13results about How to "Guaranteed global optimality" patented technology

The invention relates to the technical field of industrial intelligent manufacturing, and discloses a digital factory-oriented dynamic energy price response type production regulation and control system, which comprises a data acquisition module, a data processing module, a model establishment module, a scale regulation and control module, an instruction verification module, a visualization module and a closed-loop optimization module, the dynamic energy price is introduced into production scheduling, so that the comprehensive energy cost is reduced; a multi-agent reinforcement learning model is adopted for real-time response; an optimization strategy is decomposed into long-term, short-term and real-time instructions, macroscopic planning is carried out by using a cloud, and accurate control is executed by relying on edges; three defensive lines of safety, feasibility and stability are set for all control instructions, so that the harm of wrong instructions is effectively prevented; a decision view and a manual intervention interface are provided to ensure the reliability and credibility of the decision; the successful scheduling case is stored in a strategy library, online fine adjustment and offline training are carried out on the core model, and the continuous evolution ability is achieved.

This invention relates to the field of named entity recognition technology, and discloses a multimodal named entity recognition method and apparatus based on deep interaction of text and image features. The method includes: acquiring original text and an original image; extracting multimodal features from the original text and the original image respectively to obtain a text feature sequence and an image feature sequence fused with multi-scale features; performing cross-modal deep interaction fusion on the text feature sequence and the image feature sequence to obtain an interacted multimodal feature sequence; performing context-aware adaptive gating fusion on the interacted multimodal feature sequence to obtain a fused feature sequence; and performing conditional random field sequence decoding on the fused feature sequence to obtain an entity label sequence. This invention achieves deep bidirectional interaction and dynamic adaptive fusion of multimodal information, effectively solving the text ambiguity problem and improving the accuracy of named entity recognition.

This invention relates to the field of electrical digital data processing technology, specifically a TCR measurement method based on high and low temperature region data fusion using an improved particle swarm optimization algorithm. The method inputs the original time-series data from the high and low temperature regions, removes noise and outliers using isolated forest and Grubbs criterion, and maps the data to a unified temperature domain grid using piecewise Lagrange interpolation. An ergodic sequence is generated using Logistic chaotic mapping to initialize the particle swarm position vector, which is then transformed into node coordinate parameters in geometric space, and a cubic basis spline fitting curve is constructed. Particle fitness is evaluated using the sum of squared residuals and curvature smoothness as constraints. During iteration, the average Euclidean distance from the particle to the swarm centroid is calculated to generate a spatial dispersion factor. A multi-interval weighted decision maker based on the Mamdani framework is used to identify probe, development, convergence, and escape states, dynamically adjusting the inertia weights accordingly. This effectively solves the problems of data fragmentation and premature convergence, ultimately outputting a continuous and smooth resistance temperature coefficient distribution across the entire temperature domain.

The invention discloses a task planning method and device for a heterogeneous fixed-wing unmanned aerial vehicle cluster, equipment and a medium, and relates to the technical field of unmanned aerial vehicle cluster control, and the method comprises the steps: carrying out the task planning of the heterogeneous fixed-wing unmanned aerial vehicle cluster based on a current task target set corresponding to the heterogeneous fixed-wing unmanned aerial vehicle cluster, unmanned aerial vehicle type information and unmanned aerial vehicle constraint information; carrying out heterogeneous multivariate differential virtual potential field modeling, and based on a virtual potential field modeling result, analyzing a coupling relationship between heterogeneous unmanned aerial vehicles and a heterogeneous task target competition coupling relationship; and based on the obtained inter-potential-field coupling analysis result and the virtual potential field modeling result, the classification gradient of the heterogeneous total potential energy on the spatial position of the unmanned aerial vehicle is analyzed, and convergence analysis is carried out in combination with a preset Lyapunov function to determine target task allocation information currently corresponding to each unmanned aerial vehicle in the cluster, and the target task allocation information comprises a moving trajectory and a task target. According to the invention, unified planning of task targets and paths of heterogeneous fixed-wing unmanned aerial vehicles in a cluster is realized.

