1382 results about "Synthetic data" patented technology

The invention belongs to the technical field of earthquake disaster scene recognition, and discloses an earthquake disaster scene recognition method based on deep learning. The method comprises the following specific steps: S1, data acquisition and preprocessing; S1.1, multi-source heterogeneous data acquisition and establishment of a comprehensive database containing seismic waveform data, surface deformation data, building structure data, geographic information data and historical disaster record data; through fusion of a 3D convolutional network, a graph attention mechanism, a space-time LSTM and an adaptive cross-modal attention fusion technology, combined modeling of a seismic waveform space-time evolution law, an earth surface deformation space distribution characteristic, a building group topology vulnerability and disaster chain time sequence association is realized, the characterization capability of a complex nonlinear disaster mode is effectively improved, and the method has the advantages of high adaptability and high reliability. And disaster assessment response time is shortened to a sub-second level through mixed precision quantification and edge computing deployment, and high recognition accuracy is still kept in a scene with strong noise and data missing in combination with a multi-task classifier and a physical constraint verification mechanism.

Multi-modal models learn a joint latent space for relating data points across different modalities. To more effectively learn multi-modal models with reduced training requirements and greater benefit from limited multi-modal training data, a multi-modal model may be trained with fixed or pre-trained unimodal encoders that generate data representations in respective latent spaces. The multi-modal model is trained to learn a shared latent space while fixing the unimodal encoders, enabling training without storing the unimodal encoders in memory. Limited multi-modal data may also be augmented by generating synthetic data between commonly-labeled pairs in the respective modality's latent spaces. The effect of data diversity can also be determined by generating a diverse data set with respect to the data points in latent space, enabling measurement of performance of the multi-modal model on limited training data.

Multi-modal models learn a joint latent space for relating data points across different modalities. To more effectively learn multi-modal models with reduced training requirements and greater benefit from limited multi-modal training data, a multi-modal model may be trained with fixed or pre-trained unimodal encoders that generate data representations in respective latent spaces. The multi-modal model is trained to learn a shared latent space while fixing the unimodal encoders, enabling training without storing the unimodal encoders in memory. Limited multi-modal data may also be augmented by generating synthetic data between commonly-labeled pairs in the respective modality's latent spaces. The effect of data diversity can also be determined by generating a diverse data set with respect to the data points in latent space, enabling measurement of performance of the multi-modal model on limited training data.

The invention discloses a method and a system for collecting decision data by an intelligent agent driven by a multi-modal large model. The method comprises the following steps: collecting original data of a predetermined sensor, calculating information entropy and a normalized mass score, and generating a fusion feature vector through self-adaptive multi-modal information fusion; time sequence sensitive multi-mode alignment processing is carried out on the image, and time sequence alignment features are obtained by combining timestamp embedding and a self-attention mechanism; based on this, decision distribution is generated by using a Bayesian neural network, uncertainty is evaluated, and a final decision and index set is generated; evaluating a task score in combination with task description, and making an execution plan according to a resource state; initializing the model, generating synthetic data, optimizing the model and storing experience. According to the invention, the efficient decision-making process of the intelligent trolley is realized, and the execution capability and task completion precision of the intelligent trolley are improved.

Multi-modal models learn a joint latent space for relating data points across different modalities. To more effectively learn multi-modal models with reduced training requirements and greater benefit from limited multi-modal training data, a multi-modal model may be trained with fixed or pre-trained unimodal encoders that generate data representations in respective latent spaces. The multi-modal model is trained to learn a shared latent space while fixing the unimodal encoders, enabling training without storing the unimodal encoders in memory. Limited multi-modal data may also be augmented by generating synthetic data between commonly-labeled pairs in the respective modality's latent spaces. The effect of data diversity can also be determined by generating a diverse data set with respect to the data points in latent space, enabling measurement of performance of the multi-modal model on limited training data.

An agentic workflow system and method generate question and answer pairs and prompts that may be used to aligns generative artificial intelligence (a large language model (LLM) or a large multimodal model (LMM)) with the principles of a specific domain so that the generative artificial intelligence is better able to respond to a user query in the specific domain. The system and method may also generate aligning processes that may be used to post-train an already trained generative artificial intelligence system or fine tune the training of the generative artificial intelligence system to align that generative artificial intelligence system with the principles of the specific domain. The system and method may be used to align the generative artificial intelligence system to a plurality of different domains.

