163 results about "Virtual sample" patented technology

The invention relates to the technical field of industrial data, discloses a mining equipment state monitoring method and system based on the Internet of Things, and effectively solves the problem of insufficient model generalization ability caused by scarcity of fault samples of mining equipment. The virtual sample generation technology expands the available training data volume by 3-5 times, so that the early fault detection rate is improved to 85% or above. The self-adaptive feature selection mechanism reduces the consumption of computing resources by more than 30%, and maintains the integrity of key fault features at the same time. The multi-stage early warning system realizes accurate grading of fault severity, so that the maintenance resource distribution efficiency is improved by about 40%. The closed-loop optimization mechanism enables the model to continuously evolve in the operation process, and the annual false alarm rate is reduced by about 15%. The explainable diagnosis report provides a clear technical basis for field maintenance, and the average troubleshooting time is shortened by about 50%.

The invention discloses an island-shaped frozen soil degradation prediction method based on remote sensing and ground temperature coupling, and relates to the technical field of frozen soil degradation prediction in cold regions, and the method comprises the following steps: S1, multi-source data fusion and intelligent decomposition, S2, cross-scale dynamic parameterized model construction, S3, quantum enhanced Bayesian multi-scale calibration, and S4, three-dimensional dynamic uncertainty quantification and prediction output. According to the method, the problems of frozen soil data scarcity and remote sensing conversion precision are solved through multi-source remote sensing and actually measured data fusion in combination with CGAN, AM-PFN and virtual sample closed-loop expansion; the defects of a traditional model are overcome through a microscopic-macroscopic coupling model, DRL parameter switching and meta-learning algorithm selection; through quantum enhanced Bayes, DMD and a four-dimensional uncertainty matrix, cross-scale prediction divergence is reduced; and through a carbon release risk correlation model and a co-evolution product, a prediction dimension is expanded, and a support is provided for ecological protection of the cold region.

The invention provides a laser cutting parameter estimation and quality feature optimization method, and belongs to the technical field of laser cutting. The laser cutting parameter estimation method comprises the steps that machining parameters and quality features corresponding to the machining parameters are collected, a data set is constructed, and the data set is used for training a constructed machining quality prediction model; processing parameters input by a user are obtained, the processing parameters are input into the trained processing quality prediction model to obtain quality feature prediction values, the optimal quality feature prediction value is screened, and a corresponding optimal processing parameter combination is determined; and the cutting process is controlled according to the optimal machining parameter combination. Machining parameters are accurately optimized, cutting defects are reduced, and the laser cutting quality is improved; rapid pre-estimation and parameter optimization are achieved, trial and error are avoided, and the machining efficiency is improved; material waste and reworking are reduced, the equipment debugging time is shortened, and the production cost is reduced; the cutting process is ensured to be stable and reliable through virtual sample piece visualization, and a user can visually evaluate the cutting effect.

The invention discloses a crop spectrum inversion model adaptive construction method. The method comprises the following steps: analyzing field actually-measured water nitrogen dry matter data and a crop physiological coupling rule, constructing a water nitrogen dry matter physiological constraint feasible region in a three-dimensional state space, generating a virtual sample, and fusing the virtual sample with the actually-measured water nitrogen dry matter data to construct a physical consistent training sample set; based on the combination of the moisture level and the nitrogen level, constructing a water-nitrogen cooperation state category, independently determining a spectral feature subspace, and training to generate a water-nitrogen cooperation state submodel with independent parameters; and judging the category of the target water-nitrogen coordination state, routing to the corresponding water-nitrogen coordination state sub-model, and resolving to obtain a pixel water-nitrogen instant inversion result. According to the method, the problems of model generalization and physical consistency under small samples are solved through physiological mechanism constraints, the complex influence of water-nitrogen interaction stress on the spectrum is decoupled through a state typing mechanism, and the accuracy and robustness of crop water-nitrogen inversion are improved.

