28 results about "Algorithm convergence" patented technology

A gas temperature field reconstruction method using thermocouple measurement correction is disclosed. First, an Inventor 3D model and an ANSYS temperature field model are established for the gas temperature field. Then, a mapping relationship is established between the control parameters of the gas temperature field and the boundary conditions of the ANSYS temperature field model. Based on the principle of maximizing the influence of fuzzy boundary conditions on the temperature field reconstruction results, experiments are designed for precise boundary conditions and experimental data are obtained. Next, given the boundary conditions of the ANSYS temperature field model, the model is run to obtain the temperature field reconstruction results. A model quality assessment algorithm is run to calculate the fitting degree between the measured curve and the simulation curve, obtaining the algorithm fitting parameters. The algorithm fitting parameters are selected to form the loss function of the gradient descent algorithm. The learning rate, termination condition, and initial iteration parameters are determined. The gradient descent algorithm is run until convergence. Finally, the final iteration parameters are taken as the correction result of the fuzzy boundary conditions of the ANSYS temperature field model. This invention improves the spatiotemporal resolution of the temperature field reconstruction results.

This invention discloses a method and system for LEO satellite communication resource allocation based on dynamic weighted graph segmentation, relating to the field of satellite communication technology. Addressing the co-channel interference problem in large-scale MIMO satellite systems, this invention models the resource allocation optimization problem to maximize system sum rate as a partitioning optimization problem on a complete user graph. The weights of the edges in the graph are configured to represent the potential interference cost when two users reuse the same time-frequency resource. The optimization objective is to maximize the sum of edge weights between different partition groups. A dynamic weighted graph segmentation algorithm is employed, utilizing physical layer channel quality index feedback to introduce auxiliary weight variables. The edge weights of the graph are dynamically updated iteratively, and graph segmentation is performed until the algorithm converges. Finally, time-frequency resource allocation is completed based on the graph segmentation results. This invention can achieve near-optimal solution performance with low complexity and fast convergence speed, significantly improving the sum rate in highly dynamic satellite communication environments.

This invention discloses a learning optimization method and system for solving quadratic programming problems, relating to the intersection of mathematical optimization and artificial intelligence. The method includes: inputting convex quadratic programming parameters, transforming them into a two-block separable form through auxiliary variables and indicator functions; initializing variables and LSTM states; generating iterative approximate solutions using LSTM, and determining whether sufficient descent and gradient conditions are met, triggering a gradient descent guarantee mechanism if these conditions are not satisfied; updating auxiliary variables element-wise, adaptively updating relaxation and penalty parameters, and outputting the original and dual solutions; and training the LSTM offline using identically distributed samples. This invention combines the fast inference of neural networks with the convergence guarantee of traditional algorithms, offering high real-time performance, strong robustness, and low resource consumption. It is suitable for scenarios such as high-frequency trading, autonomous driving, edge AI, and power grid scheduling, efficiently solving dynamic parameter convex quadratic programming problems.

This invention proposes a multi-frequency electromagnetic feature fusion method for abrasive particle identification, involving signal processing, sensor information fusion, and intelligent monitoring of mechanical conditions. It solves the problems of measurement distortion caused by hardware parasitic parameter coupling, the inability to directly solve the highly nonlinear time-harmonic field model analytically, and the poor convergence of conventional optimization algorithms leading to misjudgment of abrasive particle parameters in existing abrasive particle identification methods. This invention constructs a forward analytical model of the time-harmonic field and a nonlinear objective function, and utilizes the Levenberg-Marquardt (LM) optimization algorithm to jointly invert and extract the equivalent diameter, conductivity, and relative permeability of unknown metallic abrasive particles in a multi-dimensional parameter space. This invention can accurately decouple the equivalent diameter, conductivity, and permeability of abrasive particles, accurately distinguish materials, and the LM algorithm converges quickly and does not diverge, achieving millisecond-level inversion, meeting the high precision and real-time requirements of online monitoring of industrial oil.