The invention discloses a semi-supervised identification method and system for a small target of a container, and belongs to the technical field of artificial intelligence and computer vision, and the method comprises the steps: carrying out the target detection of a container image through a pre-trained target detection model, intercepting a sub-image, and inputting the sub-image into an initial classification model, and obtaining a classification confidence coefficient; the uncertainty of the model on a sample classification result is quantified through a Monte Carlo Dropout method; a feature space distance filtering and dynamic threshold adjusting mechanism is combined, and samples with high confidence, low uncertainty and consistent feature space are screened out to serve as pseudo label data; pseudo label data and initial synthesis data are mixed, and the generalization ability of the model is gradually improved through semi-supervised iterative training. According to the method, the dependence on manual annotation can be remarkably reduced, meanwhile, the distribution difference between synthetic data and real scene data is gradually reduced, and finally, high-precision recognition and strong generalization ability of a classification model in a real scene are achieved.

The invention relates to the technical field of power equipment operation and maintenance management, in particular to a power equipment asset health management and predictive maintenance service system which comprises a data acquisition and integration module, a feature engineering module, a health assessment and prediction engine maintenance decision and early warning module and a service interface module. The data acquisition and integration module acquires equipment operation parameters through multiple types of sensors, and associates pre-stored equipment asset information to generate an equipment comprehensive data stream; the feature engineering module cleans and standardizes the equipment comprehensive data stream, and constructs a space-time correlation feature matrix; the health assessment and prediction engine comprises a health state assessment unit and a fault prediction unit, the health state assessment unit outputs a health index HI by using a gradient boosting decision tree, and the fault prediction unit outputs a fault probability and a remaining service life RUL in a future preset time period; and the maintenance decision and early warning module generates a grading early warning signal and a maintenance strategy scheme. The intelligent level of operation and maintenance of power equipment is improved, reliable operation of the equipment is guaranteed, and the operation and maintenance cost is reduced.

The invention relates to the technical field of coal mine disaster monitoring, in particular to a gas coal mine goaf coupling disaster monitoring method, an electronic device and a storage medium, and the gas coal mine goaf coupling disaster monitoring method comprises the following steps: obtaining comprehensive data of a goaf through a sensor network, and carrying out data preprocessing to obtain standardized data; constructing a digital twinborn model through a 3D modeling and geographic information system technology; performing data synchronization and model correction processing on the digital twinborn model, and optimizing parameters and states of the digital twinborn model; constructing a fluid-solid-heat multi-field coupling model through a computational fluid mechanics and finite element analysis method; constructing a risk assessment model through an analytic hierarchy process and a fuzzy comprehensive evaluation method; training a disaster early warning model by using a machine learning algorithm based on the standardized data and the risk assessment model; a digital twinborn model, standardized data, a risk assessment model and a disaster early warning model are integrated, and a visual interaction platform is constructed. According to the monitoring method, multiple disaster elements can be coupled and dynamically evolved, and the disaster prediction accuracy and timeliness are greatly improved.

The invention discloses a P4 switch processor core-oriented multi-service traffic packet scheduling method, which mainly comprises the following steps of: proposing a feature extraction method based on a statistical interval and Sketch, integrating data packet time correlation and traffic dynamic change, optimizing a Bloom filter conflict detection mechanism, and improving the adaptability and accuracy of an algorithm in a limited resource environment; a decision tree model is combined with scheduling, a decision tree is reconstructed to adapt to an operation mode and a data processing requirement of a P4 switch, and high-precision service flow classification under linear speed operation is ensured; a packet scheduling method AD-PIFO is designed based on a variable priority, the forwarding priority of a data packet is dynamically adjusted, the QoS requirements of different services are met, and the overall performance and efficiency of the switch are improved. The method is outstanding in high-priority traffic throughput and low-delay performance, is superior to similar algorithms in medium-low priority traffic throughput, is stable and effective under different load conditions, is reasonable in extra time overhead of data packet transmission, and is widely applied to SDN network optimization data packet scheduling scenes.

