state deviates from the controllability critical zone, entering S3, and otherwise, directly entering S4; s3, actively correcting the offset behavior of the large model in the training process through a controllable calibration head, so that the current state of the large model does not deviate from the controllable critical zone; and S4, dynamically adjusting a training strategy of the large model based on the controllability score and the task category. According to the method, the large model has the capabilities of long-term plasticity maintenance, continuous performance improvement and rapid adaptation to new tasks, and the fundamental upgrade from a static optimization model to a self-evolution model system is realized.

The invention provides an industrial control parameter adaptive optimization method based on deep reinforcement learning, and relates to the technical field of industrial control, and the method comprises the steps: collecting and preprocessing system operation data, dividing steady-state and fluctuation working conditions according to disturbance characteristics, and building a corresponding sub-network; adaptively switching the network output weight based on the disturbance degree to generate an initial optimization scheme; experience playback is carried out by adopting a self-adaptive parallel sampling strategy; training a deep reinforcement learning model and updating parameters; and the calculation performance improvement value is converted into a reward value to be compared with a dynamic threshold value. According to the invention, adaptive optimization of industrial control parameters can be realized, and the robustness and control precision of the system to different working conditions are improved.

The invention provides a model performance improvement method and device, electronic equipment and a storage medium, and relates to the technical field of computers and the Internet. The method comprises the steps of determining a feature detection result based on probability density distribution change between each key feature and similar application features in application data; the application data is data obtained after the model is online; the key features are M features with the maximum feature importance in the sample data; under the condition that the feature detection result represents key feature drift, training based on new sample data obtained according to a time sequence to obtain a corresponding sub-model; a new model is determined based on the sub-model and the model. According to the method, the integrated training process of the model is triggered only when drifting of the key features is detected, so that the purposes that the model is not retrained blindly, the model is trained as required to adapt to the feature drifting condition, the model is adjusted in time to prevent performance degradation of the model, and then the problem of model attenuation after online can be effectively solved.

The invention discloses a gas compressor stability extension method based on clustering modal cluster energy transient growth rate regulation and control, which comprises the following steps of: acquiring flow field data through unsteady numerical simulation of a centrifugal gas compressor, performing dynamic modal decomposition to extract modal characteristics, obtaining a modal cluster by utilizing clustering analysis and calculating energy, the method comprises the following steps: defining a modal cluster energy transient growth rate, determining a design variable and a change range, generating a sample set, calculating the modal cluster energy transient growth rate, constructing an agent model, carrying out multi-objective optimization by adopting a non-dominated sorting genetic algorithm, and finally obtaining an optimal design scheme. According to the optimization method based on clustering modal cluster energy transient growth rate regulation and control, performance improvement of the centrifugal compressor under different working conditions is achieved, the isentropic efficiency and the total pressure ratio of the compressor are remarkably improved, and meanwhile the operation stability of the compressor is enhanced.

The invention discloses a deployment-oriented wireframe analysis method and system, and belongs to the technical field of wireframe analysis, and the method comprises the steps: collecting an original image, and carrying out the preprocessing of the original image, and obtaining a to-be-analyzed image; constructing and training a wireframe analysis framework to obtain a wireframe analysis model; the method comprises the following steps: inputting a to-be-analyzed image into a wireframe analysis model, extracting multi-scale features with long-range context dependence in the to-be-analyzed image through mobile terminal line analysis ViT, local convolution and token interactive mixing processing, aligning the multi-scale features through a multi-scale feature line decoder, and performing structure perception enhancement, and analyzing to obtain a wireframe in the original image. According to the method, the structural continuity and the intersection point positioning stability in the wireframe analysis process are enhanced, clearer performance improvement is obtained in the line segment decoding process compared with a simple splicing mode, and meanwhile the real-time processing speed is kept; the method supports plug and play, does not need an additional supervision mechanism, and keeps compatibility with standard training and reasoning processes.

