145results about How to "Strong generalization" patented technology

The invention discloses a power electronic transformer working mode recognition method and device, equipment, a medium and a product, and relates to the field of artificial intelligence, and the method comprises the steps: collecting original inductive current data during the operation of a power electronic transformer; performing adaptive segmentation normalization processing on the original inductive current data to generate a normalized inductive current sequence; calculating a wavelet packet energy entropy and a time domain differential entropy of the normalized inductive current sequence, and splicing the wavelet packet energy entropy and the time domain differential entropy into a two-dimensional fusion feature vector; inputting the normalized inductive current sequence and the two-dimensional fusion feature vector into a trained deep learning model to obtain a prediction probability vector; based on the prediction probability vector, the working mode category with the maximum probability value is selected as the recognition result, and the recognition precision of the working modes of the power electronic transformer can be guaranteed under the working conditions of high noise interference or rapid mode switching.

The invention discloses a multi-layer semantic perception, distillation and semi-supervised cooperative training target detection method, and the method specifically comprises the following steps: constructing a data set: extracting a first part of images from image data, marking the first part of images to construct a supervised target detection data set, and taking the remaining images as an unmarked image data set, the data volume of the unlabeled image data set is greater than that of the supervised target detection data set; teacher model optimization: performing supervision training on the teacher model on the supervised target detection data set; constructing a teacher model, and executing self-distillation training on the teacher model; pseudo labels are generated for the unlabeled images through a teacher model, and adaptive screening is carried out based on confidence distribution; mapping a pseudo label to a strong enhanced sample through enhanced geometric transformation, carrying out semi-supervised training by using the strong enhanced sample and the pseudo label, and introducing a feature layer distillation constraint at the same time; optimizing a student model; and outputting the target detection model obtained through training.

The application discloses a kind of multi-modal hierarchical fusion network-based motor function multidimensional intelligent evaluation method and system. First, multi-modal data input is carried out, and the original data from the sensor is received;Then data preprocessing and feature extraction are performed, signal filtering, segmentation, alignment and feature extraction are completed;Efficient fusion of multi-modal data is realized by using MMHF-Net core model;Finally, multi-scale evaluation result output is realized, which is used to generate macro, meso and micro three-level evaluation results. The present application deeply fuses kinematics, electromyography, electroencephalogram and visual information, accurately restores the motor control process, and the evaluation accuracy is significantly better than that of single modal method. The proposed multi-modal hierarchical fusion network MMHF-Net can model multi-modal spatio-temporal interaction and dynamically focus on key information, while outputting scale scores, sub-project indicators and recognition results, providing fine-grained clinical insights, and achieving excellent generalization performance relying on multi-task learning and data augmentation.

The invention discloses a line loss prediction method and system based on integrated DBN-BP, and belongs to the technical field of power system data analysis. The method comprises the steps that firstly, a plurality of parallel DBN sub-networks are constructed, all the sub-networks adopt different activation functions, unsupervised pre-training is carried out with N antenna loss historical data and corresponding weather data as input, and high-dimensional robust features are automatically extracted; and then, taking the output of each sub-network as a feature, inputting the feature into a BP integrated network for supervised training, and finally fusing to obtain a high-precision line loss prediction value. Through the architecture of "unsupervised feature extraction + supervised integrated decision", the problems of strong dependency on annotated data and weak feature extraction ability in the prior art are effectively overcome; meanwhile, forward prediction can be simplified into efficient matrix operation through the full-connection structure of the model, the requirement for real-time dispatching of the power grid is met, excellent generalization ability is achieved, and reliable data support is provided for economical and safe operation of the power grid.

The application relates to a kind of OFDM-based broadband millimeter wave relay system hybrid beamforming design method, comprising: calculating the receiving signal of target node under the condition that each node is all digital processor;The receiving signal is processed using minimum mean square error criterion in target node, and MMSE matrix is calculated;Using the equivalence between maximizing sum rate and minimum weighted mean square error algorithm, the maximum sum rate problem is converted into minimum weighted mean square error problem according to MMSE matrix;Deep unfolding neural network is used to solve the minimum weighted mean square error problem, and all digital processor of each node is obtained;Each node's all digital processor is decomposed based on least square decomposition algorithm, and hybrid beamforming matrix of each node is calculated, the complexity of the algorithm is greatly reduced, and the running time of the system can be effectively improved.