The invention discloses a visual detection optimization control method, device and equipment based on digital twinning and a storage medium, and relates to the technical field of visual detection, and the method comprises the steps: obtaining an analog image and a real image in a digital twinning environment, and carrying out the frequency domain feature transformation of the analog image and the real image, thereby obtaining an image spectrum feature; a frequency domain alignment model is established based on multi-band spectrum envelope guide residual mapping, and the structure of virtual and real image spectrum features is kept aligned; further extracting features through multi-scale convolution and channel dependence mapping to obtain virtual-real fusion features; quantifying channel similarity and establishing a covariance regularization constraint, performing channel correction on the cross-domain features, and eliminating feature drift to obtain second virtual-real fusion features; and finally, performing visual detection and micro defect identification based on the features. The problem that a virtual sample and a real sample are different in local texture structure and channel distribution is solved.

The invention belongs to the technical field of visual recognition, and provides a multi-target visual recognition method and system in a dynamic scene, and the method comprises the steps: collecting a multi-target recognition image in real time in the dynamic scene, arranging the multi-target recognition image into a multi-target time sequence image sequence, constructing a background model of the multi-target recognition image, and carrying out the recognition of the multi-target time sequence image sequence; combining with dynamic background interference compensation based on optical flow field calculation, generating a moving target mask matrix, generating a target candidate frame, intercepting an image in the target candidate frame, marking the image as a moving target image, constructing and training a multi-target visual recognition model introduced into a channel space double-attention module, and recognizing the moving target image; and generating an information list of the moving targets in the moving target image according to an identification result, taking the moving targets as nodes, constructing a space-time diagram model containing the nodes and edges, realizing time sequence moving target association, judging whether fracture nodes appear or not, and if yes, constructing a conditional diffusion model to generate virtual samples as the fracture nodes to be matched with connected nodes.

The invention provides a virtual sample generation method and system based on a non-stationary neural network Gaussian process, and the method comprises the steps: obtaining original data, judging the non-stationarity, and judging whether a statistical characteristic changes with an input position or not; if not, a hidden variable model fusing the neural network and the Gaussian process is constructed, and a hidden variable space representing non-stationary distribution is obtained through training; sampling based on hidden variable space probability distribution; using the model to generate = (, Z) (Z, Z)-1X through non-stationary high-dimensional mapping, which is a non-stationary kernel function, Z is a low-dimensional hidden variable, and X is an input variable; checking the consistency of the high-dimensional virtual samples and the original data, and screening a virtual sample set meeting statistical consistency; according to the method, high-quality and high-diversity virtual samples can be generated in small sample and non-stationary scenes.

The invention discloses a monitoring camera image anomaly detection method and system, and relates to the technical field of intelligent monitoring, and the method comprises the steps: obtaining an original video stream, extracting a pixel matrix of a current frame, carrying out the joint coding of the pixel matrix and a motion vector field of an adjacent frame into a quantum state bit vector, inputting the quantum state bit vector into a pre-constructed anomaly detection operator, and carrying out the detection of the anomaly. Executing quantum state evolution calculation in the Hilbert space, and outputting a two-dimensional distribution matrix of each pixel region; integrating the marked binary image and the physical verification conclusion, and when the quantum anomaly probability value reaches the alarm standard and passes the physical verification, outputting an anomaly alarm signal; and the abnormal alarm signal is converted into a confrontation training sample, a cross-scene virtual sample is generated through quantum noise injection, and an abnormal detection operator is dynamically updated. According to the method, the Hilbert space quantum state evolution step is combined, unitary transformation is applied to the quantum state bit vector, the orthogonal projection operation is executed, the accurately quantified abnormal probability distribution matrix is generated, and accurate assessment of the abnormal risk is achieved.