This application discloses a method and related apparatus for determining a data processing model. Based on a training dataset and initial training parameters, a machine learning model is trained to obtain a target model. The performance parameters of the target model and the contribution parameters of multiple data features to the prediction results of the target model are determined. If the target model satisfies the self-optimization condition based on the performance parameters and contribution parameters, the initial training parameters are adjusted to obtain optimized training parameters. Based on the training dataset and optimized training parameters, the target model is further trained to obtain a data processing model. The data processing model is used to process financial data to obtain the corresponding prediction results. The training process considers changes in performance and contribution parameters, and automatically adjusts the model parameters to ensure that the training process is geared towards performance improvement and smooth changes in contribution parameters. The trained model has high performance and interpretability.

The invention relates to the technical field of domain control integration, and discloses a fault processing method, device and system of a high-voltage domain control system, a vehicle and a medium. All high-voltage part control units integrated by the high-voltage domain control system are divided into a power type control system and a non-power type control system according to rules related to power driving; different numbers of fault levels are set for the power type control system and the non-power type control system respectively, corresponding fault processing strategies are formulated, and when it is detected that the power type control system or the non-power type control system breaks down, corresponding fault processing can be carried out on the power type control system or the non-power type control system. Through the above system integration scheme and implementation of refined fault processing strategies on different fault levels of different types of control units, the effects of performance improvement, light weight, convenient management and control and the like brought by domain control integration can be obtained, the fault probability of vehicle power loss caused by high-voltage domain control can be reduced, and the vehicle safety is improved.

The application discloses a stage-decoupling-based large language model online inference service delay optimization method and system, belongs to the technical field of artificial intelligence, and solves the large language model online inference service delay problem under the premise of ensuring model accuracy. The method comprises the following steps: a pre-padding scheduling is an adaptive pre-padding task scheduling strategy, the batch processing scale is dynamically adjusted according to the current load condition, the input prompt length and the GPU resource occupation; a dynamic batch size mechanism adaptively adjusts the batch processing scale according to the length distribution of input requests, dynamically combines and allocates tasks; a decoding scheduling is a decoding scheduling algorithm based on a multi-level feedback queue (MLFQ), and the task priority and the time slice allocation are dynamically adjusted; and a KV cache management mechanism is used for automatically exchanging the KV cache to the host memory when the task is idle or waiting for scheduling, and the GPU is used for pre-fetching back when the task is about to be executed. The application is suitable for performance improvement of a large language model for inference service.

The invention provides a large model testing system, method and device and a medium, and relates to the technical field of artificial intelligence, in particular to the technical fields of machine learning, deep learning, large models and the like. The large model to be tested is configured to execute a plurality of rounds of code iterative optimization tasks. The system comprises an execution unit configured to execute a code sample generated by a large model in a current round in a sandbox environment to obtain an execution result; the index unit is configured to calculate a performance improvement index based on the execution result, and the performance improvement index represents code generation performance changes of the large model among a plurality of rounds; the feedback unit is configured to select a target level from a plurality of preset diagnosis levels based on the performance improvement index, and the plurality of diagnosis levels respectively represent different feedback information detail degrees; and feedback information is generated based on the target level and the execution result, and the large model executes the next round of code iteration optimization task based on the feedback information.

Techniques for managing task assignments to workers in a work environment are disclosed. The system identifies one or more workers with qualifications that match recommended qualifications to perform tasks in the work environment. The system applies a trained machine learning model to task performance data associated with the workers, such as past history of tasks performed and statistics related to the performance of the tasks. The machine learning model generates predictions of downstream effects associated with assigning the tasks to the users. The downstream effects include delays and performance improvements for subsequent tasks performed by the worker, as well as effects on tasks performed by other workers at work centers in the work environment.