The invention discloses an automatic hepatic fibrosis staging method and system based on deep learning, and the method comprises the following steps: S1, obtaining T1WI and T2WI original image data of a patient, carrying out the data preprocessing and liver segmentation of the original image data, and obtaining a T1 weighted image data set and a T2 weighted image data set; s2, a DMF-Vheat hepatic fibrosis staging model is constructed, the staging model adopts a double-sequence classification architecture and comprises a T1 branch network, a T2 branch network and a fusion network, and a heat conduction operator layer and a mixed attention module are adopted to extract and process complementary information of T1WI and T2WI sequences; s3, putting the T1 weighted image data set and the T2 weighted image data set into the staging model for training, and optimizing the staging model; and S4, outputting a hepatic fibrosis staging result through the staging model. According to the method, complementary information of T1WI and T2WI is fully utilized through the multi-mode deep learning network, and the accuracy and practicability of hepatic fibrosis staging are remarkably improved.

This invention discloses a method, system, device, and storage medium for predicting oil pipe corrosion rate based on a PCA-PSO-SVR hybrid model, specifically including the following steps: S1 Collecting corrosion detection data and operating condition parameters of the oil pipe to form a dataset; S2 Preprocessing the dataset; S3 Using the PCA model to perform dimensionality reduction on the dataset and extracting the main features affecting the corrosion rate; S4 Initializing the parameters of the PSO model; S5 Optimizing the parameters of the SVR model using the PSO model; S6 Constructing a corrosion rate prediction model based on the optimized SVR model; S7 Inputting the preprocessed dataset into the corrosion rate prediction model to obtain the prediction result and complete the prediction.

The invention provides a protein chromatography process optimization method and system based on reinforcement learning, and relates to the technical field of reinforcement learning, and the method comprises the steps: obtaining multi-source heterogeneous data, carrying out the multi-scale decomposition, building a hierarchical mechanism knowledge base, combining real-time operation data analysis, building a hybrid drive prediction model, and achieving the state prediction of a chromatography process. And dynamically adjusting model parameters by using layered feedback information. The intelligent control of the protein chromatography process can be realized, the separation purity and yield are improved, the process parameter adjustment time is shortened, and the production cost is reduced.

This invention provides a vehicle simulation speed correction method and system, relating to the field of vehicle intelligent dynamics modeling technology. The correction method includes the following steps: S1: Input and process simulation data output from the vehicle dynamics simulation model and corresponding real vehicle test data; S2: Construct a dynamic graph structure representing the interaction between state variables based on a preset physical coupling relationship of the vehicle powertrain; S3: Input the simulation data into a GCN and perform graph convolution operations under the constraints of the dynamic graph structure to extract graph embedding feature sequences representing the spatial dependencies between state variables; S4: Input the graph embedding feature sequences into a TCN and perform temporal convolution operations to learn the dynamic evolution law of state variables in the time dimension and output the correction amount of the vehicle simulation speed; S5: Output the result. Based on this, this invention solves the problem that existing correction methods have various limitations in practical applications.

The application belongs to the technical field of vehicle reliability evaluation, and discloses a vehicle component reliability evaluation method suitable for strong random working conditions. In the method, multiple random working condition parameters such as road surface grade and driving speed are uniformly represented as continuous variables, a single high-precision proxy model covering all working conditions is constructed, and advanced fast sampling algorithms (Latin hypercube sampling, mixed importance sampling, adaptive Markov chain Monte Carlo) and a segmented random task profile are combined to realize accurate and efficient evaluation of cumulative damage.

The invention belongs to the technical field of new energy power generation and grid connection, discloses a grid-side converter fault ride-through control structure modeling method, system and equipment and a medium, and aims to solve the problems of difficult modeling, high cost, poor extrapolation and lack of universality caused by dependence on manufacturer internal logic or massive experimental data in the prior art. According to the method, disturbance input and current output data under a limited working condition are acquired, a dimension raising mapping function is constructed to map the input to a high-dimensional feature space, a Koopman operator is solved based on ridge regression to establish a global linear mapping model, active and reactive current reference components are predicted, and the current reference components are optimized. And finally, inputting the current inner loop time domain model to generate a current transient response curve during the fault period. According to the method, high-precision modeling and prediction can be realized only by a small number of samples, and the method has good extrapolation and universality and can be adapted to various fault ride-through strategies.

