65 results about "Workload prediction" patented technology

The invention provides a computing system and method for GPU computing power scheduling, and the method comprises the following steps: 1, obtaining computing task data, and carrying out the information collection of each task through receiving a to-be-processed computing task queue; according to the invention, through task information collection and deep analysis, and GPU resource omnibearing monitoring and workload prediction, deep understanding of tasks and GPU resources is realized, highly accurate computing power distribution can be carried out, resource waste is avoided, and the resource utilization rate is improved; advanced technologies such as a knowledge graph, mixed integer programming and deep reinforcement learning are fused through the intelligent scheduling model, the fusion mode is innovative in the GPU computing power scheduling field, the limitation of singleness of a traditional scheduling model is overcome, the method can better adapt to a complex and changeable computing environment, the task execution efficiency is optimized, and the task scheduling efficiency is improved. The adaptability and competitiveness of the whole system in a complex environment are remarkably improved, and an excellent solution is provided for the GPU computing power scheduling problem in the field of high-performance computing.

The invention discloses a data center energy consumption optimization method based on electricity price and load prediction, and relates to the technical field of scheduling optimization of a demand side, and the method comprises the steps: collecting parameters related to the energy consumption of a data center, constructing a data center energy consumption model based on workload transfer, proposing an electricity price and workload prediction model based on an LSTM neural network, and optimizing the energy consumption of the data center. The method comprises the following steps: constructing a data center load optimization scheduling decision model in a market environment by taking minimization of power purchase cost as an objective function and considering constraint conditions, and performing simulation verification on the model by utilizing an IEEE30 node system to obtain the electricity price of a prediction day, the arrived workload, the workload scheduling condition and the power purchase cost. According to the data center energy consumption optimization method based on the electricity price and load prediction, through introduction of the LSTM model, short-term and long-term fluctuation of the electricity price can be accurately predicted.

The invention relates to a method and a device for reducing response time delay of a deep learning model and a storage medium, which are applied to the technical field of artificial intelligence, and comprise the following steps: acquiring a historical load through sliding window sampling, acquiring initial load prediction data by using a dynamic joint prediction mechanism based on the historical load, obtaining an error compensation value through the corresponding actual load data sequence, and performing error compensation on the initial load prediction data to obtain final load prediction data; determining the total number of instances required for a period of time in the future based on the final load prediction data sequence; the number of instances which are currently operated is adjusted to match the total number of instances required in a period of time in the future, so that the response time delay of large-scale deep learning model reasoning work is reduced; through accurate workload prediction and resource scheduling, the response time delay of model reasoning can be effectively reduced, the achievement rate of a service level target is improved, and the service cost is reduced.

A workload allocation engine is configured to allocate a workload in a cloud architecture. A data center for an enterprise is identified, and for the data center, a plurality of servers within the data center and a plurality of virtual machines (VMs) running on the servers are identified. Based upon at least one predetermined factor, a plurality of clusters of the servers generated. For each of the clusters, a plurality of time-series variables are tracked. For each of the clusters and the workload, carbon emissions generated by a particular cluster and for the workload are predicted. The workload is assigned to the particular cluster based upon the predicted carbon emissions associated with the particular cluster, and the workload is performed by the particular cluster.

The invention discloses a database self-adaptive mixed data partitioning method and system based on reinforcement learning, and belongs to the field of artificial intelligence. According to the method and the device, the problems that a large amount of redundant data needs to be accessed and self-adaptive data partitioning cannot be automatically performed according to the dynamic working load when the query load is executed in an existing data partitioning scheme are solved. According to the method, a workload prediction model is established, the occurrence frequency of various types of queries in query workloads of a database in a future period of time is predicted based on historical workloads, a mixed data partition problem is modeled as a Markov decision process, and the state, action and reward of an intelligent agent are defined; according to the predicted occurrence frequency of various types of queries and the trained reinforcement learning model, an appropriate mixed data partitioning scheme is adaptively recommended for the arriving query workload, and the amount of data needing to be accessed when the query load is executed is reduced. The method can be applied to self-adaptive mixed data partitioning of the database.