Systems and methods for Data Loss Protection (DLP) utilizing distilled models include receiving a plurality of general data predictions from a teacher model; determining one or more strengths of the teacher model based on the received general data predictions; generating a synthetic dataset based on the one or more strengths of the teacher model; providing the synthetic dataset to the teacher model and receiving a plurality of synthetic data predictions from the teacher model based thereon; and performing knowledge distillation on a student model based on the synthetic data predictions received from the teacher model to produce a distilled model. The distilled model is then used in production for classifying inputs to a DLP system.

The invention discloses a lithium ion battery health state estimation method and system, a medium and equipment, and the method comprises the steps: enabling each client of a battery mechanism participating in collaborative modeling to extract a health factor based on the original data of local battery monitoring, constructing and training a deep convolution generative adversarial network, and obtaining local synthetic data; each client uploads the local synthetic data to a central server, and the central server aggregates the local synthetic data of each client to form global synthetic data and distributes the global synthetic data to each client; each client carries out federated transfer learning training based on local data and the global synthesis data, uploads a local model to a central server after local training is finished, carries out dynamic weighted aggregation to form a global model and issues the global model, and the local training and federated aggregation are continuously and iteratively executed until the whole training process is finished; and each client carries out personalized fine tuning on the global model based on local data to realize estimation of the health state of the battery.

The invention discloses a computer memory bank fault prediction method and system based on deep learning, and relates to the technical field of computer hardware fault diagnosis, and the system comprises a multi-source time sequence data collection module which is used for obtaining memory bank operation state data in real time; the dynamic feature enhancement module is based on a composite architecture of a generative adversarial network and transfer learning, comprises a fault mode generator, and generates synthetic data consistent with real fault distribution by using an LSTM network; aligning feature spaces of different hardware platforms through a maximum mean difference loss function; the multi-modal fusion deep learning model comprises a space-time convolutional network, a graph attention network and an adaptive weight adjustment mechanism; and the fault early warning analysis module is used for analyzing a fault probability predicted value, an interpretable thermodynamic diagram and a maintenance suggestion. According to the invention, passive maintenance is changed into active prevention and control, and preposition and precision of fault management are realized through dual mechanisms of long-term trend prediction and short-term risk early warning.

The present application relates to the field of behavior recognition technology, and discloses a behavior recognition method, device, equipment, storage medium and computer program product, including: obtaining multi-source heterogeneous data collected by a multimodal sensor network, and performing spatiotemporal synchronization and fusion processing on the multi-source heterogeneous data based on a spatiotemporal synchronization algorithm to obtain target comprehensive data; extracting low-dimensional feature vectors from the target comprehensive data, and combining the attention mechanism to perform time series modeling on the low-dimensional feature vectors to construct a behavior dynamic model; performing behavior recognition on the real-time collected operation process information according to the behavior dynamic model, and outputting the personnel behavior category. By integrating multi-source heterogeneous data in the complex environment of the power plant, and accurately extracting key features, and capturing the dynamic change laws of behavior and state, the model can accurately distinguish various types of personnel behaviors under normal operating procedures or complex and changeable working conditions, and realizes real-time and accurate recognition of violations and abnormal behaviors in power plant operations.

Techniques for determining a presence of an object, especially an object such as animal or debris, in a path of a vehicle, are discussed herein. For example, sensors of various modalities, which may include multispectral sensors, may capture data representing an environment the vehicle is traversing. In examples, one or more trained machine learned (ML) models, operating on a vehicle computing system, may detect and / or classify objects in the environment, based on input data of one or more modalities or spectral bands. The ML models may be pre-trained using training data including real sensor data, synthetic data, and / or augmented data, along with auto-generated annotations. In some examples, hyperspectral data may be used to identify materials associated with detected objects. A confidence score associated with the detection of the object may also be computed. The vehicle may be controlled based on detection of the object and its classification.

