160 results about "Differential evolution" patented technology

The invention discloses a visual large model Token adaptive optimization method, system and device based on differential evolution and a medium, and the method comprises the steps: carrying out the data processing of an image classification data set, an instance segmentation data set and a saliency target detection data set, and obtaining all Tokens corresponding to each image through a Patch Embedding and position coding method; obtaining a plurality of groups of Tokens corresponding to each image through a random selection mode, and performing data processing to output all Tokens corresponding to each image and the plurality of groups of Tokens selected from each image; constructing a Token adaptive selection module, a self-attention optimization module and a downstream task output module; a complete Token adaptive optimization visual large model is constructed; training a reconstruction model and a complete Token self-adaptive optimized visual large model; performing model reasoning to obtain an image classification result, an instance segmentation result image and a saliency target detection result image; the system, the equipment and the medium are used for implementing the method. The method can be widely applied to various visual tasks such as image classification, instance segmentation and saliency target detection.

The invention designs a workshop dynamic scheduling method based on an improved genetic algorithm and multi-objective optimization. According to the method, in order to solve the problem that a scheduling scheme fails after dynamic events such as equipment failure, order insertion or material delay occur in a discrete manufacturing workshop, a greedy strategy is adopted for pre-scheduling, and rapid rescheduling is performed based on an improved genetic algorithm after the dynamic events occur. The algorithm improves search efficiency and scheduling stability through multi-population parallel evolution, differential evolution self-adaptive parameter adjustment and an elitist retention mechanism. And taking minimization of the maximum completion time, the total delay time and the equipment change frequency as multiple targets, and obtaining a comprehensive optimal solution through a weighted summation method. According to the method, the scheduling response speed and stability of the workshop in a dynamic environment can be remarkably improved, and the workshop production efficiency and the equipment utilization rate are improved.

The invention belongs to the technical field of machine learning models, particularly relates to a personnel performance evaluation method based on IWOA-SVM, and solves the problems that a traditional support vector machine (SVM) is low in precision, difficult in parameter selection and the like in performance intelligent evaluation. The method comprises the steps that Tent chaotic mapping and a pseudo-opposition learning strategy are utilized to increase the diversity and quality of an initial population, and the whale algorithm (WOA) is prevented from falling into local optimum; the global optimization capability of the WOA is improved by adopting a differential evolution mechanism; a penalty factor and kernel function parameters of the SVM are optimized through an improved whale algorithm (IWOA), and performance evaluation can be effectively carried out while optimal parameters are obtained. According to the method, the whale algorithm can be improved by using Tent chaotic mapping, pseudo-opposition learning and a differential evolution strategy, SVM parameters are searched in a global range, and better model performance is obtained.

The invention provides a cable testing method and a small handheld cable tester. The method comprises the following steps: carrying out wavelet transform noise reduction, Z-score standardization and time sequence dynamic regularization on original cable data; then constructing a geometric deep learning network by using a dispersed self-organizing structure of the rotating cube set, and extracting multi-modal depth features; fusing different modal features through a spiking neural network gating mechanism and a cross-modal attention mechanism; and finally, fault classification is carried out based on a differential evolution optimized neural network, and fault location is realized by combining a time sequence generative adversarial network and a dynamic probability neighborhood growth clustering algorithm. The system also evaluates the data contribution degree of each modal through information entropy and mutual information analysis, and optimizes the model performance by using adaptive weight distribution and lightweight neural network technologies. According to the invention, various fault types such as cable breakage, short circuit, insulation aging, poor contact and the like can be identified and positioned with high precision, and the efficiency and the accuracy of cable maintenance are remarkably improved.