The invention provides a multi-granularity GPU-CPU cooperative scheduling method for a heterogeneous computing system, and belongs to the technical field of heterogeneous computing resource management. The method comprises the following steps: S1, constructing a distributed cluster; s2, setting sampling parameters, and obtaining a working load time sequence of the distributed cluster; s3, resampling and aggregating the workload time sequence based on the sampling parameters to obtain a plurality of training subsets; s4, generating a workload prediction model by using each training subset and a time attention selection algorithm; s5, expanding the set evaluation function through the workload prediction model; s6, generating a scheduling rule based on the dynamic threshold parameter and the evaluation function; and S7, generating a scheduling decision based on the scheduling task and the scheduling rule, and performing scheduling based on the scheduling decision. The heterogeneous computing resource scheduling method has the advantages that the adaptive capacity, the resource utilization rate, the timeliness and the reliability of heterogeneous computing resource scheduling are greatly improved, and the energy consumption cost is greatly reduced.

The present disclosure relates to a deep learning-based method for predicting a high-dimensional and highly-variable cloud workload, including the following steps: Step S1: obtaining historical workload data of a cloud data center, and carrying out preprocessing; Step S2: on the basis of a raw data set, predicting a future workload of a central processing unit by using a deep learning based prediction algorithm for cloud workloads (L-PAW) integrating a top-sparse auto-encoder (TSA) and a gated recurrent unit (GRU), and transmitting a predicted result to a cloud service provider (CSP); and Step S3: determining, by the CSP, a resource allocation strategy according to the predicted result, such that the cloud data center achieves load balancing. The present disclosure realizes adaptive and effective workload prediction, thereby effectively improving the efficient resource allocation efficiency in cloud computing.

The present invention discloses an internal and external surgical nursing workload prediction system based on big data, including a data acquisition module, a data preprocessing module, a multidimensional workload factor extraction module, a workload prediction model establishment module, a model training optimization module and an internal and external surgical nursing work management module. The present invention relates to the field of nursing work information data processing technology, and specifically refers to an internal and external surgical nursing workload prediction system based on big data. This solution innovatively proposes a dynamic feature weight adjustment mechanism and spatiotemporal feature fusion to effectively capture the spatiotemporal correlation of the nursing workload; innovatively proposes a multi-level prediction architecture that can comprehensively capture the relationship between each level, thereby improving the accuracy of the entire nursing workload prediction result; innovatively enhances the training optimization effect of the model through a cross-level error compensation mechanism and an improved optimization algorithm, and significantly improves the application performance in nursing workload prediction.

The present invention provides a method for optimizing cooling resources in a data center based on workload prediction and a system thereof. The method includes the steps of: continuously collecting data in real-time from a plurality of sensors distributed across various sections of the data center; predicting future workload demands and corresponding cooling requirements over time and space in each section of the data center using historical and real-time collected data; determining whether to allocate the predicted workload demand to a section that meets the corresponding cooling requirements or to adjust the cooling resources to achieve the required cooling levels, whichever results in more optimized energy usage; and dynamically adjusting the cooling resources in each section of the data center based on real-time and predicted workload demands over time to minimize energy consumption and operational costs, thereby ensuring optimal cooling without over-provisioning.

The invention provides a distributed computing power center load and energy consumption collaborative scheduling method and system based on artificial intelligence, a storage medium and electronic equipment, and relates to the technical field of load and energy consumption collaborative scheduling. According to the method, a computing power center node workload prediction model, a computing power center node workload energy consumption prediction model and a cooling system energy consumption prediction model are respectively constructed by utilizing GAT and LSTM, the three models cooperatively work through a cascade architecture, and the output of a preorder model is used as the input of a subsequent model, so that the data interaction capability among the models is enhanced; and the overall prediction performance is improved. Besides, the strong language generation and knowledge reasoning capabilities of the generative artificial intelligence are utilized to generate a flexible and efficient scheduling strategy, so that the flexibility and the intelligence degree of the system are remarkably improved, and the system can better adapt to complex and changeable computing power demand scenes.

The specification provides a workload prediction method, device, medium and computer system in a cloud computing service, the method comprising: in response to a first trigger event, obtaining first time series data corresponding to system state data in a first time period; based on the first time series data, performing a first prediction for a future first time domain range to obtain a first prediction result; in response to a second trigger event, obtaining second time series data corresponding to system state data in a second time period; based on the second time series data, performing a second prediction for a future second time domain range to obtain a second prediction result; evaluating the accuracy of the first prediction result to obtain an evaluation result; and determining a target prediction result of the workload according to the evaluation result, the first prediction result and the second prediction result.