The invention provides a multi-data fusion fire-fighting early warning method and platform, and relates to the technical field of fire fighting, and the method comprises the steps: collecting a multi-dimensional environment data set of a target region; performing noise filtering on the multi-dimensional environment data set to obtain a multi-dimensional environment de-noised data set; comprehensively analyzing the plurality of data sources based on the multi-dimensional environment de-noised data set to generate fire-fighting comprehensive data, and fusing the multi-dimensional environment de-noised data set to obtain a multi-dimensional environment fused data set; building a fire risk model according to the fire-fighting comprehensive data, synchronizing the multi-dimensional environment fusion data set to the fire risk model for evaluation according to a plurality of data sources, and obtaining a fire risk level of the target area; and generating a fire-fighting early warning instruction based on the fire risk level, and sending the instruction to the remote control terminal. According to the method, the technical problem that a traditional fire early warning method is single in monitoring means and is easily interfered by environmental factors to cause false alarm or missing alarm is solved, and the accuracy and response speed of fire early warning are remarkably improved.

The invention relates to the technical field of wind field simulation, and provides a wind field numerical simulation method based on multiple meteorological data sources. The objective of the invention is to solve the problems of large simulation error, rough terrain boundary processing and poor turbulence model parameter adaptability caused by non-uniform coverage of a single data source, insufficient precision and unscientific multi-source fusion. The method is characterized by comprising the following steps: step 1, constructing a CFD three-dimensional grid based on a geometric model of a target area; 2, collecting original data of a ground station, satellite remote sensing, numerical forecasting and the like; 3, performing standardized preprocessing (abnormal value elimination, missing interpolation, radiation / geometric correction and resampling), determining a multi-source fusion weight by combining historical data analysis, and generating comprehensive meteorological data by adopting a weighted average method; and 4, inputting the comprehensive data into a CFD-RANS model, dynamically adjusting turbulence parameters, accurately setting terrain boundary conditions, and obtaining a wind field space-time distribution result through numerical solution. According to the invention, through combination of multi-source data fusion and CFD simulation, the wind field simulation precision and stability are improved.

The invention relates to a power system scheduling method and device based on multi-modal data fusion. The method comprises the following steps: acquiring multi-modal data of a power system, wherein the multi-modal data comprises power grid topology data, meter information, wind and light energy output prediction data, maintenance plan text data and energy comprehensive data; performing feature extraction, space-time alignment processing and hierarchical attention learning on each modal data to obtain a fusion feature vector; training the scheduling model by using the fusion feature vector until a total loss function meets a preset condition, the total loss function comprising a power flow equation constraint term; the method comprises the following steps: acquiring real-time multi-modal data of a power system, and obtaining a real-time fusion feature vector through feature extraction, space-time alignment processing and hierarchical attention learning; and inputting the real-time fusion feature vector into the trained scheduling model to obtain a real-time scheduling decision, and scheduling the power system according to the real-time scheduling decision. The method can improve the dispatching precision of the power system.

The invention relates to the technical field of monitoring and identification, and discloses a rainfall type group landslide risk area and hidden danger point comprehensive identification and evaluation method. The method comprises the following steps: determining a landslide danger index and a disaster-bearing body vulnerability index of each calculation unit in a target area according to an updated comprehensive database; determining a risk evaluation result according to a landslide risk index of each sub-region determined by the landslide risk index and the disaster-bearing body vulnerability index, wherein the risk evaluation result comprises a landslide risk grade corresponding to each sub-region; determining a deformation area in the target area according to synthetic aperture radar data; determining a hidden danger point identification result corresponding to the deformation area, wherein the hidden danger point identification result comprises disaster hidden danger points in the deformation area and hidden danger risk levels corresponding to the disaster hidden danger points; adjusting a hidden danger point identification result according to the risk evaluation result; and adjusting a risk evaluation result according to a hidden danger point identification result and a disaster hidden danger point treatment result. According to the method, dynamic feedback and optimization of risk and hidden danger assessment can be realized, and misjudgment is reduced.