The invention discloses a structure construction deviation automatic identification and safety evaluation method, which comprises the following steps: acquiring multi-modal original signals such as space measurement, environment, equipment operation and images from multiple positions of a structure construction site, and establishing a trend drift feature distribution model through time sequence alignment, normalization, noise reduction and feature extraction; then through combination with generative artificial intelligence, virtual samples of historical and extreme working conditions are expanded, and construction of a mixed working condition data pool is realized; through distribution drift analysis and multi-target parameter optimization, a deviation detection threshold value is dynamically and adaptively adjusted, parameters and a judgment result are pushed to a digital twinning and safety control platform in real time, and meanwhile, whole-process log and performance archiving are carried out, so that the sensitivity, adaptability and system safety of structure construction deviation detection are improved, and the construction safety is improved. And the method has relatively high data fusion efficiency and algorithm generalization ability.

The invention provides a gear tooth surface precision detection method based on a deep learning network, a medium and equipment, and mainly solves the problem of insufficient model generalization ability caused by scarcity of real labeled data and domain difference in an existing method. The method comprises the following steps: collecting a plane workpiece image marked with a roughness level as a source domain data set, and pre-training an intelligent detection model; based on a gear shaping machining meshing motion relation, establishing a tooth surface morphology geometric simulation model, and generating virtual gear tooth surface images of different roughness grades; extracting a real gear tooth surface area image as a target domain sample; a domain adaptation method based on clustering optimal transmission is adopted, the feature distribution difference between a source domain and a target domain is calculated and minimized, model parameters are optimized, and self-adaptive migration of a feature space from a plane image to a gear tooth surface is achieved. According to the method, virtual sample generation and a domain adaptation technology are combined, so that the data acquisition cost is effectively reduced, and the detection precision and the model generalization ability under the small sample condition are remarkably improved.

The embodiment of the invention belongs to the technical field of network operation and maintenance, and particularly provides a network operation and maintenance strategy determination method and related device.In the determination method, processed static features and dynamic features are input into a heterogeneous feature fusion network containing a feature importance real-time calibrator, and a quantized network state vector is obtained; generating a scene feature vector by using a scene feature enhancement module based on the quantized network state vector; inputting the scene feature vector into a multi-model dynamic fusion module to generate a candidate operation and maintenance strategy; and the multi-model dynamic fusion module performs tracking on the candidate operation and maintenance strategies, calculates a strategy confidence score, and adjusts parameters of the multi-model dynamic fusion module through the closed-loop feedback tuning module until a target strategy meeting network operation and maintenance requirements is generated. According to the embodiment of the invention, through dynamic feature processing, virtual sample generation and multi-model fusion, the adaptability and generation efficiency of a network operation and maintenance strategy to a complex scene are remarkably improved.

The invention relates to the technical field of hydrogen energy storage, in particular to an underground hydrogen storage system, which is characterized in that a hydrogen storage sensing module and a processing module are integrated in the system, a multi-input multi-output monitoring channel is constructed by using array type electrode nodes, and the processing module constructs a dual-drive prediction model through superposition of a physical state space and a neural network to obtain reference hybrid prediction output; generating a virtual sample set by using random sampling to construct a dynamic random tolerance envelope band; and finally, leakage is judged according to the permeability of the envelope band according to the actually measured signal. According to the invention, the problem of high false alarm rate of the system due to the fact that in-pipe nonlinear background triboelectric charge noise covers tiny leakage characteristics under a fluctuating power supply charging working condition is effectively solved; millisecond-level self-adaptive accurate diagnosis of the air tightness of the pipeline in a complex dynamic environment is realized, and the detection capability of weak leakage signals is remarkably improved while electromagnetic interference is filtered out.