The invention discloses a lake pollution traceability analysis method based on multi-source data assimilation and reverse diffusion, and relates to water pollution traceability analys.The lake pollution traceability analysis method comprises the steps that chemical pollution is monitored at the downstream of a lake, monitoring data are imported into a hydrodynamic force-water quality model after Kalman filtering, and spatial and temporal distribution data of pollutants in a water body are obtained; pollution source inversion is carried out based on a differential evolution Markov chain Monte Carlo method, and pollution source parameters are obtained; forward simulation is carried out, a flow velocity field and water depth are solved through a hydrodynamic force-water quality model, then a pollutant transport model is generated, pollution source parameters are used as input, a finite volume method discrete control equation is used based on the pollutant transport model to predict pollutant concentration space-time evolution, and pollution duration under natural attenuation is evaluated; and comparing real-time monitoring data with an expected effect, and evaluating a pollution traceability analysis effect. Pollution source positioning errors are reduced, pollution contribution rate calculation accuracy is improved, traceability efficiency is improved, the risk of excessive reagents is avoided, and treatment cost is reduced.

The invention relates to the technical field of satellite navigation signal processing, in particular to an integer ambiguity resolving method based on a hybrid adaptive differential evolution grey wolf algorithm, which comprises the following steps of: loading GNSS (Global Navigation Satellite System) original data to carry out decorrelation, and calculating an adaptive integer search space boundary; calculating the solution of each individual, storing, comparing the fitness of each individual, and updating and recording globally optimal, suboptimal and third optimal solutions; executing a variation-crossover-selection process of differential evolution to generate a new generation of population; parameter self-adaption and Levy flight are executed; checking whether the current number of iterations reaches a preset maximum number of iterations, if not, adding one to the counter, returning and repeating the steps of comparing the fitness of each individual, updating and recording the global optimal solution, the suboptimal solution and the third optimal solution, and if yes, starting local search to find a better global optimal wolf; and the final calculation result finally recorded in the global optimal wolf variable is output, so that the premature convergence problem in the traditional algorithm solution is solved, and the solution success rate is improved.

The invention provides a civil aviation fleet-oriented maintenance plan and work package integrated optimization method, which comprises the following steps of: firstly, taking minimization of total maintenance cost as a target function; constructing a mathematical planning model for integrated optimization of the fleet maintenance plan and the work package by taking a maintenance item number constraint, a maintenance item and maintenance work package corresponding relation constraint, a maintenance resource constraint, a maintenance interval constraint, a variable relation constraint, a variable value constraint and the like as constraint conditions; then, auxiliary variables are introduced to carry out linearization processing on the established mathematical programming model; and finally, solving the linearized mathematical programming model by adopting a differential evolution-adaptive large-scale neighborhood search two-stage heuristic algorithm, namely, continuously iteratively solving through a designed two-stage interaction strategy, so as to achieve the purpose of integrated optimization. According to the method, the dual goals of reducing the total maintenance cost and improving the maintenance scheme making efficiency are achieved, and a scientific solution is provided for the complex optimization problem in civil aviation maintenance management.

The invention provides an online defect detection method and system for a cable insulation layer, and relates to the technical field of cables, and the method comprises the steps: obtaining an extrusion monitoring data set based on cable production; insulating layer defect online detection is carried out according to the extrusion monitoring data set; performing trend prediction according to the insulating layer defect first map; performing feedback adjustment on the insulation extrusion scheme according to the insulation layer defect second map to obtain an extrusion adjustment space; performing defect risk optimization on the extrusion adjustment space to obtain an extrusion candidate population; and based on the multi-dimensional defect risk prediction model, performing differential evolution optimization on the extrusion adjustment space according to the extrusion candidate population to obtain an extrusion adjustment optimization result, and executing cable production online optimization according to the extrusion adjustment optimization result. According to the invention, the technical problem of low defect detection accuracy of the cable insulation layer in the prior art can be solved, and the technical effect of improving the defect detection accuracy of the cable insulation layer is achieved.