The invention provides a method for predicting the quality of a sparse small sample industrial process based on a stacked network model. The method comprises the following steps: collecting end point quality report data of an industrial production process; performing hierarchical processing on the acquired data according to the missing rate, and removing abnormal data in combination with a quartile method and production experience; generating a synthetic data expansion small sample data set by adopting a conditional generative adversarial network; obtaining a first-layer basic model based on an accumulated contribution rate screening method of an SHAP value; constructing a first layer of a stacked integrated learning model and adjusting hyper-parameters by using Bayesian optimization; constructing a Ridge meta learning device to integrate the output of the basic model and constructing a second-layer network; a six-fold cross validation training model is adopted; predicting performance through a multi-index quantitative model based on the test set; and verifying the prediction precision of the end-point phosphorus content and the temperature by using real converter production data. The method can realize high-precision prediction of the end point quality index of the complex industrial generation process, and is beneficial to ensuring the product quality and improving the production efficiency.

The invention relates to the technical field of artificial intelligence, and discloses a neural network model training method based on virtual reality and simulation and related equipment. The method comprises the steps of generating a three-dimensional visual virtual reality scene based on a target real scene and constructing a dynamic physical simulation environment; deploying the neural network model to a test vehicle digital twin execution model for reasoning, and outputting a reasoning result; identifying a to-be-optimized target test scene type of the neural network model based on a reasoning result; dynamically adjusting scene parameters of the dynamic physical simulation environment to simulate a target test scene type, and generating a high-fidelity synthetic data set; and performing iterative training on the neural network model by mixing the high-fidelity synthetic data set and the real data set to form a closed-loop optimization mechanism until the generalization ability of the neural network model in the complex scene reaches a preset threshold. Based on the method, the training efficiency and generalization ability of the neural network model in a complex scene can be improved.

The invention provides a double-chain fusion product quality tracing method and system applied to a production line. The method comprises the following steps: firstly, collecting comprehensive data of each production node and attribute information containing a unique stage number; according to the total number of production stages, randomly selecting nodes with different numbers as initial clustering centers, and grouping all the nodes according to stage attributes by using a reverse clustering algorithm to form a node cluster; and constructing a basic traceability chain for each node cluster, and integrating data in corresponding stages to generate a distributed account book. And then screening core nodes from each basic chain to establish a cross-stage quality tracing chain, and realizing synchronous updating and inter-chain data verification of the core nodes on basic account books through an intelligent contract. And finally, a whole-process data tracing service is provided based on a quality tracing chain, and the information credibility is guaranteed by using the non-tampering characteristic of the block chain. According to the invention, product quality tracing with high reliability and high credibility can be realized.

Systems and techniques are described herein for adapting a pretrained machine learning model. For instance, a process can include encoding a training image into a first feature vector, the training image including a first object located at a first location; generating a second feature vector based on a set of sinusoidal functions using a set of weights; combining the first feature vector with a second feature vector to generate a combined feature vector; processing the combined feature vector using a visual language model to obtain a second location for the first object; and adjusting the set of weights based on a comparison between the first location and the second location.

A bridge component identification method based on unmanned aerial vehicle point cloud reconstruction and three-dimensional synthetic data is characterized in that a point cloud is generated through video acquisition of an unmanned aerial vehicle, accurate identification of bridge components is realized by combining with an improved PointNet + + model, and a technical process of data acquisition-point cloud generation-model training-component identification is constructed. The method comprises the following specific implementation steps: (1) acquiring a bridge member field video by using an unmanned aerial vehicle; (2) building a three-dimensional reconstruction framework to generate dense point cloud of bridge components; and (3) generating diversified bridge point cloud synthetic data in batches. (4) constructing a point cloud semantic segmentation data set with labels, and preprocessing and enhancing the point cloud semantic segmentation data set for model training; (5) realizing point cloud identification based on an improved PointNet + + model; and (6) post-processing an identification result, and extracting geometric parameters of the component for bridge structure state evaluation. According to the invention, efficient generation and high-precision identification of the point cloud of the bridge member can be realized, and reliable data support is provided for quality evaluation of the bridge structure.