The application discloses a kind of robot dexterous operation systems, methods, equipment and media for transparent object, belong to robot dexterous operation and three-dimensional perception field, system includes: point cloud data construction module, can be constructed and noisy to obtain noisy hand-object interaction point cloud by physical simulation;Point cloud feature extraction module can extract high-dimensional latent features from noisy hand-object interaction point cloud;Query prediction module can infer object shape and pose by query point generation and decoder decoding;Perception coding module can extract features from global point cloud, hand-object interaction point cloud and tactile information to obtain multi-source perception features;Multi-modal feature fusion module can realize feature integration through self-attention mechanism and cross-attention mechanism calculation, and output fusion features;Action generation strategy module can generate differentiated action instructions for robot arm and dexterous hand based on fusion features.The system can achieve higher adaptability and stronger generalization capability in transparent object operation task.

The application relates to the technical field of image processing, in particular to an image super-resolution reconstruction method based on a multi-scale content perception mixer, which is realized by using an adaptive processing mechanism. The method comprises the following steps: shallow feature extraction is performed on a low-resolution image to be reconstructed, so as to obtain an initial shallow feature map; feature enhancement based on a feature pyramid and an attention mechanism is performed on the shallow feature map, so as to obtain a deep feature map; multi-scale content perception prediction is performed based on the deep feature map, so as to generate guide information for guiding calculation allocation, the guide information comprising a window classification binary mask and a window size; different image regions are allocated to different calculation paths for processing based on the guide information; the feature maps output by the calculation paths are recombined and fused, and then enlarged to a target resolution, so that a high-resolution image is finally obtained. The method realizes accurate classification of image regions and on-demand allocation of calculation resources, and significantly reduces the calculation complexity and the memory occupation.

The present application relates to a kind of forest leaf instance segmentation method and system based on deep learning, method includes: obtaining vegetation image, vegetation image is input into leaf instance segmentation model, obtains leaf instance segmentation prediction result;Leaf instance segmentation model is trained using training set;Training set includes: vegetation original image;Feature extraction and enhancement are carried out using backbone module in leaf instance segmentation model, and adaptive spatial fusion mechanism in progressive feature pyramid network is integrated to dynamically adjust feature weight, generate dynamic fusion feature;Through the dynamic asymmetric spatial perception mechanism built-in in dynamic anomaly regression head module, the corresponding multi-source deformation feature layer of dynamic fusion feature is obtained, and the feature fusion strategy of top-down cascaded decoding module is used to optimize multi-scale feature, obtain multi-source fusion feature layer, further using multi-source fusion feature layer, generate leaf instance segmentation prediction result.The present application solves the problems of data scarcity, poor adaptability and low efficiency.

The invention discloses a battery state and early fault diagnosis system based on multi-source data fusion and artificial intelligence. The battery state and early fault diagnosis system comprises a data acquisition module, a feature extraction module, an artificial intelligence processing module and an early warning and execution module. The data acquisition module is responsible for synchronously acquiring voltage, current and temperature data from a battery pack in real time to form a multi-source data set for describing the real-time running state of the battery; the feature extraction module receives an original data stream from the data acquisition module. According to the method, the limitation of a traditional method based on a simple model or threshold judgment is broken through, the strong nonlinear mapping capability of the artificial intelligence model is utilized, and the deep correlation with the internal health state and the early fault of the battery is accurately mined from the high-dimensional features; according to the method, the estimation precision of the health state of the battery is remarkably improved, early weak fault symptoms which cannot be perceived by a traditional method can be identified, the crossing from post-event alarm to pre-event early warning is realized, and the safety of the system is greatly improved.