The invention relates to the technical field of cloud computing, and provides a cloud resource workload prediction method based on multi-scale self-adaption in a computing network environment, and the method comprises the steps: carrying out the data collection of a workload sequence of a cloud data center through employing a sample division method based on a sliding window, and obtaining the cloud workload data of different time scales; carrying out noise reduction processing on the cloud working load data by adopting an adaptive threshold noise reduction method based on wavelet multi-scale decomposition, and carrying out Fourier transform to extract frequency domain features; carrying out sample division on the processed cloud working load data by adopting an adaptive mode clustering algorithm, and outputting a mode data set; adopting a Mixup data enhancement method based on mode driving to generate a mixed sample, adding the mixed sample into the mode data set, and constructing an enhanced mode data set; and obtaining a cloud resource workload prediction result by adopting a similarity alignment dynamic integrated prediction method. Finally, an ablation experiment proves that the method has higher prediction precision.

The invention provides a self-supervised multi-scale cloud workload prediction method and system in the technical field of cloud computing resource management. The method comprises the steps of S1, collecting time sequence data of workloads of physical machines in a cloud environment; s2, performing data enhancement operation on each time sequence data to obtain enhanced sequence data; s3, constructing a multi-scale encoder based on the multi-scale patch embedding layer and the scale sensing layer; s4, performing self-supervised training on the multi-scale encoder based on the enhanced sequence data; step S5, adding a prediction head in the multi-scale encoder, introducing an adapter with a specific scale, obtaining a preset number of annotation data, performing fine adjustment on the prediction head and adapter parameters of the adapter based on each annotation data, and then constructing a cloud workload prediction model; and S6, performing workload prediction based on the cloud workload prediction model. The method has the advantage that the workload prediction precision is greatly improved under the condition of scarcity of labeled data.

The invention discloses a workload prediction adaptive neural network model for adjoint computation, and relates to the technical field of workload prediction, and the workload prediction adaptive neural network model comprises a federated learning module which is used for training a deep neural network model in a specific scene. And aggregating the models of the local clients through an aggregation strategy to obtain a global model. According to the workload prediction adaptive neural network model for adjoint calculation, the data is deeply analyzed, the prediction model can capture the change rule and trend of resource use, and the resource margin of the node in a period of time in the future is estimated accordingly.

According to some embodiments, methods and systems may be associated with a cloud computing environment. A workload prediction framework may receive observed workload information associated with a database in the cloud computing environment (e.g., a DataBase as a Service (“DBaaS”)). Based on the observed workload information, a Statement Arrival Rate (“SAR”) prediction may be generated. In addition, a host variable assignment prediction may be generated based on the observed workload information. The workload prediction framework may then use the SAR prediction and the host variable assignment prediction to automatically create a workload prediction for the database. A physical database design advisor (e.g., a table partitioning advisor) may receive the workload prediction and, responsive to the workload prediction, automatically generate a recommended physical layout for the database (e.g., using a cost model, the current physical layout, and an objective function).

A method for incentivizing contact center agents with scheduling perks that includes: receiving a workload forecast and staffing plan for a current shift; receiving a service level target; receiving types of scheduling perks; performing a selection routine for determining a select scheduling perk for offering to a select agent; and sending a scheduling perk offer via electronic communication to the select agent. The selection routine may include: selecting a proposed scheduling perks and a proposed agent; modifying the staffing plan to create a proposed staffing plan that reflects the proposed scheduling perk; predicting a proposed forecasted target adherence using the modified staffing plan; determining whether the proposed forecasted target adherence satisfies a threshold defined by an acceptable forecasted target adherence and, if so, deeming the proposed scheduling perk as select scheduling perk and proposed agent as select agent.

The application provides a kind of self-supervision multi-scale cloud workload prediction method and system in the technical field of cloud computing resource management, method includes: step S1, the time series data of each physical machine workload in cloud environment is collected;Step S2, each time series data is subjected to data enhancement operation to obtain enhanced sequence data;Step S3, based on multi-scale patch embedding layer and scale perception layer, multi-scale encoder is constructed;Step S4, based on enhanced sequence data, the self-supervision training of multi-scale encoder is carried out;Step S5, in multi-scale encoder, prediction head is added, and the adapter of specific scale is introduced, obtain preset quantity of labeled data, based on each labeled data, the adapter parameters of prediction head and adapter are fine-tuned, and then cloud workload prediction model is constructed;Step S6, based on cloud workload prediction model, workload prediction is carried out.The application has the advantages that in the case of labeled data scarcity, the workload prediction accuracy is greatly improved.