The application discloses a method for cooperative jamming based on intelligent optimization algorithm, comprising: constructing a jamming decision model, the jamming decision model comprising a cooperative jamming decision matrix, a gain matrix, a jamming matrix, a jamming gain matrix, a jamming bandwidth ratio factor, a jam-to-signal ratio, and a jamming benefit; establishing an objective function and a constraint condition of the jamming decision model according to the jamming benefit; and using an artificial bee colony algorithm to take the jamming benefit as a fitness function and optimize the cooperative jamming decision matrix A. In different complex electromagnetic spectrum environments such as limited spectrum resources and the same frequency band shared by jamming devices and illegal users, the limited jamming resources are reasonably distributed under the condition that the jamming device of the own side can normally communicate, so that greater jamming benefit is achieved; the algorithm convergence speed and search ability are improved, and the method is helpful for making a decision with higher jamming benefit in a shorter time.

This application discloses a multi-strategy adaptive particle swarm optimization method and apparatus for dynamic parking space scheduling, belonging to the field of parking space scheduling technology. The method includes: cleaning flight and parking space data to construct a flight-parking space compatibility matrix; mapping particle positions using continuous real-number encoding and initializing the population through an OBL (Optimal Boundary Learning) strategy; adaptively adjusting the inertia weight AIW based on population diversity and combining it with a random migration RI (Increase in Randomization) strategy to avoid local optima; applying boundary constraints and conflict resolution to particle positions, and using a DA (Data Determination) mechanism based on congestion distance to preserve non-dominated solutions; and outputting the optimal parking space scheduling scheme through multi-attribute decision-making after the iteration meets the termination condition. This application improves the algorithm's convergence speed and global optimization capability through multi-strategy collaborative optimization, reduces the proportion of infeasible solutions, increases parking space utilization and flight docking rate, and balances passenger travel experience with airport operational efficiency.

The present disclosure proposes a network representation learning method across medical data sources, comprising: S1, generating medical network data comprising a source network and a target network; S2, randomly sampling a set number of nodes from the source network and the target network; S3, obtaining an L-layer neural network, and calculating the structural features and expression features of the source network and the target network respectively for each layer, and calculating the distance loss between the network features of the source network and the target network; S4, obtaining the output of the source network in the L-layer neural network, and calculating the loss value according to the classification loss and the distance loss, and updating the parameters of the algorithm according to the back propagation algorithm; S5, repeating steps S2-S4 until the entire algorithm converges, so that the accuracy of the algorithm for disease classification no longer rises within multiple iterations. The present disclosure considers the problem of inconsistent data distribution between different hospital data sources, and through the extraction of network structural information and node attribute information and the minimization of feature distance, the information loss is compensated, which has a wide application space.

The application discloses a dynamic programming-game hybrid method for solving satellite cooperative task scheduling, introduces a sequence constraint interval dynamic programming and an elite asynchronous updating mechanism on the basis of a traditional game theory cooperative method; a complex single-satellite best response problem is converted into a path optimization problem with side swing maneuver and illumination constraints through the sequence constraint dynamic programming, so that a task sequence with the maximum marginal contribution is obtained to improve the quality of local decision, and the algorithm has the ability to process complex space-time constraints; through the elite node strategy updating mechanism, system shock caused by multi-agent cooperative conflict is greatly reduced to improve the algorithm convergence speed, and the distributed network is more effectively guided to converge to a high-quality Nash equilibrium state; compared with a traditional greedy strategy or a heuristic algorithm, the method has better global optimization performance, the algorithm has faster convergence speed and stronger stability, and the cooperative scheme planned by the method has higher system total income and fewer resource conflicts.

This invention discloses a low-latency, low-communication-overhead navigation method for resisting sudden link interruptions, relating to the fields of wireless communication, cooperative positioning, and graph signal processing. The invention constructs a momentum-accelerated graph filter in a distributed network, accelerating convergence by introducing an inertial momentum term into node state updates, and adaptively adjusting the polynomial truncation order using residual energy detection to reduce latency and communication overhead. Simultaneously, a spatiotemporal two-dimensional trust region model is constructed, combining temporal consistency and spatial geometric verification to perform real-time trust scoring of neighboring node links. Finally, based on the trust score, a robust M-estimator based on the Huber kernel function is used to update the graph displacement operator. This invention solves the problems of slow convergence and high communication overhead in traditional algorithms, and can effectively cope with sudden link interruptions and measurement drift in highly dynamic scenarios, achieving high-precision, highly robust distributed cooperative navigation.