The invention discloses an elevator steel belt damage detection and quantitative analysis system based on eddy current and magnetic flux leakage dual-mode detection, and aims to solve the problems of single-mode information loss and serious industrial field strong noise interference in the existing steel belt detection. The system synchronously integrates an eddy current sensor, a magnetic flux leakage sensor and an encoder through a multi-probe array adapter; and multi-dimensional damage physical information in the steel strip is obtained. An improved wavelet packet transform-empirical mode decomposition (WPT-EMD) collaborative noise reduction algorithm is adopted, and in combination with a sub-band energy entropy and a self-attention mechanism, non-stationary mechanical noise is effectively filtered out. A multi-rule feature extraction engine is used for extracting smooth residual errors, derivative mutation and other features, a support vector machine (DE-SVM) model introducing physical priori knowledge weights is constructed, and the small sample recognition problem is solved in combination with a virtual sample generation technology. The method can realize high-precision positioning and quantitative evaluation of steel strip damage under complex working conditions, and has the characteristics of strong anti-interference capability, high identification accuracy and good generalization performance.

The invention provides an industrial assembly action recognition and error correction training method based on deep learning, and the method comprises the steps: achieving the synchronous collection of the whole assembly process through the deployment of a multi-mode perception collection system; a standard assembly database is constructed, a three-dimensional simulation platform is introduced, random disturbance is applied through a parameterized digital human body model, field disturbance is simulated, and a virtual sample enhancement training set is generated; based on real and virtual samples, constructing a deep neural network model, fusing a graph convolutional network, an LSTM and a convolutional network to extract skeleton, trajectory and mechanical features, and realizing assembly action recognition, anomaly detection and key frame positioning; and an assembly quality scoring function is constructed through the track consistency, the attitude offset and the mechanical anomaly index, multi-stage error correction feedback is generated, and personalized training is guided. The method is high in recognition precision of errors in assembly actions of assembly operators, can achieve timely and intelligent feedback, and is suitable for assembly quality optimization and skill training in an industrial scene.

The invention discloses a riverway water quality prediction method based on generative data enhancement, and the method comprises the steps: (1) arranging automatic water quality monitoring stations on each monitoring section of a riverway, collecting multi-site and multi-index historical water quality data, combining with meteorological driving factors, sorting according to a time sequence, preprocessing, constructing a standardized multi-dimensional sequence, and obtaining an actual measurement sample; (2) establishing a water quality time sequence generation model based on a conditional variation auto-encoder, and generating a virtual sample with time-space consistency; (3) physical constraint and distribution consistency screening are applied to the virtual samples generated in the step (2), unreasonable data are removed, and high-quality virtual samples are obtained; (4) combining the high-quality virtual sample and the actual measurement sample to form an enhanced data set, and re-dividing the enhanced data set into a training set, a verification set and a test set; and (5) training a water quality prediction model based on the enhanced data set, and outputting a future multi-time-interval water quality index prediction result.

The invention relates to the technical field of reservoir prediction, and discloses an oil-containing sandstone prediction method based on a CNN-BILSTM hybrid model, and the method comprises the following steps: data preparation: carrying out the standardization processing of seismic data; constructing a three-dimensional sample library, and realizing sample equalization through virtual sample generation and dynamic boundary adjustment; hybrid model construction: constructing a deep learning model comprising a multi-scale convolutional layer, a bidirectional LSTM layer and a cross-modal attention fusion layer; model training and optimization: carrying out model training by adopting a dynamic batch strategy and a mixed loss function; and performing whole-region prediction, inputting seismic data into the trained model, outputting probability distribution of the oil-containing sandstone, and performing post-processing to generate a final prediction result. According to the method, a three-dimensional sample library with strong geological representativeness is constructed through a virtual sample generation and dynamic equilibrium strategy of geological-geophysical cooperative constraint, the problems of actual sample sparsity and category imbalance are solved, and the sensitivity and discrimination capability of the model to hidden oil-bearing sandstone under a complex geological background are remarkably enhanced.