The invention discloses an optimization method and device based on a neural network model, a medium and a program product, and the method comprises the steps: (1) carrying out the modeling of an automobile rear subframe based on SFE-Concept, constructing a first-order modal maximization mathematical model according to an adaptive penalty function, generating an optimization population based on Latin hypercube, and carrying out the optimization of a first-order modal maximization mathematical model; performing first-order modal and rigidity simulation analysis on the optimized population in an Isight multidisciplinary optimization design platform; (2) generating an optimal candidate sub-population and a successful design variable vector through differential evolution based on a cubic kernel radial basis function machine learning model; (3) training the Dropout neural network model to obtain evolution parameters; (4) updating evolution parameters based on the Dropout neural network model; and (5) updating and optimizing the population based on the evolution parameters, if a convergence condition is met, outputting an optimal rear subframe, otherwise, returning to the step (2). According to the method, the evolution parameters are adaptively adjusted according to the Dropout neural network model, and the adaptability to the modal optimization problem of the rear subframe of the automobile is high.

The invention belongs to the technical field of medical data identification, and particularly relates to a depression identification method and system based on differential evolution feature selection. Obtaining original electroencephalogram data, and preprocessing the obtained original electroencephalogram data; based on the preprocessed electroencephalogram data, selecting an optimal feature subset by adopting a feature selection method based on differential evolution; and constructing an adaptive hierarchical fusion network model based on Transform, and training based on the optimal feature subset to obtain a depression identification reference model and a reference identification result. The purpose of performing feature selection on the electroencephalogram data is to remove redundant features and irrelevant features so as to obtain an optimal feature subset, and the screened feature subset is put into training, so that not only can the training time be saved, but also the recognition rate higher than that of original data training can be obtained.

The invention relates to a cloud and mist shielding method and device based on an unmanned aerial vehicle. The cloud and mist shielding method based on the unmanned aerial vehicle comprises the following steps: performing shielding amount analysis on current environment data to obtain information of a shielding area to be generated; performing cloud distribution analysis on the to-be-generated shielding area information by adopting a cloud distribution strategy library to obtain an accurate cloud distribution position; judging whether a track corresponding to the accurate cloud distribution position exists in the current voyage decision library or not: if not, sequentially performing unified coding, cost calculation and differential evolution on the current unmanned aerial vehicle according to the accurate cloud distribution position to obtain tracks corresponding to all the unmanned aerial vehicles; if yes, the formation reinforcement learning model is adopted to optimize the flight paths corresponding to all the unmanned aerial vehicles, the optimized flight paths are obtained and distributed to all the unmanned aerial vehicles, and the unmanned aerial vehicles are made to fly to the target position according to the optimized flight paths and release the camouflage cloud and mist. According to the cloud and mist shielding method based on the unmanned aerial vehicle, the concealment and safety of our military actions are enhanced.

The invention discloses a filtering antenna design method of a naked mole algorithm based on SHAP, and relates to the crossing field of a radio frequency front end of a wireless communication system and a computer algorithm. According to the method, the SHAP value is utilized to quantify the influence of the antenna structure parameters on the antenna performance, the search strategy in the NMRA is dynamically adjusted, and the exploration capability of the algorithm in a high-dimensional space is enhanced. Firstly, the SHAP value of each antenna parameter is initialized, and then the NMRA algorithm stage is entered. In the working stage, SHAP weighted differential evolution is carried out, and in the breeding stage, all breeders update their own states according to a certain probability, and breeders with poor performance are eliminated; self-adaptive variation is carried out according to the probability pm so as to prevent from falling into local optimum; and finally, evaluating the new population, updating the current optimal solution, and outputting the optimal solution after the maximum number of iterations is reached. The method is mainly applied to the fields of short-wave satellite communication, radars and the like, the limitation that a uniform search strategy is adopted for all features in the high-dimensional optimization problem of a traditional NMRA algorithm is overcome, and it is guaranteed that the performance of a filtering antenna is good.