The application discloses a kind of for the method, device, equipment and medium of tab abnormality detection in battery cell production process, comprising: the tab area image collected in battery cell production process is segmented to generate tab mask;Based on the tab mask extraction multi-scale pixel-level feature vector, and the multi-scale pixel-level feature vector is input into unsupervised probability GMM model to identify pixel-level abnormal point;The pixel-level abnormal point is analyzed to obtain candidate abnormal area, and the parameter index of the candidate abnormal area is obtained;Determine whether the parameter index meets the set tab abnormality judgment rule, if meet, it is judged as tab abnormality, and tab abnormality result is output.Therefore, the application can be widely applied to the detection of tab folding, crease, fracture, misplacement and other abnormalities on battery cell production line.

This invention relates to the field of intelligent brain disease diagnosis technology, specifically providing an auxiliary diagnostic method and system based on multimodal decoupled dynamic graph learning. The method includes: acquiring and preprocessing multimodal data (such as neuroimaging, genetic markers, etc.) of the subject; extracting common pathological information and modality-specific features through a shared encoder and modality-specific encoders respectively, and optimizing the separation process using a decoupling loss function; furthermore, fusing all modality embeddings using a multi-head self-attention mechanism with a masked matrix to generate initial node representations, where the mask is used to suppress modality self-attention; subsequently, performing hierarchical dynamic graph convolution based on the node representations: in each layer, dynamically updating the graph adjacency matrix by combining the current node representation with the original features, and iteratively optimizing the node representations through message passing; finally, inputting the optimized representations into a classifier to obtain disease prediction results. This invention improves the automation performance and reliability of diagnosis.

The invention discloses a rock slice image classification method based on domain self-adaption, and the method comprises the steps: collecting rock slice image data, the method comprises the following steps: acquiring a rock slice image, performing multi-scale data preprocessing and geological field data enhancement, extracting multi-level visual features of the rock slice image by utilizing a pre-trained DINOv3 model, performing field specialized adaptation on general visual features through a geological field adaptive Adapter module, and enhancing rock slice discriminative feature representation by adopting a double-path attention mechanism. Constructing a progressive hierarchical classification head to realize coarse-to-fine rock classification; and designing a multi-stage progressive training strategy to optimize the overall performance of the model. According to the method, mineral composition and structural features of the rock slices under different scales can be accurately captured, and multi-scale features and an attention mechanism are fully utilized, so that accurate classification of the rock slices is realized, and the accuracy and reliability of rock slice identification are remarkably improved.

This invention provides a method for identifying the authenticity of wheat flour based on the fusion of Raman and near-infrared spectroscopy, belonging to the field of wheat flour authenticity identification technology. The method includes: acquiring Raman and near-infrared spectral data of the wheat flour sample to be tested, and preprocessing them separately; constructing and training a fusion detection model, which includes a data-level fusion module, a feature-level fusion module, and a decision-level fusion module; the data-level fusion module generates full-spectrum fusion data; the feature-level fusion module concatenates core features into a comprehensive feature set; the decision-level fusion module inputs the comprehensive feature set into a hybrid model; and the trained fusion detection model processes the comprehensive feature set to output the authenticity identification result of the wheat flour sample to be tested. This invention, through a three-level spectral fusion architecture and a dynamic adaptive adversarial mechanism, can effectively achieve high-precision and high-robust identification of wheat flour authenticity.

A tool for compressor impeller and volute assembly relates to hoisting tool technical field. In order to solve the existing tool can only hoist one level impeller, there is weak universality, cannot satisfy the assembly demand of multilevel impeller. The utility model discloses a crossbeam, vertical beam, connecting piece and connecting disc, crossbeam horizontal setting, the top of crossbeam is connected with the lifting lug, and the lifting lug has the lifting hole, and the vertical beam is vertically installed at the bottom of crossbeam, and the vertical beam and crossbeam are welded connection, and the lifting lug and crossbeam are welded connection, and the connecting piece and connecting disc are detachably connected, the connecting piece is horizontally installed on the vertical beam and is located at the same side with crossbeam, the center hole of connecting disc can carry out the stop opening positioning to impeller, after positioning, first, connecting disc and impeller are fixed, then connecting disc is fixed on connecting piece, and hoisting hole can hoist impeller, and the assembly of multilevel impeller can be realized by replacing different connecting disc, and the generalization is strong.

