18 results about "Network classification" patented technology

Provided are a method for regulating a wind-photovoltaic-storage power station based on electricity, green certificate, and carbon price prediction, a device, a medium, and a product. The method includes: inputting acquired historical price data into a price prediction model, and outputting a predicted price; determining a deviation vector of price data based on the historical price data and the predicted price, and generating an uncertainty set of the predicted price by using a multi-kernel-based one-class support vector machine algorithm; classifying the uncertainty set of the predicted price by using a neural network classifier, to obtain multiple types of price scenarios; solving, based on predicted prices under the multiple types of price scenarios, a joint clearing model by using a Pied Kingfisher Optimization (PKO) algorithm, to obtain an operation strategy for the wind-photovoltaic-storage power station; and regulating the wind-photovoltaic-storage power station based on the operation strategy for the wind-photovoltaic-storage power station.

The application relates to the technical field of artificial intelligence, and provides a classification prediction method based on table data, which comprises the following steps: inputting a characteristic value, a characteristic type corresponding to the characteristic value and statistical information corresponding to the characteristic value into an embedding layer of a neural network classification model to obtain an embedding representation corresponding to the characteristic value; then determining an embedding representation sequence according to the embedding representation; inputting the embedding representation sequence into an encoding layer of the neural network classification model to obtain a semantic vector sequence; inputting the semantic vector sequence into an output layer of the neural network classification model to obtain a prediction result; determining a loss value according to the prediction result and a labeled label, and training the neural network classification model according to the loss value to obtain a target neural network classification model. Through the above scheme, the neural network model can understand the global statistical information of the characteristics like a tree model, the performance and accuracy of the neural network model in performing a classification prediction task based on table data are improved, and the satisfaction of users in the experience of financial products is improved.

This invention provides a method for identifying transient processes in traction networks based on phase space reconstruction and neural networks. The method includes: acquiring transient current signals; calculating the maximum Lyapunov exponent of the transient current signals; if the exponent is greater than zero, proceeding to the next step; for transient current signals that meet the phase space analysis conditions, adaptively determining the optimal delay time using the average mutual information method and adaptively determining the optimal embedding dimension using the pseudo-nearest neighbor method; reconstructing the phase space of the transient current signals to obtain a reconstructed phase space vector; extracting dynamic attribute features reflecting the dynamic characteristics of the transient process and constructing a joint feature vector; inputting the joint feature vector into a pre-established and trained radial basis function neural network classification model, and outputting the category identification result of the transient process through the model. This invention can solve the problems of poor identification ability, difficulty in fully revealing the deep dynamic characteristics of transient signals, and limited adaptability to complex scenarios in existing technologies.

This invention relates to the field of weld defect type identification technology. Existing weld defect detection methods fail to achieve intelligent and efficient detection due to the diverse types of weld defects. This invention provides a low-cost self-learning neural network design method for ultrasonic detection of weld defects. It performs high-dimensional spatial domain feature representation on the initial one-dimensional ultrasonic signal, enriches the feature expression of weld defect data, selects the feature domain with better performance, constructs an adaptive scaling network for weld defect detection, and uses a multi-objective, training-free network search and evaluation method to iteratively search and evaluate candidate networks for weld defect detection, obtaining a balanced solution on both network classification accuracy and parameter quantity. Ultimately, it achieves the selection of a weld defect type detection network with superior overall performance without any training.

This invention relates to the field of circuit breaker fault diagnosis technology, specifically a method based on a large language model and a fuzzy prototype network. By fusing signal features and semantic information, and considering both small-sample learning capability and classification robustness, it achieves fault diagnosis for high-voltage circuit breakers. The method first performs multi-domain sequence feature encoding, then episode-driven dynamic semantic generation, followed by bidirectional attention and fuzzy membership interaction, and finally fuzzy prototype network classification. By introducing semantic priors to constrain the feature space, this invention effectively characterizes the relationship between complex fault features and category semantics under few-sample conditions, reducing the impact of fuzzy boundary samples on classification results. This improves the accuracy and generalization ability of high-voltage circuit breaker fault diagnosis, making it suitable for intelligent diagnosis scenarios involving novel faults and complex operating conditions.

Approaches to improving frame rate and visual smoothness in real-time video streams through multi-frame interpolation are revealed. A neural classification network analyzes two consecutive images and outputs confidence scores indicating the reliability of motion data for each pixel. These scores determine whether the motion of a pixel is accurately described by the motion vectors or should be treated as static. The classification results are reused to generate intermediate frames by warping the original images based on their motion properties. Blended weights are calculated by combining the confidence scores of the warped motion vector with static values, and a second neural network refines the alignment and blending of the candidate images.This second network predicts intermediate streams and generates new blend weights that are used to warp and blend the candidate images, ultimately producing a final interpolated image that improves the visual smoothness and consistency in the video stream.