The invention discloses a synthetic aperture radar online trajectory planning method based on an adaptive network. The method comprises the following steps: S1, establishing a bistatic SAR system model and a task coordinate system; s2, dividing a task space; s3, establishing a multi-objective optimization model; s4, performing independent evolution in each task block by using a joint multi-objective evolutionary algorithm of a block propagation strategy, and performing joint optimization through an inter-block elite individual sharing mechanism; s5, generating a trajectory sample data set; s6, introducing a cooperative differential evolution optimization mechanism based on a regularization extreme learning machine algorithm, and constructing a large-scale adaptive network; s7, carrying out real-time track prediction and online updating; and S8, outputting an optimal trajectory result meeting the constraint. According to the method, the problems of long time consumption, high calculation complexity and insufficient network generalization performance of a multi-objective evolutionary algorithm in real-time trajectory planning are solved, and the real-time performance and precision of trajectory planning are improved.

The present application relates to geological exploration technical field, disclose a kind of thick loess plateau area coalfield 3D seismic exploration method, comprising the following steps: S101, artificial seismic source shooting record and superficial Rayleigh wave detector data are collected;S102, grid division is carried out to target area, and set to be inverted parameter;S103, vertical ray path time estimation, multilayer medium dispersion forward and poststack energy estimation are carried out;S104, according to first wave arrival time observation value, Rayleigh wave observation dispersion curve, theoretical ray path time, phase velocity and poststack energy generates multi-target error function;S105, target area is divided into three subdomains, differential evolution iteration is carried out, until local convergence;S106, establish a plurality of parallel annealing chain, set chain temperature and gradually decrease by chain, optimize to be inverted parameter, output optimal to be inverted parameter;S107, the static correction time difference and absorption compensation coefficient of each acquisition point are calculated.The present application improves coal seam structure imaging precision and exploration data reliability under complex surface conditions.

The invention relates to the technical field of power equipment state monitoring and fault diagnosis, and discloses a fusion type high-voltage switch cabinet insulation state quantitative monitoring method and system, and the method comprises the steps: firstly obtaining a partial discharge ultrahigh frequency and sound wave signal, and calculating a generalized Renyi spectrum entropy and an entropy gradient index; carrying out nonlinear correction on the medium sound wave propagation velocity by using the entropy gradient index, establishing an equivalent sound velocity model in a medium degradation state, and obtaining a standardized equivalent time difference; then, constructing a two-dimensional phase plane state space taking the standardized equivalent time difference and the generalized Renyi spectral entropy as dimensions, and analyzing a differential evolution trajectory of a state vector on a time sequence; and finally, constructing a comprehensive state loss function, mapping to generate an insulation health degree index, and carrying out life prediction. According to the invention, through medium degradation wave velocity constitutive mapping and phase plane conjoint analysis, monitoring errors caused by medium aging are solved, and dynamic accurate quantitative evaluation of the insulation state is realized.

The invention discloses a permanent magnet synchronous motor locked-rotor temperature estimation method and device, and relates to the technical field of synchronous motors, and the method comprises the steps: collecting a preset number of motor states obtained under different locked-rotor working conditions and corresponding temperature data, and constructing a corresponding training data set; according to different motor cooling structures, constructing a motor stalling temperature estimation architecture based on the correlation between motor states and the influence factors of the motor states; a differential evolution training method based on fuzzy control uses the training data set to train the motor locked-rotor temperature estimation architecture to obtain a motor locked-rotor temperature estimation model; and obtaining a motor state of a to-be-estimated motor, and inputting the motor state of the to-be-estimated motor into the motor locked-rotor temperature estimation model to obtain the current temperature of the motor under the locked-rotor working condition. According to the invention, the problem that the accuracy is low when the stalling temperature of the permanent magnet synchronous motor is estimated in the prior art is solved.

The application relates to the technical field of battery life prediction, and discloses a lithium ion battery capacity prediction method, which comprises the following steps: obtaining lithium ion battery data, performing normalization processing and variational mode decomposition preprocessing on a battery capacity sequence to obtain a data set serving as model input; performing multi-scale decomposition on the battery capacity sequence obtained by S01 by using a self-adaptive mode selection mechanism based on an MAPE criterion, to obtain intrinsic mode functions (IMFs) of the best decomposition mode number; the lithium battery data is preprocessed by using variational mode decomposition (VMD), the battery capacity sequence is multi-scale decomposed by using a self-adaptive mode selection mechanism based on an MAPE criterion, and intrinsic mode functions (IMFs) of the best decomposition mode number are obtained; a self-adaptive step Gaussian random walk strategy, an auxiliary correction strategy and a differential evolution strategy are introduced to improve a white whale optimization algorithm (WOA), so that the robustness of the algorithm and the precision of the final solution are improved.