The present application discloses a method, device and storage medium for quantifying the workload of an imaging report, including: obtaining a target imaging report corresponding to a patient, and extracting multiple first key features in the text content of the target imaging report using an NLP model; encoding each first key feature using a pre-set encoding rule, and determining multiple feature codes corresponding to each first key feature and multiple encoding values ​​corresponding to each feature code; generating a corresponding first data set based on multiple feature codes and multiple encoding values; and inputting the first data set into a pre-set workload prediction model, and determining the workload corresponding to the target imaging report. The technical effect of truly reflecting the workload of doctors when filling out diagnostic reports based on target imaging reports and enhancing the work enthusiasm of doctors is achieved.

The application discloses a server energy efficiency dynamic optimization method based on workload prediction and a deep learning model, relates to the cross field of server energy efficiency optimization and deep learning technology, and comprises the following steps: collecting server multi-source data, performing space-time alignment on the server multi-source data, extracting causal characteristics, constructing a standardized data set, constructing a time series graph convolution network, combining a multi-head attention mechanism, outputting a probabilistic load prediction interval, constructing an energy efficiency objective function based on a Lyapunov optimization framework, quantifying performance loss and temperature drift, and outputting multi-agent optimization constraints, defining a multi-reinforcement learning agent, outputting a collaborative adjustment instruction through a counterfactual baseline algorithm, constructing a digital twin shadow model to deduce energy efficiency, combining a meta-learning fine-tuning model, forming a closed-loop optimization of physical and digital double verification, realizing dynamic and accurate optimization of server energy efficiency, and adapting to multiple computing power scenes.

The embodiment of the present invention relates to a computing resource allocation method, device and electronic device for a data middle station, including: receiving a computing task request from a user, and determining the computing task type and the corresponding computing resource demand based on the computing task request; obtaining basic operation status information of a computing resource pool, the basic operation status information including the workload of the computing resource pool; determining multiple computing resource allocation schemes based on the basic operation status information, the computing task type, the computing resource demand and a pre-built computing resource allocation utility model; determining the workload value of the computing resource pool based on the workload and a pre-built workload prediction model, and converting it into a cost vector; determining the best computing resource allocation scheme from multiple computing resource allocation schemes based on the computing resource allocation utility model and the cost vector, and allocating the best computing resource allocation scheme to the computing task request. The problem of low computing resource utilization is solved.

The invention discloses a self-guiding lift-off maneuvering network work load prediction method, and relates to the technical field of network load prediction, the self-guiding lift-off maneuvering network work load prediction comprises a resource prediction model, a federated learning module and a local data set, the federated learning module is specifically a distributed training architecture and is used for training a deep neural network in a specific scene, and the local data set is used for predicting the work load of the self-guiding lift-off maneuvering network. Aggregating the models of the local clients through an aggregation strategy to obtain a global model; the resource prediction model comprises a non-ground mobile network module, a local prediction module and an available resource cooperation module; the non-ground mobile network module is specifically a communication computing network constructed in the maneuvering process of the unmanned aerial vehicle group; according to the self-guiding lift-off maneuvering network work load prediction method, the non-ground mobile network has a good supplement effect on the ground network due to the flexibility of the structure, the maneuverability of the nodes and the diversity of the carried resources, and meanwhile, the resource guarantee requirement of an imminent scene is met.

Techniques for multi-layered prediction of computing workloads are disclosed. A system identifies a level of granularity associated with a request for predicting the computing workload of a particular entity. The system obtains attribute data for computing resources at the identified level of granularity. The system determines whether computing resources not specified in the request should be included in the workload prediction. The system applies a time series prediction model to time series data obtained from the computing resources associated with the request. The system presents one or more workload predictions for the computing workload associated with the request.

The invention discloses an accompanying cloud workload prediction algorithm based on deep learning, and relates to the technical field of network resource prediction. The non-ground mobile network is specifically a flexible and efficient communication and calculation network constructed by an unmanned aerial vehicle carrying a server and a 5G mobile base station in a maneuvering process, and can meet the communication and calculation requirements of resource-limited Internet of Things equipment in a presence scene; according to the adjoint cloud workload prediction algorithm based on deep learning, both computing resources and network resources are provided through a non-ground mobile network, and the algorithm is mainly applied to providing resource guarantee for service requirements of highly-imminent scenes. The scene generally has no preset infrastructure, and the nature, the social and the information environment show stronger uncertainty. The intelligent service also embodies the obvious characteristic of rapid change of time, space and resource demands.