This invention relates to the field of construction engineering technology, specifically to a method and system for global optimization generation of engineering plans based on a multi-objective ant colony algorithm. The method includes initializing engineering and algorithm parameters, employing a non-uniform pheromone initialization strategy, constructing engineering activity paths based on an improved state transition probability formula, calculating path adaptability through a multi-objective fusion fitness function, iteratively optimizing using a dual pheromone update strategy (local and global), and outputting the optimal solution after verification and correction. The system adopts a modular design, supporting the entire process of method implementation. This invention achieves multi-objective collaborative optimization of schedule, cost, and resource utilization, improves algorithm convergence speed and optimization accuracy, adapts to various scenarios such as construction engineering and production line debugging, reduces implementation difficulty, provides reliable technical support for the efficient advancement of engineering plans, and possesses strong industrial application value.

This invention belongs to the field of wireless communication, specifically relating to a STAR-RIS-assisted 6G integrated sensing and communication beamforming design method. The method includes constructing a STAR-RIS-assisted integrated sensing and communication system model; under constraints of transmit power, sensing performance, and multi-user service quality, establishing an optimization problem aimed at maximizing multi-user communication and speed by jointly designing the base station transmit beamforming and the transmission-reflection coefficients of STAR-RIS; transforming the optimization problem into an easily tractable equivalent form based on a weighted minimum mean square error equivalent transformation; decoupling the optimization problem based on the equivalent form into two independent sub-problems using an alternating optimization method; solving the sub-problems using a successive convex approximation method and a semi-definite relaxation method, and iteratively updating the algorithm until convergence; this invention effectively improves system communication performance while ensuring sensing performance and multi-user communication quality.

This application discloses a U-Q-Learning-based mobile robot path planning method in unknown environments, comprising: designing a composite reward function including a repetitive movement penalty, a distance reward, and a state transition heuristic reward; initializing a Q-table, an action selection count table, and a state visit count table, as well as an exploration rate, a learning rate, and a discount factor; calculating the upper confidence bound of actions based on the Q-value of the state, the number of action selections, and the number of state visits, and dynamically adjusting the exploration rate; if the current exploration rate is large, randomly selecting an action from the action space; otherwise, selecting the action with the largest upper confidence bound; the robot executing the selected action, transitioning to the next state, and calculating the immediate reward; adaptively adjusting the learning rate, updating the current Q-value by combining the immediate reward, the discount factor, and the maximum Q-value of the next state, updating the state to the next state, repeating multiple iterations until the algorithm converges, and outputting the optimal path based on the optimized final Q-table. This application can improve the efficiency and stability of path planning.

The application discloses a multi-unmanned aerial vehicle (UAV) cooperative sensor multi-modal task allocation method, and belongs to the technical field of UAV cluster cooperative control. The UAV platform, multi-modal sensor and cooperative mode thereof are subjected to system modeling; a decision variable is designed to mathematically represent a platform-sensor-mode-role-task allocation relationship, and a multi-target optimization function is constructed; constraint conditions are defined; a chromosome coding strategy of coupling multi-agents is adopted to map the allocation scheme into a chromosome, and an initial population satisfying the constraint is randomly generated; selection is performed according to a selection operator, and a crossover operator and a mutation operator are executed to explore a new solution; a probability adjustment operator is used to dynamically and adaptively adjust a crossover probability and a mutation probability; when a maximum iteration number is satisfied, a current optimal chromosome is output, and after decoding, a final task allocation scheme which can be directly executed is obtained. The application has refined resource scheduling, intelligent adaptation in concealment, and high algorithm convergence and robustness.

The application provides a distributed solving method for a multi-region interconnected system with carbon emission constraints, which has the characteristics of including multi-region interconnected system low-carbon economic dispatching model construction, low-carbon economic dispatching model decomposition, information network topology structure building, distributed iteration optimization solving framework design and evaluation index establishment, and evaluating the algorithm convergence and system emission reduction effect. The method is scientific and reasonable, has strong applicability and good effect, and the like.