The present application relates to the field of photovoltaic technology, and particularly relates to a photovoltaic cell model parameter extraction method and related products, the method comprising: establishing a photovoltaic cell mathematical model, and determining model parameters to be optimized in the mathematical model; constructing a fitness function of the photovoltaic cell mathematical model; constructing a parameter optimization model based on a whale optimization algorithm and a differential evolution operator, and taking the fitness function as an objective function; executing the parameter optimization model, and obtaining an optimal parameter set corresponding to the model parameters to be optimized in the mathematical model; obtaining a final photovoltaic cell mathematical model through the optimal parameter set, and verifying errors of the final photovoltaic cell mathematical model; the present application firstly constructs a mathematical model of a photovoltaic cell, and then determines parameters to be optimized in the mathematical model, and then defines a fitness function based on multiple error indicators and physical parameter constraints, and then forms a parameter optimization model suitable for multimodal optimization by combining a whale optimization algorithm and a differential evolution operator, and finally obtains an optimal parameter set of the photovoltaic cell model.

The present application belongs to the technical field of medical data recognition, and particularly relates to a depression recognition method and system based on differential evolution feature selection. Original electroencephalogram data is acquired, and the acquired original electroencephalogram data is preprocessed; based on the preprocessed electroencephalogram data, a differential evolution-based feature selection method is used to select an optimal feature subset; a self-adaptive hierarchical fusion network model based on Transformer is constructed, and training is performed based on the optimal feature subset to obtain a depression recognition reference model and a reference recognition result. The purpose of feature selection on electroencephalogram data is to remove redundant features and irrelevant features, thereby obtaining an optimal feature subset. The screened feature subset is put into training, which not only saves training time but also obtains a higher recognition rate than original data training.

The application provides a gradient-oriented differential evolution method and system based on layer optimization difference, relates to the technical field of machine learning, and comprises the following steps: obtaining a neural network model to be trained; identifying key layers and non-key layers in the neural network model based on preset evaluation indexes; individual parameters of the key layers are stored in multiple groups of individualization, and shared parameters of the non-key layers are stored in a single group of sharing; according to the individual parameters of the key layers, a niche is divided by using the Euclidean distance, and the fitness value of each complete individual is evaluated; based on the fitness value, individuals in the niche are classified into optimal individuals, better individuals and poor individuals, and different updating strategies are adopted to update the parameters of the key layers; according to the optimal individual on the current verification set, the shared parameters of the non-key layers are updated by gradient descent; and when a preset iteration termination condition is met, a trained neural network model composed of the optimal individual parameters of the key layers and the shared parameters of the non-key layers is output.

A method and system for automatically generating prompts is disclosed. In some embodiments, the method includes providing user input to large language models (LLMs) utilizing meta prompting to generate a set of prompts represented by vectors. The method includes the LLMs identifying differential vector(s) from the vectors, mutating the vectors with the differential vector(s), and using first algorithm(s) to determine mutated prompt vector(s).The method includes generating an intermediate prompt by combining the mutated prompt vector(s) with the set of prompts and selecting a prompt vector using second algorithm(s). The method also includes dividing a task of validating the intermediate prompt into subtasks. The method further includes performing the subtasks by the software-based agents as part of a chain of thought (CoT) process to validate the intermediate prompt and outputting suggestion(s), and generating a final prompt by refining the intermediate prompt using the suggestion(s).