The invention discloses a small sample SOC prediction enhancement method and system based on a generative adversarial network, and belongs to the technical field of battery state-of-charge prediction. The method comprises the following steps: a small sample data acquisition and fine preprocessing stage: acquiring multi-source real-time state data and label data, and preprocessing to obtain a standardized real small sample data set; a GAN model construction and adversarial training stage: generating a virtual sample data set under an extreme working condition through adversarial training; an enhanced data set construction and SOC prediction model training stage: constructing an enhanced data set, and training an SOC prediction model based on LSTM by using the enhanced data set; and a real-time SOC prediction stage: preprocessing battery data of a to-be-predicted battery collected in real time, splicing the battery data according to a time sequence window, inputting the spliced battery data into the trained SOC prediction model, and outputting a real-time SOC prediction value. According to the method, the GAN generator is combined with the fault data seed and the noise disturbance, so that the virtual sample conforming to the physical law can be generated, and the extreme working condition coverage is greatly improved.

The invention is applied to the technical field of industrial process control, and particularly discloses a multi-view image feature fusion multi-mode process virtual sample generation method, which comprises the following steps of: S1, collecting sensor data in an industrial process through an offline detection and distributed control method, and establishing an industrial process database; s2, performing normalization processing on the collected sensor data based on a Z-Score method to obtain a normalized data set; according to the multi-mode process virtual sample generation method based on multi-view image feature fusion, mode discrimination is realized through a Gaussian mixture model, and a measurement learning method of mode preserving embedding is combined, so that the problem of data missing caused by insufficient initial data of a new process, sensor faults or working condition switching in an industrial scene is solved; feature distribution of different process modes can be accurately reflected, and after the feature distribution is combined with an original sample for training, a soft measurement model can capture more comprehensive industrial process dynamic characteristics.

The invention discloses a grain hyperspectral image classification method based on incremental learning, and belongs to the field of image classification, and the method comprises the following steps: S1, the construction of an experimental data set: selecting a plurality of old-class rice grains and a plurality of new-class rice grains, collecting a fixed number of samples for each class, dividing a historical training set, a historical test set, a new category training set and a new category test set, converting the hyperspectral original data into a Tensor format, and normalizing the hyperspectral original data to [0, 1]; s2, initial model training; s3, new category incremental training: extracting historical category statistical parameters from a feature memory module to construct an anchoring pool, screening Top-3 similar historical categories, embedding a diffusion model noise adding process, injecting spectrum and space priori knowledge to generate virtual samples, performing dual dynamic verification, and combining real and virtual samples to form a new category training set; and calculating the distillation weight based on the spectrum similarity, and training the model through the dual-target distillation loss.

The invention provides a stainless steel corrosion rate prediction method based on virtual sample generation and transfer learning, and relates to the technical field of data-driven prediction models, and the method comprises the steps: S1, obtaining target stainless steel material corrosion data and low alloy steel corrosion data, and carrying out the standardization processing; s2, determining the direction and range of virtual sample data generation based on an SMOTE virtual sample generation method, and generating a stainless steel material data synthesis sample; s3, constructing a cross-domain transfer learning model of a stainless steel material, training an artificial neural network model by using low alloy steel corrosion data, and transferring to a target domain model; and S4, constructing a corrosion performance prediction optimization model of the target stainless steel material, and performing optimization output to obtain a corrosion prediction result of the target stainless steel material. Cross-domain corrosion rule migration is realized through a virtual sample generation technology and migration learning, and an efficient and reliable solution is provided for stainless steel corrosion rate evaluation by increasing the basic data volume.

The invention provides a dioxin emission prediction method based on an RF-PSO integrated algorithm, and relates to the technical field of dioxin emission prediction. Comprising the following steps: S1, constructing a data set; s2, preprocessing a data set, performing standardization and dimension reduction processing on input features in the data set, and dividing the data set into a training set and a test set; s3, training and testing a random forest model, and taking the trained random forest model as a dioxin emission initial prediction model; s4, generating virtual sample input; expanding the range of the input features through multi-distribution overall trend diffusion, and constructing a virtual input data set; and S5, optimizing the generated virtual sample through an RF-PSO (Radio Frequency-Particle Swarm Optimization) integration algorithm. According to the method, even if the performance of the initial prediction model is weak, the reliable virtual sample can still be generated, and the high generalization ability and prediction precision of the final dioxin emission prediction model are ensured.