The invention discloses a method for testing dynamic comprehensive data of an electric cylinder in real time, and relates to the technical field of mechatronics test.The method comprises the steps that vibration signals, temperature data and electromagnetic interference data of the electric cylinder are collected and preprocessed, and a multi-modal data stream is generated; defining a spline basis function based on the vibration signal, and extracting a transient feature tensor of the vibration signal by using the spline basis function; constructing a transient environment state transition model based on the multi-modal data stream, inputting the transient characteristic tensor, the temperature data and the electromagnetic interference data into the transient environment state transition model, and generating a displacement compensation amount and a hidden state sequence; and constructing a digital simulation model, inputting the displacement compensation amount into the digital simulation model, and generating multi-physics field coupling simulation displacement. By constructing the spline basis function and the transient environment state transition model, the problem of vibration signal feature distortion is solved, and the extraction precision and the time resolution of the transient impact features are remarkably improved.

The invention discloses a semi-supervised seismic wave impedance inversion method and system based on Swin Transform iterative gradient correction, and belongs to the technical field of earthquakes in geophysical exploration. According to the method, the impedance gradient is iteratively corrected in the model space through the deep neural network, and complex nonlinear prior information is captured; and each anisotropic structure direction regularization item is established, and constraint is carried out along the direction of the stratum structure, so that the space consistency of the prediction result is ensured. Meanwhile, the USTNet network designed by the invention combines the advantages of the local feature capture capability of the convolutional neural network and the global modeling capability of Transform, and reduces the non-uniqueness of solutions. The network is used for inverting synthetic data and actual data, and the effectiveness of the method is verified.

The invention discloses a robot operation track generation method based on structure perception and knowledge enhancement reasoning, and belongs to the field of artificial intelligence and machine learning, and the method comprises the steps: building a large-scale multi-modal database through obtaining sampling data of a real robot operation scene and synthetic data generated by a simulation environment; multi-source three-dimensional perception data is obtained, a large language model with a structure perception capability is input, and a machine execution instruction system with relatively strong three-dimensional space interaction understanding and dynamic modeling capability is obtained; multi-task progressive training is carried out based on spatial information understanding and interactive operation, capacity progressive unlocking is achieved, and robot tasks are strongly oriented; a large model knowledge graph enhancement method based on a three-dimensional scene object is introduced to assist in large model reasoning, and a large model which has reasoning ability and structure sensing ability and is higher in robustness to complex scenes is constructed and used for generating robot operation tracks.

The invention discloses a wind power prediction system for optimizing a neural network based on a genetic algorithm, relates to the technical field of new energy power system prediction, and improves the precision and adaptability of wind power prediction by fusing the genetic algorithm and a deep neural network. The system adopts multi-objective genetic optimization, randomly initializes a neural network parameter combination, evaluates the fitness by taking a prediction error and model complexity as double objectives, and screens out an optimal network architecture through evolution operation; in the aspect of neural network training, the system adopts an LSTM and TCN hybrid network as a basic model, dynamic weighting input features of a meteorological attention mechanism are combined, a learning rate and regularization parameters are optimized by using a genetic algorithm, model convergence is accelerated, and overfitting is prevented; in addition, for the space-time imbalance of the wind power data, a generative adversarial network is introduced to generate synthetic data in an extreme weather scene, and the generalization ability of the model is enhanced.

Systems and techniques are described herein for adapting a pretrained machine learning model. For instance, a process can include encoding a training image into a first feature vector, the training image including a first object located at a first location; generating a second feature vector based on a set of sinusoidal functions using a set of weights; combining the first feature vector with a second feature vector to generate a combined feature vector; processing the combined feature vector using a visual language model to obtain a second location for the first object; and adjusting the set of weights based on a comparison between the first location and the second location.

The invention discloses a personalized federated learning method based on knowledge fusion distillation and a storage medium, and belongs to the technical field of personalized federated learning, and the method comprises the steps: a server carries out the pre-training and distributes a diffusion model to each client, and generates a local synthesis data set, initializing a global model as a student model of a client to perform knowledge fusion distillation training, and guiding the student model training by the teacher model optimized in the last round; and after training is completed, the client uploads student model parameters to the server for federal aggregation to update the global model, and meanwhile, the personalized feature extraction capability and reliability of the next round of teacher model are enhanced by using synthetic data, so that a closed-loop learning framework of global cooperation and local personalized collaborative optimization is formed. According to the method, the problems of client model drifting, performance attenuation and convergence rate slowing caused by data heterogeneity can be solved.