The application provides a topology optimization method and device for embedded chip cooling. The application comprises the following steps: obtaining multi-physical field distribution data and material distribution factors, and normalizing to construct a condition characteristic vector; inputting the characteristic vector into a pre-trained generative model, and obtaining a latent variable characteristic increment through low-dimensional latent space single-step denoising reasoning; obtaining a material distribution factor differential evolution amount through a variational autoencoder decoder reconstruction; superimposing the updated material distribution factor and topology configuration, and performing cyclic iteration calculation until convergence. The scheme improves the topology evolution efficiency and multi-physical field adaptability, and optimizes the iteration convergence speed.

This invention addresses the challenge of complex scheduling tasks in industrial manufacturing, where multiple tasks exist simultaneously with coupled constraints, leading to high optimization difficulty. It proposes an intelligent multi-task optimization method based on constraint coupling strength indices. This method aims to minimize the number of vehicles used and the total transportation cost. First, a knowledge graph network is constructed to store task information and constraints. Then, a constraint coupling strength index is calculated based on the relationship between constraints and decision variables. Subsequently, during the multi-task differential evolution search, candidate solutions are grouped according to constraint coupling strength, and differentiated cross-tabulation, mutation, and cross-task knowledge transfer strategies are employed to guide the search process towards efficient convergence. This method effectively characterizes complex constraint structures, enhances multi-task collaborative optimization capabilities, and provides an efficient and feasible optimization solution for transportation scheduling in industrial manufacturing.

The invention provides an aircraft slip-off time prediction method based on differential evolution and XGBoost, which comprises the following steps of: firstly, screening key slip-off time prediction data characteristics based on a sliding motion process of a civil aircraft in an airport scene in a departure process, acquiring corresponding data, and preprocessing to construct an aircraft slip-off time prediction data set; then, an XGBoost model is adopted as an aircraft slip-out time prediction model, and model hyper-parameters needing to be optimized and a search range of the model hyper-parameters needing to be optimized are selected; optimizing the selected XGBoost model hyper-parameters by using a differential evolution optimization algorithm to obtain an optimal model hyper-parameter set; and finally, training by using the optimal model hyper-parameter set to obtain an aircraft slip-off time prediction model. According to the method, the intelligent optimization algorithm is used for searching the optimal model hyper-parameters for the prediction model, so that the comprehensive prediction capability of the model on the aircraft slip-off time is improved.

The invention discloses a short baseline constraint optimization positioning method based on a magnetic moment vector modulus ratio, a measurement array and a system, and belongs to the technical field of magnetic detection and positioning. The method comprises the following steps: firstly, constructing a regular triangular prism-shaped measurement array, collecting magnetic field vector data generated by a magnetic target at each vertex measurement point of the array, and calculating a magnetic field vector at a coordinate origin of the measurement array and a vector mode gradient of the magnetic field vector; the spatial position coordinates of the magnetic target are used as decision variables, a constraint optimization model used for magnetic dipole positioning solution is established, and the target function of the constraint optimization model is a magnetic moment vector modulus ratio function; solving the constraint optimization model by adopting an adaptive differential evolution dung beetle optimization algorithm to obtain an optimal spatial position coordinate of the magnetic target; and based on the optimal spatial position coordinates, performing inversion calculation on three components of the magnetic moment vector of the magnetic target. According to the invention, the space volume of the positioning system is reduced, the positioning system is relatively simple, and the positioning error caused by the uncertainty of magnetic conductivity is also eliminated.

The invention discloses an industrial software service combination method based on an improved multi-objective grey wolf algorithm, and the method comprises the following steps: building an industrial software resource service combination optimization problem framework, and completing the evaluation of QoS indexes including service time, service cost, service reputation, service delivery quality and the like and availability indexes; an SCOS model with optimal service quality (QoS) and availability as targets is constructed; tent chaotic mapping and a Levy flight improved differential evolution strategy are fused into a multi-target grey wolf algorithm model, the improved multi-target grey wolf algorithm model is used for solving an SCOS model, and the effectiveness and feasibility of the method are verified through algorithm performance verification and example testing. According to the improved multi-target grey wolf algorithm model, the convergence speed can be effectively increased, manual combination is liberated, and the problem of local optimization of industrial software combination is effectively solved.