This invention belongs to the field of soft measurement modeling technology and discloses a soft measurement modeling method based on mode-related spatiotemporal diffusion generation, which includes the following steps: (1) acquiring data of multi-mode dynamic processes; (2) data partitioning and preprocessing operations; (3) establishing a mode-related spatiotemporal diffusion model and generating virtual samples; (4) predicting pressure variables of three-phase flow processes and evaluating model performance. This invention proposes a soft measurement modeling method based on mode-related spatiotemporal diffusion generation. By using the DSTN-Net noise prediction network to capture the dependence of dynamic data in the time and space dimensions, the mode learner learns the multi-mode distribution characteristics and distinguishes the data distribution of different modes, which can generate virtual samples with high similarity to the original samples, thereby improving the prediction performance of the soft measurement model in the case of small samples.

The invention discloses a numerical control system-oriented defect visual classification detection method and system, and the method comprises the steps: obtaining an original defect sample covering a numerical control system part, carrying out the multi-modal image collection, carrying out the weighted fusion of visible light, near-infrared light and ultraviolet light images in an original sample library through a multispectral fusion algorithm based on the original defect sample, and carrying out the detection of the defect visual classification. The method comprises the following steps: generating an original defect sample, generating an enhanced sample, acquiring three-dimensional geometric information of the surface of a part by utilizing a 3D laser scanning device, transmitting and receiving ultrasonic waves through an ultrasonic detection module, detecting defects in the part, generating a virtual sample based on the original defect sample, the enhanced sample and a 3D model sample, and combining the original defect sample, the enhanced sample, the 3D model sample and the virtual sample to obtain a three-dimensional image. And training a defect classification model based on the mixed sample library. Defect evolution under different working conditions can be simulated, virtual defect samples conforming to real physical laws are generated, the defect state coverage range is wide, and defect data which are difficult to obtain by real samples are supplemented.

The invention belongs to the technical field of fault diagnosis, and particularly relates to a gearbox fault twinborn diagnosis method based on a virtual-real coding mapping optimization model, and the method comprises the steps: obtaining envelope spectrums of actual measurement and simulation signals of a gearbox under two different working conditions, and constructing a source domain actual measurement and virtual sample set and a target domain test and virtual sample set; building a virtual-real coding mapping network, and obtaining a virtual-real coding mapping optimization model by adopting a Bayesian optimization algorithm; training the optimization model by utilizing the source domain actual measurement and the virtual sample set until convergence; inputting target domain virtual and actual measurement samples into the trained optimization model, extracting corresponding coding features of the target domain virtual and actual measurement samples, training a k-nearest neighbor (KNN) classifier by adopting the coding features of the target domain virtual samples, inputting test sample features into the trained KNN, and outputting a diagnosis result; according to the method, high accuracy can be achieved in small-sample cross-working-condition gearbox fault diagnosis, and the method has good application prospects in the field of twin data driven gearbox fault diagnosis.

The invention discloses a product quality defect analysis method and device and a medium, and relates to the technical field of data processing. The method comprises the following steps: firstly, establishing a prediction model from a raw material performance index to a product quality index based on historical data by using a random forest algorithm; when product defects are detected, initial raw material data of the batch of products are obtained and subjected to multi-round random disturbance, a large number of virtual samples are generated, and the quality of the virtual samples is predicted through a model. Then, qualified samples are screened out from the virtual samples, and a sample subset with the highest model prediction reliability is further selected from the samples. And finally, performing statistics on the consistency of the disturbance direction of each raw material performance index in the subset, and judging the raw material index with the highest consistency as a key factor which is most likely to cause the current quality defect. Noise in industrial data can be effectively processed, product quality problems can be quickly and accurately attributed to specific raw material characteristics, and accurate guidance is provided for production control and process optimization.