This application discloses a method and system for locating and verifying carbon emissions at the mesoscale level using a carbon dioxide lidar system. The method includes the following steps: obtaining CO2 concentration data within a preset mesoscale region using a CO2 lidar system; identifying carbon sources based on the CO2 concentration data; locating and tracking carbon sources based on the identification results; and verifying the carbon emissions of the carbon sources based on both the location and identification results. This application utilizes machine learning to train a neural network classification model to identify and classify carbon sources, track and locate them, accurately calculate the carbon emissions of each source, and verify their emissions. This enhances the management and supervision of carbon-emitting enterprises and improves the atmospheric environment.

This invention discloses a meta-learning-based social bot detection method. First, it quantifies and clusters the behavioral features of social platform accounts, reconstructing coarse-grained binary labels into fine-grained multi-class label sets. The meta-learning dataset is then divided, and a standardized N-way K-shot few-shot meta-task is generated. Next, multi-modal features of the accounts are extracted, mapped by an encoder, and a dual-branch encoding network combining inductive MLP and transductive R-GCN is constructed. This network adaptively fuses the output to produce a high-dimensional joint discriminative embedding. Finally, the embedded features are input into a prototype network classifier to calculate fine-grained category prototypes, measure sample similarity, and generate classification probabilities. Model optimization is achieved based on episodic cross-entropy loss. This invention addresses the lack of meta-task diversity in traditional meta-learning detection methods, enabling the model to learn the essential representation of account behavior, significantly improving its generalization ability to unknown social bot variants, and adapting to few-shot cold-start detection scenarios.

This invention relates to a method and system for extracting maize planting area based on deep learning, comprising: determining multiple vegetation indices based on acquired optical images and determining polarization feature indices based on radar images; constructing a set of statistical feature images corresponding to the entire growing season based on the multiple vegetation indices and polarization feature indices; inputting the median EVI and median RVI temporal images in the set into a temporal convolutional network (TCN) to extract target phenological features; aligning, stitching, and standardizing the target phenological features with the mean NDVI image, maximum NDRE image, and standard deviation RVI image in the set according to pixels to obtain a target multidimensional classification feature matrix; inputting the target multidimensional classification feature matrix into a prototype network classifier for preliminary identification of maize planting areas; and performing connected component analysis, inverse hole filling, and quantified statistical area based on the preliminary identification results to obtain the total maize planting area of ​​the target region.

This invention discloses an auxiliary diagnostic method and system for autism based on a dual-person brain functional connectivity map and graph attention network. It comprises a dual-person fNIRS hyperscanning data acquisition module, a preprocessing and graph construction module, and an edge-weighted enhanced graph attention network classification module. The system has the following technical features: synchronously acquiring brain activity signals of the subject and examinee during social interactions using hyperscanning technology; constructing a dual-person brain functional connectivity map including nodes in both brain regions and the strength of inter-brain connections; and dynamically aggregating and learning node features and edge weights in the graph using an edge-weighted enhanced graph attention network. This invention can objectively capture abnormal inter-brain synchronization in autistic individuals during real social interactions, providing automated, rapid, and quantifiable auxiliary diagnostic results, effectively improving the objectivity and accuracy of diagnosis, and reducing reliance on professional clinical experience.

The application is based on a brain function network classification method and system based on adversarial graph contrastive learning, the method comprising: acquiring resting-state functional magnetic resonance image data, preprocessing and constructing a functional connection matrix X; inputting X into an adversarial graph contrastive learning classification model for classification. The model comprises a graph augmenter, a feature extraction layer, a projection head and a classifier. The graph augmenter has a trainable encoder built-in, generates edge deletion probability matrix P and augmented graph X' through the encoder; the feature extraction layer extracts the feature representation of X and X'; the projection head maps the feature representation to the contrastive learning space to obtain the optimal weight parameter; and the classifier classifies the brain function based on the feature representation. The application has the advantages of realizing data-driven and task-oriented dynamic augmentation, retaining classification-related functional connections, deleting redundant connections, enabling the model to distinguish the functional specificity of different brain regions, avoiding the problem that the traditional model cannot distinguish the functional differences of different brain regions due to node permutation invariance, resulting in poor classification performance and loss of practical significance of the explanatory nature.

Disclosed embodiments includes a network classification system (NCS) that generates a set of machine learning (ML) features from information about information objects accessed by various users, and determines an organization (org) type associated with the network address based on the set of ML features. Obtained network events may include the information about the accessed information objects. A content consumption monitor (CCM) generates consumption scores for the network addresses based on the identified org types. The CCM can generate more accurate intent and consumption data by filtering out events unrelated to content consumption for that org type. The NCS and the CCM may be implemented as the same network function, or the NCS and CCM may be implemented as separate network functions. Other embodiments may be described and / or claimed.

