10 results about "Utility computing" patented technology

The invention relates to the technical field of computers, in particular to a load awareness and task migration scheduling method of a heterogeneous multi-core power terminal and related equipment. A real-time load state vector is constructed by collecting core load state data, tasks are classified, dynamic priorities are calculated, and task priority data are generated. And calculating a global load imbalance degree according to the real-time load state vector, and triggering a task migration scheduling process when the global load imbalance degree exceeds a set threshold value. In the process, a utility function model is constructed based on task priority, performance gain and migration cost, utility calculation is carried out on a task and target core combination, an optimal migration scheme is screened, migration operation is executed, and efficient distribution of tasks and dynamic optimization of system energy consumption are achieved. According to the method, real-time load balancing and optimal performance scheduling of the heterogeneous multi-core power terminal are realized, so that the task processing efficiency and the operation stability are remarkably improved.

The invention relates to an unmanned aerial vehicle group electric energy guarantee resource optimization distribution method comprising the following steps: S1, collecting unmanned aerial vehicle states and environment data, and building an initial model; s2, using STGNN to predict future supply demands, and integrating a telecommunication traffic prediction thought to process spatial-temporal dynamics; s3, distributing an electric energy guarantee vehicle according to the alliance formation game and the innovation utility function; s4, monitoring and adjusting distribution in real time; and S5, evaluating system performance and performing iterative improvement. According to the method, a bee colony resource allocation behavior is simulated, and dynamic and predictive resource optimization is realized by using a space-time diagram neural network STGNN prediction model; and edge utility calculation is introduced, so that the stability and efficiency of distribution are ensured.

The invention belongs to the technical field of heat dissipation of electronic equipment, and particularly relates to a CDU cooling optimization method and system based on temperature feedback, and the method comprises the steps: collecting the real-time operation data of a cold plate type liquid cooling system; the deviation between the water supply temperature and a preset target temperature is calculated, an original control output quantity is calculated through a proportional-integral-derivative controller, and a total heat dissipation demand signal is determined according to the original control output quantity and an output upper limit value; according to the real-time power consumption of the water pump and the fan and the heat dissipation marginal utility, the cooperation deviation at the current moment is calculated, the energy efficiency distribution weight at the previous moment is updated based on the logarithmic cooperation control law, and the energy efficiency distribution weight at the current moment is obtained; and according to the total heat dissipation demand signal and the energy efficiency distribution weight at the current moment, the target rotating speed of a water pump and the target rotating speed of a fan are calculated respectively. According to the invention, collaborative optimization control of the water pump and the fan can be realized, and system energy consumption is reduced.

The invention discloses an intervention task closed-loop scheduling method and system based on a learning condition risk radar, and belongs to the technical field of online learning education, and the method comprises the steps: collecting original learning condition log data, carrying out the preprocessing, and obtaining a smooth feature vector; carrying out trend risk estimation and mutation risk modeling on the smooth feature vector to obtain a fusion risk score; based on the questions, the knowledge points and the error cause labels, constructing an error cause link diagram, and obtaining an error cause link priority; constructing a candidate intervention task set; performing utility calculation on the candidate intervention task set to generate a constraint matrix; performing feasibility solution on the candidate intervention task set by adopting a multi-strategy king butterfly optimization algorithm, and searching an optimal task combination of a current period; performing result distribution on the optimal task combination of the current period to obtain a receipt set; and carrying out weight adaptive adjustment according to the receipt set, carrying out iterative loop until a preset convergence condition or a maximum number of iterations is reached, and outputting the intervened task combination.

The invention relates to the technical field of quantum key distribution, and particularly discloses a quantum key distribution-based power load management data transmission security method, which comprises the following steps of: firstly, deeply analyzing a power data packet, and marking security and service quality requirements of the power data packet according to a preset security service level protocol; meanwhile, the system carries out continuous perception and prospective prediction on the state of the quantum key pool, not only predicts a future key supply trend through a time sequence model, but also calculates an expected key gap in combination with a power business demand so as to quantify a future resource risk. And finally, the dynamic scheduling decision engine dynamically selects an optimal encryption mode, namely quantum key encryption or classic key encryption, for each data packet through utility calculation according to the security level of the data packet and the expected key gap. Thus, limited quantum key resources can be intelligently and preferentially allocated to key services, service congestion caused by misuse of the resources is avoided, and the resource utilization efficiency is optimized while the high security level is ensured.

The invention relates to the technical field of data resource matching, and discloses an end-to-end resource matching system and method based on differential privacy federated learning, and the system comprises a model aggregation module, a utility calculation module, a client first screening module, a model sorting module, a client second screening module and a resource distribution module. Constructing a kernel matrix based on the contribution values of the screened client models to calculate the calculation effectiveness of the kernel matrix, and carrying out secondary screening based on the effectiveness, the resource cost and the privacy budget; for each client, the expansion utility of the client is calculated according to the number of times that the client is selected in the previous iteration round, then each selected client is sorted, and corresponding data resources are distributed to all clients based on the sorting. According to the method, the resource output of each client is dynamically optimized and adjusted, so that the change sensitivity of multiple attributes of the client is reduced, accurate measurement is provided for real-time contribution of the client, and a reasonable resource allocation scheme is constructed.

The invention discloses an electric vehicle charging load prediction method and system fused with a multi-dimensional decision, and relates to the technical field of power load prediction, and the method comprises the steps: firstly collecting historical load, travel trajectory, traffic, dynamic electricity price, predicted waiting time, SOC and other original features; calculating the selection probability of each site by the user based on the weighted comprehensive utility; modeling the charging station into a graph structure, carrying out spatial weighted aggregation on node features by adopting graph convolution, fusing the node features with original features to form spatial-temporal features, and inputting the spatial-temporal features into LSTM to obtain a future load; and estimating a variance by using a residual error, and outputting point prediction and a confidence interval. According to the scheme, the flow migration and competition relation between the stations is captured through graph convolution, and misjudgment caused by single-point abnormity is remarkably reduced; utility functions can be explained, parameters can be calibrated, and cross-city and new site rapid migration is facilitated. A confidence interval is output in combination with uncertainty quantification, and a risk boundary is provided for power grid dispatching, charging queuing guidance and a price strategy.

The invention discloses a federated learning training node hierarchical selection method and system, and relates to the technical field of federated learning and edge computing cross, and the method comprises the steps: estimating the time delay of each edge node based on a time delay model, and dividing all edge nodes into a plurality of layers based on the estimated time delay; based on the utility of the edge node, calculating the comprehensive utility of each layer, and randomly selecting any layer as a time delay upper limit layer of the current round by taking the normalized comprehensive utility as a weight; in the edge nodes of the time delay upper limit layer and the layer with the time delay not exceeding the time delay upper limit layer, randomly selecting a set number of edge nodes to participate in the current round training by taking the utility of the normalized edge nodes as a weight; and the global model is distributed to the selected edge node, the edge node uploads the updated model, the center node aggregates all the updated models to generate a new global model, and iteration is performed until model training is finished. Edge node energy efficiency levels are considered, and efficient edge node selection is realized through dynamic layering and utility optimization.