This invention relates to the field of multi-agent cooperative control technology, and in particular to a switching multi-agent cooperative tracking control method based on deep reinforcement learning. The method includes building a two-layer independent architecture for each independent agent, constructing a multi-level grouping architecture for the overall multi-agent cluster, proactively adjusting the internal structure of the multi-level grouping architecture based on prediction results, and using causal analysis technology to trace and determine the root causes of sudden changes and select an execution strategy. This invention adopts a two-layer independent architecture scheme, configuring a leader layer and an executor layer for each agent. Through an asynchronous decision-making mechanism with unidirectional instruction transmission, it achieves complete decoupling between global planning and local execution, ensuring the global optimality of the cooperative tracking task, meeting the real-time control requirements in dynamic environments, eliminating the single-point failure risk of centralized decision-making, and improving the cluster decision-making efficiency and operational stability.

The invention discloses a new energy positive and negative sequence voltage coordinated optimization method adaptive to high-power fault impact, which comprises the following steps of: constructing an optimal unbalanced voltage suppression model of a doubly-fed wind turbine generator under an asymmetric fault of a power grid, and taking the minimum voltage unbalance degree of a grid-connected point as a target function by the model to suppress the optimal unbalanced voltage of the doubly-fed wind turbine generator under the asymmetric fault of the power grid; the capacity constraint of the rotor side converter and the synchronous stability constraint of the system are included; based on the optimal unbalanced voltage suppression model, reconstructing a double-sequence synchronous rotating coordinate system; carrying out convexity processing on the reconstructed model, relaxing the capacity constraint of the rotor-side converter into a second-order cone form, and introducing an auxiliary variable to convert a target function and related constraints into a standard second-order cone programming problem; and solving the second-order cone programming problem, obtaining an optimal voltage dynamic control instruction, and controlling the new energy generator set. According to the method, the inter-sequence coupling effect can be accurately considered, so that the optimal suppression of the voltage imbalance is realized on the premise of ensuring the safety and stability of a machine-network system.

This application discloses a dynamic load power balancing control method for microgrids, relating to the field of power system automation and intelligent control technology. It includes: acquiring the power fluctuation amplitude, interruption duration, and priority dynamic decay rate of the load to form an initial feature set; quantifying the coupling degree between power fluctuation and interruption duration using fuzzy logic to determine a time-period fluctuation index; when the time-period fluctuation index exceeds the limit, calculating a weight adjustment coefficient by fusing multiple features using a neural network; searching for a target power supply sequence for high-weight loads using a genetic algorithm; updating load priorities through secondary fuzzy inference based on real-time power fluctuations on the supply side; calculating a response delay correction factor using a neural network; and adjusting the target power supply sequence to obtain the final power allocation scheme. This invention integrates multiple algorithms to achieve closed-loop control, improving the power balance and dynamic adaptability of the microgrid, ensuring stable power supply, and is applicable to microgrid load control scenarios.

The application relates to an aluminum template installation precision control method, system, device and medium. The method comprises the following steps: extracting the actual local characteristics of the installed aluminum template, matching the actual local characteristics with the predefined key local characteristics, and giving corresponding semantic labels; calculating six-degree-of-freedom deviation parameters and weight coefficients to construct a multi-objective optimization function; based on the function, based on the physical connection relationship and the physical stroke limit value, a mathematical optimization problem is constructed; the problem is numerically solved to obtain the optimal adjustment amount of each aluminum template unit, generate and execute the adjustment instruction, obtain the verification point cloud data of the adjusted aluminum template, and verify and iteratively optimize until the new structured deviation data set meets the preset precision requirement. The method constructs a multi-objective optimization model, combines the physical connection relationship and the stroke limit value, realizes high-precision and intelligent adjustment of the aluminum template installation, significantly improves the precision and efficiency of the aluminum template installation, and enhances the global optimization capability and intelligent level of the adjustment process.