The application discloses a kind of considering rubber-oil fluid-structure interaction hydraulic elastic support structure design method, the full parameterization geometric model of hydraulic elastic support is established, the performance prediction of hydraulic elastic support is carried out, parameterization analysis workflow is constructed on simulation platform, optimization problem is defined, initial sample points are generated in design variable space using experimental design method, input-output data based on all sample points are used to construct response surface approximation model, and multi-objective optimization algorithm is used to iterate optimization on response surface, and response surface prediction result is obtained;After optimization algorithm converges, the output pareto front solution set is selected, the candidate optimal design is selected, high-fidelity verification analysis is performed on the selected optimal design point, a complete geometric model is regenerated, and the final optimization design of the hydraulic elastic support is obtained.The application can significantly improve the accuracy of hydraulic elastic support performance prediction, realize the collaborative optimization of stiffness, strength and other performance indicators, and shorten the product development cycle.

This invention relates to the field of communication network technology, specifically to a slice-aware dynamic resource allocation method based on deep reinforcement learning. By constructing a system model that integrates wireless channel characteristics and multi-slice service requirements, the differentiated quality of service requirements for the coexistence of eMBB and URLLC are formalized as a joint optimization problem aimed at maximizing system service satisfaction. Through the design of slice-aware states, actions, and rewards, the method achieves coordinated dynamic allocation of time-domain, frequency-domain, and power resources. This solves the problem that in existing technologies, when considering resource allocation in the time, frequency, and power domains simultaneously, the action space dimension is often too high, leading to difficulties in model training, slow algorithm convergence, and potentially the curse of dimensionality. The method achieves adaptive allocation of multi-service resources in a dynamic network environment, balancing spectral efficiency, user fairness, and differentiated quality of service assurance.

The application discloses a kind of composite briquette adding phase geometric structure and space distribution optimization method, adopt ellipsoid equation to simultaneously describe round ball type, whisker type, sheet type filling method, realize the optimization design of different geometric shapes;Specific steps are as follows:1, give the relevant condition of briquette;2, establish the r-θ2D section analysis model of briquette;3, given the discrete geometric number of adding phase;4, generate corresponding ellipse in model using random algorithm;5, determine whether the ellipse generated in 4 has problem, if there is problem, repeat 4;6, repeat steps 4, 5, generate M sets of data containing random geometry, space distribution;7, carry out mesh division;8, use numerical analysis method and solve in parallel, obtain M groups of key variables;9, based on the result in 8, using genetic algorithm and other optimization means, evaluate the pros and cons of M groups of design, then generate M groups of data again in combination with steps 4-6;10, repeat steps 4-9 until the optimization algorithm convergence condition is reached;11, output optimization result.

This invention discloses a method and apparatus for optimizing parameters of lightweight hydrogen storage cylinders based on adaptive optimization. The method includes: constructing a multi-agent collaborative optimization framework to define the functional boundaries of ant colony agents, NSGA-II agents, quantum annealing agents, and surrogate model agents, and clarifying the information interaction rules between agents; performing parameter collaborative optimization based on the multi-agent collaborative optimization framework according to collected design parameter samples and corresponding performance data of the hydrogen storage cylinder; dynamically adjusting the collaborative weights and optimization strategies of each agent by real-time monitoring of surrogate model prediction errors, population diversity, and algorithm convergence speed; selecting candidate solutions from the final Pareto front for real-world simulation verification; if the performance meets the design requirements, outputting the optimal parameter combination for the hydrogen storage cylinder; if not, feeding back the simulation results to the surrogate model agent for updating, and re-performing the iterative optimization process until the optimal parameter combination that meets the requirements is obtained.

The application discloses a sparse MIMO array design method based on genetic optimization, which is used for realizing high-resolution virtual array construction under the condition of reducing the number of transmitting and receiving array elements, and comprises the following steps: firstly, multiple geometric subarray structure constraints are constructed according to array performance evaluation indexes; then, population and system parameter settings are initialized; secondly, a multi-objective constraint model is established with the optimization target of maximizing fitness; subsequently, the model is solved by using an improved genetic algorithm until the algorithm converges, and finally, a sparse MIMO array structure satisfying the performance optimization and the quantity constraint is obtained. Compared with the prior art, the application enhances the controllability of the array structure by introducing the geometric subarray constraint, and guarantees the expected aperture form of the virtual array; the improved genetic algorithm can realize global search under the condition of fixed antenna number, and avoids local optimization; compared with the traditional method, the application has good engineering implementation and algorithm universality.