This invention discloses a small-sample Bayesian optimization sampling method and system based on cloud droplet data augmentation, belonging to the field of transportation engineering material design. Addressing the problems of limited initial samples and poor fit of the Bayesian optimization surrogate model in modified asphalt formulation design, this invention acquires initial small-sample data; constructs a clustering cloud model to obtain expectation, entropy estimates, and hyperentropy estimates; generates cloud droplet virtual samples using a forward cloud generator and merges them with the original samples to expand the dataset; trains a Gaussian process regression surrogate model based on the expanded data; constructs an expectation-improved acquisition function to optimize and solve candidate formulations; updates the data after physical testing verification; and iterates repeatedly until the termination condition is met to output the optimal formulation. This invention mines the distribution information of small-sample data through cloud droplet data augmentation, improves the accuracy and sampling efficiency of the surrogate model, significantly reduces the number of expensive physical tests, and lowers the cost of formulation design. It can be widely applied to the formulation optimization of modified asphalt and similar high-cost experimental materials.

The invention discloses an industrial product visual quality detection and parameter optimization method based on digital twinning, and relates to the technical field of digital twinning and machine vision. According to the method, a virtual sample library is generated by constructing a digital twin model, and a detection model is trained in combination with multi-feature fusion and transfer learning; deploying to an edge end for real-time detection, positioning a defect root cause through a digital twin re-occurrence process, optimizing parameters by adopting a multi-objective optimization algorithm, and performing closed-loop iteration. The method does not need to depend on a large-scale real defect sample, improves the detection precision and generalization capability, and achieves the continuous optimization of the assembly quality and efficiency.

The invention discloses a sample screening method and device, equipment, a medium and a product, and the method comprises the steps: obtaining an initial task model, the initial task model is obtained through the training of a first sample set, and the first sample set comprises a plurality of real samples; adding a plurality of first virtual samples into the first sample set to obtain a second sample set; in a current training period of training the initial task model, training the initial task model by adopting the second sample set to obtain a reasoning result of each first virtual sample; and determining a plurality of target virtual samples according to the first virtual samples of which the confidence coefficients of the reasoning results are smaller than a confidence coefficient threshold value in the plurality of first virtual samples. And the virtual sample for improving the model performance can be screened out, so that the model training effect is improved.

The application relates to the field of data processing and provides an engineering problem solving method and system based on a fusion model, which comprises the following steps: performing data enhancement on real samples through physical constraint noise disturbance to obtain an enhanced sample set; performing parameter optimization on a physical constitutive model according to the enhanced sample set to obtain optimal constitutive model parameters; performing interpolation on the enhanced sample set through the physical constitutive model optimized by the optimal constitutive model parameters to obtain a virtual sample set; performing multi-source collaborative training on a neural network model according to the real samples, the enhanced sample set and the virtual sample set to obtain a hybrid training model; and performing prediction and solving on an engineering problem to be solved through the hybrid training model to obtain a solving result. The application realizes collaborative optimization of physical driving and data driving, improves the adaptability of the model to samples outside the data distribution under the premise of ensuring the credibility of generated data.

The invention discloses a wind power plant power curve AI calibration method considering environmental factors and performance degradation, and belongs to the field of new energy power generation. The objective of the invention is to solve the problem of large prediction deviation caused by insufficient consideration of environment dynamic change and lack of a fan degradation monitoring correction mechanism in a traditional power curve. The method comprises the steps of firstly constructing an environmental stress-microscopic damage mapping library, then generating a physical credible virtual sample to fill a data blind area, then establishing an implicit degradation-power coupling equation set, and finally solving in real time through edge side fast and slow double circulation. The fast cycle calibrates the power in a second level, the slow cycle periodically updates the degradation initial value and the correction parameter, and the double-cycle cooperation ensures the calibration accuracy. The method can integrate the environment and degradation factors, improves the power prediction precision, guarantees the stability of a power grid, improves the benefits of a wind power plant, and is suitable for wind power plant power optimization control.