The application relates to a cloud service workload prediction method and system based on a convolution enhanced Transformer, which comprises the following steps: collecting cloud server workload data and preprocessing the cloud server workload data to decompose the cloud server workload data into a trend component and a residual component; constructing a workload prediction model, training the workload prediction model by using the trend component and the residual component, and obtaining a trained workload prediction model; predicting the cloud server workload by using the trained workload prediction model to obtain a workload prediction value; predicting the trend component by using a trend information capturing module to obtain a prediction result of the trend component; predicting the residual component by using a convolution enhanced Transformer encoder module to obtain a prediction result of the residual component; and fusing the prediction result of the trend component and the prediction result of the residual component by using a feature fusion module to obtain a final workload prediction value.

The embodiment of the invention discloses a workload prediction method and device for an operation process, equipment and a storage medium, and relates to the technical field of power management, and the method comprises the steps: obtaining the process information of the operation process of a preset process category in a preset time interval from a process database, and obtaining the initial process information; on the basis of the associated construction event information, screening out operation procedures with the number of associated construction events larger than a preset number threshold value from the initial procedure information to obtain intermediate procedure information, and on the basis of the actual workload, screening out operation procedures with the actual workload not in a preset workload interval from the intermediate procedure information to obtain sample procedure information; according to the method, the average process workload corresponding to the preset process category is determined based on the actual workload in the sample process information, and the reference workload corresponding to the preset process category is determined based on the average process workload, so that the reference workload of the operation process can be accurately predicted, and a judgment basis is provided for judging whether the progress of the operation process is reasonable or not.

The invention belongs to the technical field of intelligent traffic, and discloses a driver workload evaluation method based on a skin electric signal under various hearing interferences, which inspects various different driving scenes to research the comprehensive influence of navigation frequency and music rhythm on the driver workload, combines subjective evaluation with the skin electric signal by adopting a double-index method, and improves the driver workload evaluation accuracy. Strong and consistent correlation is shown between SCL distribution of the galvanic skin signals and subjective evaluation scores, a stable basis is provided for the galvanic skin signals serving as main physiological indexes of a workload prediction model, and driver workload prediction based on the galvanic skin signals under various auditory interferences is more accurate.

The invention discloses a database query pre-optimization method, system and device and a medium, belongs to the technical field of database and AI fusion, and aims to solve the technical problem of how to reduce database query optimization overhead and improve database query response performance. A workload prediction framework adapting to a dynamic environment is designed, logic plan cache management is optimized based on a prediction result, accurate perception and quick response to query modes and resource demand changes are realized, and database performance and expansion capability are comprehensively improved; the method specifically comprises the following steps: predicting a workload trend: performing semantic-based query template characterization and workload sequence prediction based on a multi-scale attention mechanism to realize prediction of a future arrival query SQL (Structured Query Language) of a current workload; and logic plan caching: effectively caching the logic plan of the query statement output in the workload sequence prediction stage based on the multi-scale attention mechanism.

The invention discloses a project progress optimization method and system based on data analysis, and relates to the technical field of project progress optimization, and the method comprises the steps: obtaining the plan information of a target building project, and constructing a first workload prediction model based on historical project data; and based on the activity and position data collected by the wearable device, monitoring the task execution state of each worker in real time, and constructing a second progress assessment. When it is detected that the overall progress deviates from the target threshold value, the task scheduling sequence and the distribution strategy of the remaining workload are dynamically adjusted, and closed-loop optimization control over the construction period is achieved. According to the project progress optimization method based on data analysis provided by the invention, the workload and progress can be evaluated more comprehensively by combining multi-dimensional information such as historical data, environmental factors and staff efficiency, the limitation of depending on a single data source or simple average is avoided, and the adaptability of a project is remarkably improved. And a multi-objective optimization algorithm and reinforcement learning are adopted.

To enable accurate determination of the work effort required to port large-scale software projects across disparate platforms.SOLUTION: A method of predicting porting work effort is provided. A total porting work effort to port a software project from a source platform to a disparate target platform is predicted using a first porting work effort, a second porting work effort, and a third porting work effort. The software project is ported from the source platform to the disparate target platform provided that the total porting work effort is less than a defined maximum porting work effort threshold level.SELECTED DRAWING: Figure 1

The application belongs to the technical field of intelligent transportation, and discloses a driver workload prediction method based on multiple physiological signals under multiple auditory disturbances, which is created by combining three music rhythms (none, slow and fast), two navigation modes (high frequency and low frequency) and three traffic scenes (regular road, school road and construction site road), collects multi-modal data including skin electricity signal EDA, electrocardiogram signal ECG, electroencephalogram signal EEG, driving behavior characteristics and subjective workload evaluation, and the data is used to build a workload prediction model, and a stacking model is adopted, and the stacking model integrates multiple machine learning models to improve the prediction performance.