The invention provides a piston type air pump performance simulation method. The method comprises the steps that firstly, a fluid domain is extracted and imported into a model; step 2, segmenting and combining the model; 3, dividing grids for the first-stage piston region and the second-stage piston region; step 4, dividing grids for the first-stage inlet valve area and the second-stage inlet valve area; 5, dividing grids for a secondary outlet valve area; step 6, dividing grids for fluid domains of other regions; 7, setting and selecting an interaction surface and a boundary condition; the performance of the piston type air pump is calculated by adopting pumplinx, various performance indexes of the air pump under different working conditions can be accurately evaluated, flow, pressure and efficiency key parameters of the air pump can be accurately predicted, and a reliable basis is provided for design, optimization and performance improvement of the air pump.

The application discloses an employee performance improvement method, device, equipment and medium, the employee performance improvement method comprises the following steps: determining the work post of the employee to be promoted, and identifying the performance evaluation type of the employee to be promoted according to the work post; according to the performance evaluation type, the performance evaluation data of the employee to be promoted is obtained; based on the performance evaluation data, the performance improvement scheme of the employee to be promoted is generated through a large model; the execution feedback data of the employee to be promoted executing the performance improvement scheme is obtained, and based on the execution feedback data, the performance improvement result of the employee to be promoted is obtained. The method dynamically adapts the post characteristics and the performance evaluation type, utilizes the real-time analysis and intelligent decision-making capability of the large model on the multi-source data, generates a personalized performance improvement scheme, so that the employee to be promoted continuously obtains accurate guidance and positive motivation in the work process, and then actively improves the employee performance.

The invention belongs to the technical field of data processing and analysis, machine learning and recommendation systems, and discloses a priori knowledge migration-based recommendation fine arrangement model construction method, which specifically comprises the following steps of: 1, training sample production; step 2, feature engineering; step 3, model training; and 4, reducing the influence on the front-end link. Through model training optimization, rich front-end data is utilized to improve the training accuracy of a rear-end sparse link, the model preferentially stabilizes a front-end link target at the initial stage, then the rear-end link loss is gradually increased, the overall performance is improved, in order to reduce the interference of the front end on rear-end training, the reverse transmission of the front-end output is blocked before the front-end output is taken as the rear-end input, and meanwhile, the rear-end sparse link training accuracy is improved. The sparse target is processed by adopting functions such as huberl oss, and the influence of abnormal values is reduced by setting hyper-parameters, so that the model can quickly adapt to non-mainstream distribution data, the interference of abnormal data is reduced, and the prediction effect is further enhanced.

Aspects of the disclosure are directed to enhancing RAG through a structured schema for RAG that captures a wider range of RAG techniques to serve as a foundation for improving performance and an optimization framework for efficiently serving the wider range of RAG techniques based on significant performance variability across RAG workloads. The optimization framework may achieve up to a 2× increase or more in query per second (QPS) per chip compared to alternative RAG frameworks built on LLM-system extensions.

The application relates to the technical field of industrial equipment control, and discloses an industrial equipment intelligent control system and method based on deep reinforcement learning. The system comprises five modules, namely, an equipment state monitoring module, a reinforcement learning decision module, a control parameter adjustment module, an abnormal intervention module and an equipment performance optimization module. The equipment state monitoring module collects real-time operation data of equipment, analyzes state changes by using a deep reinforcement learning algorithm, and generates a state feature data set; the reinforcement learning decision module generates a control strategy parameter set by using a deep reinforcement learning model algorithm to control the strategy; the control parameter adjustment module adjusts parameters to optimize operation efficiency, and generates an optimized control parameter set; the abnormal intervention module monitors abnormalities, identifies intervention by using deep reinforcement learning, and generates an abnormal intervention adjustment data set; and the equipment performance optimization module optimizes performance indexes, and generates a performance optimization parameter table. The system realizes accurate equipment monitoring, intelligent strategy optimization, timely abnormal intervention and performance improvement, and meets the intelligent management and control requirements of equipment.