This invention relates to aircraft propulsion systems, specifically to an online real-time monitoring system and method for aircraft solenoid valves and self-locking valves. The system provided by this invention comprises a host computer, valve testing equipment, solenoid valves and / or self-locking valves, a solenoid valve and / or self-locking valve actuator, GNC monitoring equipment, cables, and network cables. The solenoid valve and / or self-locking valve actuator, GNC monitoring equipment, and solenoid valves and / or self-locking valves are connected to the valve testing equipment via cables; the valve testing equipment is connected to the host computer via a network cable. This connection method integrates the valve testing equipment with the solenoid valve and / or self-locking valve control pathways, enabling long-term online real-time monitoring of the solenoid valve and / or self-locking valve's operational characteristics data, eliminating the need for offline, post-processing interpretation. Furthermore, this invention offers advantages such as multiple parallel testing pathways, high time accuracy, long operating time, strong versatility, high real-time performance, strong compatibility, and low latency.

The application belongs to the technical field of deep learning and remote sensing image processing, and specifically discloses a city land use mapping method based on multi-modal data collaborative perception, which comprises the following steps: performing parcel division on high-resolution remote sensing images of a research area to obtain irregular city parcels and remote sensing spectral features thereof; setting street sampling points along a road network, extracting street perception features and extracting interest point semantic features; constructing a heterogeneous graph structure based on the street sampling points, the street perception features and the interest point semantic features falling into the same parcel, and then extracting parcel-level multi-source geographic features; inputting the remote sensing spectral features and the parcel-level multi-source geographic features into a full sparse topic model for semantic alignment and fusion to generate fused parcel feature representations; and performing classification based on the fused parcel feature representations to output city land use mapping results. The application can realize high-precision and high-robustness city land use recognition under the condition that multi-source data is unevenly distributed or sparse.

The invention discloses a multi-modal federal cross-domain fault diagnosis method and application based on prototype comparative learning, and the method comprises the steps: each source client receives an initialized global model and a prototype, and carries out the local model training through the local data; in the training process, a source client side mines potential essential association in multi-source heterogeneous data through double-layer prototype comparison, the barrier of modal isomerism is effectively broken through, and alignment of multi-modal fault features is achieved. Then, local model optimization is carried out by minimizing total comparison loss and classification loss, and a local prototype and a local modal prototype are generated; and the central server aggregates the local prototype and the local modal prototype to generate a global prototype and a global modal prototype of the comprehensive fault information, and aggregates the local model to construct a global model with good generalization ability. According to the method, modal isomerism and data distribution difference existing in multi-modal federated learning are solved, and the accuracy and generalization of global model cross-domain fault diagnosis are effectively improved.

The invention discloses a deep learning-based intelligent recognition method for a motion mode of an unmanned surface vehicle. The method comprises the following steps: acquiring original sensing data in real time through a multi-source sensor system mounted on the unmanned surface vehicle, wherein the multi-source sensor system at least comprises an inertial measurement unit (IMU), a global navigation satellite system (GNSS) receiver, a laser radar and a visible light camera; performing time synchronization and coordinate system alignment processing on the original sensing data to generate a fused sensing data stream with a unified timestamp and a spatial reference system; dividing the fused sensing data stream into time window segments with fixed lengths, wherein each time window segment corresponds to one to-be-identified motion state sample; the IMU, the GNSS, the laser radar and the visible light / infrared camera are synchronously integrated, quadruple sensing dimensions are formed, the IMU captures instantaneous acceleration and angular velocity changes and reflects control instruction execution, the reliability of single-point observation is improved, and the system can still maintain the basic recognition capability when part of sensors fail.

The invention relates to the technical field of agricultural remote sensing monitoring, in particular to a method and system for measuring the number (area) of plants in a tobacco field based on visible light images of an unmanned aerial vehicle, and is suitable for automatic and high-precision growth monitoring and resource accounting of a large-scale tobacco field. Comprising six steps of tobacco field visible light image acquisition, image preprocessing, tobacco field region segmentation, tobacco field plant target detection and plant number statistics, tobacco field area calculation, and result post-processing and output, and through combination of an improved target detection algorithm and an image segmentation technology, high precision of tobacco field plant number statistics and high accuracy of area determination are realized. Meanwhile, the detection efficiency is guaranteed, the actual requirement of large-scale tobacco field monitoring is met, data acquisition, processing, analysis and output can be automatically completed in the whole process, manual intervention is not needed, the monitoring efficiency is greatly improved, and the labor cost is reduced.