This invention provides a cable testing method and a small handheld cable tester. The method includes: performing wavelet transform noise reduction, Z-score normalization, and time-series dynamic normalization on the raw cable data; then constructing a geometric deep learning network using the dispersed self-organizing structure of a rotating cube set to extract multimodal deep features; next, fusing different modal features through a spike neural network gating mechanism and a cross-modal attention mechanism; finally, performing fault classification based on a differential evolution optimized neural network, and achieving fault location by combining a temporal generative adversarial network and a dynamic probabilistic neighborhood growing clustering algorithm. The system also evaluates the contribution of each modality of data through information entropy and mutual information analysis, and optimizes model performance using adaptive weight allocation and lightweight neural network technology. This invention can accurately identify and locate various fault types such as cable breaks, short circuits, insulation aging, and poor contact, significantly improving the efficiency and accuracy of cable maintenance.

The invention relates to the technical field of magnetic element modeling and control, in particular to a multi-factor magnetic core loss prediction method based on machine learning. The method comprises the following steps: firstly, acquiring magnetic flux density data of a magnetic element, extracting distribution characteristics and shape characteristics of the magnetic element, and realizing excitation waveform identification by constructing a classification model; secondly, introducing a temperature index correction term based on a Steinmetz equation, and establishing a temperature sensitive loss model; further performing importance evaluation on multiple factors by using an entropy weight method and statistical analysis; constructing a magnetic core loss prediction model by adopting an integrated learning model of a random forest and LightGBM; and with minimum loss and maximum magnetic energy transmission as targets, a single-target optimization model is established by fusing differential evolution and a simulated annealing algorithm. According to the method, the generalization ability and prediction precision of the model are improved, the magnetic core loss modeling requirements under the conditions of multiple materials, multiple waveforms and multiple working conditions can be met, and the method is suitable for application scenes such as high-frequency magnetic element design and power electronic system optimization.

The present application relates to the technical field of magnetic element modeling and control, and particularly relates to a multi-factor magnetic core loss prediction method based on machine learning. The method first acquires the magnetic flux density data of the magnetic element, extracts its distribution characteristics and shape characteristics, and realizes excitation waveform recognition by constructing a classification model; secondly, a temperature index correction term is introduced based on the Steinmetz equation to establish a temperature-sensitive loss model; further, the entropy weight method and statistical analysis are used to evaluate the importance of multiple factors; an integrated learning model of random forest and LightGBM is used to construct a magnetic core loss prediction model; and a single-objective optimization model is established by combining differential evolution and simulated annealing algorithm with the objectives of minimum loss and maximum magnetic energy transmission. The method improves the generalization ability and prediction accuracy of the model, can adapt to the modeling needs of magnetic core loss under multiple material, multiple waveform and multiple working condition conditions, and is suitable for application scenarios such as high-frequency magnetic element design and power electronic system optimization.

This invention relates to a method for optimizing process parameters of enameled wire stranding based on particle swarm optimization (PSO). The method includes collecting process parameters to construct a multi-objective prediction model, using a dynamic weight fitness function to adapt to different operating conditions, employing a hybrid PSO-ADE optimization approach, verifying the closed-loop and re-searching based on sensitivity, and outputting the optimal control signal. This PSO-based method for optimizing enameled wire stranding process parameters achieves nonlinear mapping by collecting process parameters to construct a multi-objective prediction model, setting dynamic weights and adaptive penalties to adjust the optimization objective according to operating conditions. It uses particle swarm optimization as the main loop, condition-triggered differential evolution to enhance diversity, balance exploration and development, introduces a verification closed loop, and re-searches based on the deviation direction and sensitivity when the target is not met. It determines variable priority based on sensitivity, granting greater search freedom. This method solves problems such as difficulty in describing multi-variable coupling, incompatibility of fixed weights with disturbances, premature convergence of single algorithms, prediction errors leading to non-compliance, and ambiguity in the direction of secondary optimization.