The application discloses a kind of off-site computing power dynamic collaborative optimization methods based on multi-agent reinforcement learning, including the following steps: constructing resource topology diagram;Based on resource topology diagram, get node-level state feature vector and system-level state feature vector;Multi-agent environment is constructed;Local observation, global information and agent action set are input into improved CTDE model, output policy network parameters and value network parameters and construct training batch;Based on training batch, get converged policy network parameters and converged value network parameters;Get execution result;Obtain dynamically updated policy network parameters and dynamically updated value network parameters, realize the dynamic collaborative optimization of task acceptance, resource allocation, task migration, replica start-stop and bandwidth matching.

The present application belongs to the technical field of mesh model simplification, and particularly relates to a three-dimensional scene mesh model simplification method and system based on iterative search. The method comprises the following steps: selecting a mesh to be simplified; calculating a single shrinkage weight of each edge in the mesh to be simplified based on a shrinkage quadratic error and curvature; calculating a final shrinkage weight of each edge in the mesh to be simplified based on the single shrinkage weight of each edge in the mesh to be simplified through iterative search; selecting an edge with the minimum final shrinkage weight in the mesh to be simplified as a collapse edge to gradually simplify the mesh to be simplified until the compression rate of the mesh to be simplified meets the requirement; and reselecting the mesh to be simplified and repeating the above steps until the compression rates of all meshes meet the requirement. The present application can simplify the mesh while retaining the visual features of the three-dimensional scene mesh model.

This invention relates to the fields of photovoltaic new energy and industrial intelligent control technology, specifically a photovoltaic support system energy consumption modeling and control method based on industrial cloud computing. This method collects real-time operational and environmental data of the photovoltaic support cluster at the edge, uploads it to an industrial cloud platform to construct an energy consumption model that integrates static attributes and dynamic operating conditions, achieving accurate energy consumption prediction. With the dual objectives of maximizing power generation and minimizing operational energy consumption, cluster-level global optimization is completed in the cloud, generating an optimal tracking strategy and distributing it to the edge controller for execution. Through real-time data feedback, online iteration of the energy consumption model and closed-loop optimization of the strategy are achieved. This invention solves the problems of inaccurate energy consumption modeling, high energy consumption due to excessive tracking, and weak cluster collaboration capabilities in existing photovoltaic support system control, effectively improving the comprehensive power generation efficiency and full life-cycle management capabilities of photovoltaic power plants. It is applicable to complex application scenarios such as large-scale ground-mounted, mountainous, and distributed photovoltaic power plants.

The invention provides an MAS-based formation vehicle-mounted electric power interconnection system energy management method and system, and belongs to the technical field of formation vehicle-mounted electric power interconnection systems, and the method comprises the steps: constructing a multi-Agent technology control structure of a formation vehicle-mounted electric power interconnection system; constructing a micro-grid energy management and scheduling device for generating an energy management and scheduling scheme which is constructed according to the battery management parameters and the scheduling algorithm parameters and managed by a time sequence, and obtaining an optimal energy management and scheduling scheme through an improved litsea rotundifolia optimization algorithm; according to the improved Golden Litsea Optimization algorithm, a goose group optimization algorithm and a whale optimization algorithm are introduced into an exploration stage and a development stage of an original Golden Litsea Optimization algorithm; and based on the energy management and scheduling scheme, performing energy scheduling and management on the formation vehicle-mounted electric power interconnection system through the global energy management and scheduling Agent of the dominant micro-grid. According to the method, the optimal energy management scheme of the formation vehicle-mounted electric power interconnection system based on the MAS can be quickly generated.

The application provides an AI-based network resource optimization method, system and device, relates to the technical field of artificial intelligence, and solves the technical problem that the global decision accuracy is reduced due to the cross-level feature information transmission loss of network resources in the prior art. The method specifically comprises the following steps: acquiring bottom state data of a target network; inputting the bottom state data into a multi-level scheduling architecture, wherein the multi-level scheduling architecture comprises a plurality of functional levels in vertical cascade, and a direct connection channel connecting a bottom functional level and a high-level functional level across an intermediate functional level; transmitting original features generated by the bottom functional level based on the bottom state data to the high-level functional level through the direct connection channel, and fusing the original features with abstract features output by the intermediate functional level to generate global decision features; and generating a resource scheduling strategy for the target network based on the global decision features through an AI resource scheduling model. The application is used for optimizing network resources.