The invention discloses a method for detecting small defects on the surface of lightweight steel, and belongs to the technical field of industrial machine vision. The method is based on an improved YOLOv8n architecture and cooperatively works through three core technical means: firstly, a Swin Transform module is introduced into a backbone network to model long-distance spatial dependence and suppress complex background interference; secondly, a high-resolution P2 detection branch is constructed, shallow details and up-sampling semantic features are fused, and microdefect characterization is enhanced through bidirectional refining circulation; finally, P2 exclusive adaptive threshold focus loss ATFL is adopted, threshold update is only limited to P2 detection branch samples, and precise optimization of difficult and tiny defects is achieved in cooperation with a gradient directional return mechanism. According to the scheme, the detection recall rate and the positioning precision of the tiny defects are effectively improved while the model parameter quantity is remarkably reduced, and high-precision and light-weight industrial deployment is realized.

The application provides a lightweight steel surface defect detection method and device and processing equipment, which are based on a YOLOv11 model, deep adaptability optimization is performed, the steel surface defect detection task has better detection precision, the parameter quantity and the calculation quantity can be effectively reduced, the generalization ability is good, when the industrial terminal is deployed, the high performance and the lightweight can be considered, and thus the industrial application prospect is good.

The invention discloses a machine learning privacy auditing method and system in a trusted execution environment, and belongs to the technical field of machine learning. The method comprises the steps of uploading a private data set to a remote machine learning service deployed in a trusted execution environment to train a target model, and obtaining measurement information of private data in each training round of the target model; synthesizing the measurement information in each training round to generate a measurement sequence of a plurality of training rounds; extracting dynamic characteristics of the measurement sequence; performing classification based on the dynamic features to obtain member relationship prediction results of the private data in a plurality of training rounds; and determining whether to continue training, suspend training or exit training according to a member relationship prediction result in the plurality of training rounds. According to the method, a privacy risk measurement and auditing mechanism which is fine in granularity, low in overhead and capable of being dynamically evaluated can be realized.

The invention provides a blast furnace molten iron silicon content prediction method based on an improved grey goose algorithm, and relates to the technical field of intelligent prediction, and the method comprises the following steps: collecting multi-dimensional process parameters in a blast furnace operation process, and synchronously recording corresponding silicon content data; performing abnormal value elimination and normalization processing on original parameter data, dividing a training set and a test set, constructing an initial prediction model of the BP neural network, optimizing training parameters of the BP neural network by using an improved grey goose algorithm, and dynamically adjusting step parameters in the algorithm along with the distance between an individual and a global optimal solution. Fusing a velocity field constructed based on a local curvature factor and a local density factor, updating an individual position in a global search stage, reconstructing a final prediction model according to an optimization result, and performing convergence training by using a training set; and finally inputting the test set into the prediction model, and outputting a blast furnace molten iron silicon content prediction result. The method is suitable for blast furnace molten iron component control and has the advantages of being stable in optimization, small in error and high in real-time performance.

The application discloses an active defense method, device and equipment for diffusion speech conversion and a medium, and relates to the technical field of speech security. The active defense method comprises the following steps: obtaining source speech and reference speech, introducing a constrained protection disturbance into the reference speech, constructing protected speech, and ensuring that the protection disturbance satisfies a disturbance imperceptibility constraint. The source speech and the protected speech are input into a speech conversion model with the reference speech as a speaker condition, content information analysis, speaker condition constraint acoustic generation and waveform synthesis are completed by the speech conversion model, and a protection generated speech is output. A joint loss function is constructed based on speaker embedding representations of the protection generated speech and the reference speech, gradients of a current disturbance position and a predicted midpoint position are calculated according to the joint loss function and are fused to obtain an update direction, the protection disturbance is iteratively optimized, and projection or clipping is performed to satisfy the disturbance imperceptibility constraint. Finally, the protected speech is output.