This invention relates to the field of scheduling technology for cloud computing and distributed computing, specifically to a serverless job scheduling method based on a large language model. It includes: determining the constraints of a solution and a method for calculating the fitness value of the solution; initializing and generating an initial population using a hybrid initialization strategy; constructing prompt words containing example code; calling program code from different large language models to generate multiple different candidate local search strategies, and performing syntax analysis and interface checks; selecting the optimal local search strategy; sequentially performing selection, crossover, and mutation operations on the current population to generate new solutions, updating the current population and the current global optimum; executing the optimal local search strategy on the current global optimum, updating the current global optimum; and determining the running time. This invention has the positive effects of automating the generation and optimization of local search strategies, and improving the convergence speed and solution quality of the algorithm.

This invention discloses a method, equipment, and readable medium for offshore wind farm layout targeting a single type of wind turbine. Belonging to the field of offshore wind farm planning, design, and intelligent optimization algorithm technology, this invention focuses on minimizing the levelized cost of electricity (LCoE) over its entire lifecycle. It employs a Gaussian wake model for accurate and efficient evaluation of wake effects. A soft-constraint penalty mechanism integrates wind farm engineering constraints into the objective function, preserving gradient information to explore high-quality solutions at the feasible region boundary. A multi-strategy guided mutation operator incorporating engineering knowledge is designed, coupled with an active population diversity restoration mechanism, to achieve full-process optimization of wind turbine layout. This invention, using the aforementioned method, equipment, and readable medium for offshore wind farm layout targeting a single type of wind turbine, can improve algorithm convergence speed and search efficiency, adapt to irregular site boundaries in offshore wind farms, and enhance the overall lifecycle economics of offshore wind farms.

The invention discloses a casting polishing robot trajectory optimization method based on a new GWOA algorithm, and the method comprises the steps: collecting and preprocessing a casting point cloud, and extracting a flash burr region as a polishing target; generating an initial trajectory based on the target point coordinates, and determining a robot motion trajectory through a visual odometer; synchronizing the six-dimensional force sensor and IMU data, and performing timestamp alignment and EKF fusion to obtain a robot pose and polishing force estimation value; the total grinding time consumption, the surface roughness and the grinding head abrasion loss are calculated through combination of the number of grinding points, a grinding force model and an Archard formula; taking the weighted sum of the three as an optimization target, solving an optimal trajectory from an initial pose by using a genetic whale hybrid algorithm under trajectory and process constraints, and completing polishing in sequence; according to the method, the positioning and modeling precision of the casting polishing robot can be improved, the multi-target balance effect is optimized, the algorithm convergence speed is increased, and the problems of local optimum and slow convergence of a traditional algorithm are solved.

The application discloses a mechanical arm path planning method based on an improved fast extended random tree and belongs to the technical field of industrial robot control. According to a set safety distance, a collision-free six-axis joint value sequence set from a starting point to an end point is quickly obtained. In initial path searching, a space preset tree strategy is used to guarantee the connection between an explored mechanical arm pose and initial and final mechanical arm poses and to make different poses as possible as to spread in space; in a path searching algorithm, a bidirectional search tree algorithm, a target bias, a greedy strategy and a variable step length strategy are used to accelerate algorithm convergence; a loop pruning algorithm is used in path optimization to greatly reduce a redundant path and to guarantee the local optimality of the path in an iteration process. The method improves the path planning efficiency of the mechanical arm in a three-dimensional space, solves the problems of long kinematics solving time of the mechanical arm in a multi-dimensional space and a redundant path and realizes a collision-free motion task in a joint space.

This invention discloses a reactive power optimization method for distribution networks based on the accelerated gradient alternating direction multiplier method, relating to the field of power system operation and control technology. The technical solution includes the following steps: S1, acquiring the topology and node operation data of the distribution network, and constructing an improved similarity matrix; S2, dynamically partitioning the distribution network using an adaptive spectral clustering algorithm; S3, establishing a local reactive power optimization model for each sub-region obtained from the dynamic partitioning; S4, introducing globally consistent variables for the voltage amplitude and phase angle of boundary nodes, establishing consistency coupling constraints between regions, constructing the augmented Lagrangian function for each region, and solving the local optimization problem; S5, updating the globally consistent variables using an accelerated gradient strategy; S6, calculating the original residual and dual residual. This invention achieves dynamic partitioning of the distribution network through adaptive spectral clustering, significantly improving the algorithm's convergence speed, effectively reducing system network losses, and eliminating the risk of node voltage exceeding limits.