This invention proposes a gear fault diagnosis method based on an enhanced relational network. The method involves: 1) acquiring multiple sample gear vibration signals and dividing them into a support set and a query set; 2) converting the gear vibration signals in the support and query sets into two-dimensional video images using continuous wavelet transform; 3) enhancing the relational network's ability to obtain strong features by incorporating a CA attention mechanism; 4) constructing and training a relational network model for gear fault diagnosis using the acquired data; 5) monitoring and collecting gearbox vibration signals in real time and preprocessing the vibration signals; and 6) inputting the preprocessed vibration signals into the relational network classification model, and obtaining the gearbox's operating status through analysis and comparison, thereby completing the gearbox fault diagnosis. This invention fully utilizes the CA attention mechanism to enhance the relational network's ability to extract features, reducing the problem of low diagnostic accuracy due to a small sample size.

The present application relates to the technical field of data processing, and particularly relates to a packaging machine fault diagnosis method and system, the method comprising: obtaining operation data of multiple measuring points in a packaging machine; constructing a device heterogeneous graph, generating a causally enhanced directed acyclic graph combined with the causal relationship strength between nodes, generating a spatio-temporal fusion feature of a node by using a cross-node time series attention mechanism combined with a probability feature vector representing the potential operation state of the node; performing a graph readout operation on the spatio-temporal fusion features of all nodes to obtain a graph-level feature vector representing the overall operation state of the packaging machine; inputting the graph-level feature vector into a fully connected network classifier to output the fault type and confidence of the packaging machine. The present application can comprehensively and dynamically represent the overall health state of the device.

This invention discloses an eddy current flaw detection method for rectangular cross-section pipes. First, the pipe is straightened multiple times using a precision straightening machine to strictly control longitudinal curvature, flatness, and corner radius deviations. Then, a flexible array probe containing planar and corner modules is driven by an inflatable airbag encased in silicone, with a pressure closed-loop control to maintain a fill coefficient ≥85% and to match differentiated eddy current parameters. Visual positioning and vacuum adsorption conveying ensure coaxiality. Defects are classified using wavelet-based denoising and a CNN neural network, and an encoder achieves ±1mm positioning. Finally, laser marking and pneumatic sorting are performed. This invention improves flaw detection accuracy and efficiency and is applicable to the inspection of pipes of various specifications.

This invention discloses an industrial defect detection method based on incremental training and CBAM enhancement, belonging to the field of industrial defect detection technology. The method includes the following steps: acquiring surface images of industrial products containing existing and newly added defects; performing annotation, data augmentation, and dataset partitioning to obtain existing defect datasets and newly added defect datasets; constructing an incremental lightweight network integrating a CBAM module and initializing the network weights; training the network based on the existing defect dataset, optimizing the network weights, and saving the initial model and Fisher information matrix; when adding a new defect type, expanding the network classification layer dimension, loading the initial model weights, and incrementally training and updating the network using the newly added defect dataset in conjunction with EWC regularization; preprocessing the image to be detected, inputting it into the trained network, and outputting the defect category and confidence level.

This invention discloses a method for UAV radio frequency fingerprinting based on a lightweight network, comprising: collecting UAV spectrum data for complex scenarios; segmenting the collected UAV radio frequency signals into K segments of I / Q data; extracting radio frequency signal features based on the collected UAV I / Q data; replacing the classifier in the lightweight network MobileNetV4-Small model with a KAN network classifier based on B-spline functions, and setting the hyperparameters of the trainable network to construct a UAV radio frequency fingerprint model based on a lightweight network; training the network using training set data and fine-tuning the network model parameters using validation set data; inputting test set data into the trained UAV recognition network model and outputting the UAV category prediction result. This invention, through the KAN network classifier, leverages its powerful feature representation learning ability and the interpretability of the network structure to not only improve the accuracy of UAV prediction but also enhance the robustness of the model.

The application discloses a painting style recognition method and system based on a lightweight deep learning model, and belongs to the field of artificial intelligence and artistic image recognition. The application designs an IRA-HCT-KAN end-to-end classification network, integrates six IRA (inverse residual attention) modules, three HCT (convolution and Transformer hybrid) modules and a KAN (Kolmogorov-Arnold network) classification layer, extracts, converts and classifies painting style categories through features, and realizes efficient local feature extraction and down-sampling through the IRA module. The HCT module fuses the advantages of convolution local perception and Transformer global modeling, and the KAN classification layer replaces the fully connected layer to improve the interpretability. The application considers the recognition performance, lightweight and decision transparency, can be deployed on mobile or edge devices, is suitable for the fields of art history research and digital cultural relics, and is helpful to intelligent understanding of cultural and artistic images.