16results about How to "High data efficiency" patented technology

The invention discloses a thermodynamic system analysis method and system based on a physical information neural network. The method comprises the following steps: constructing a thermodynamic system mathematical model; converting a solving problem of the mathematical model of the thermodynamic system into an optimization problem taking a minimum control equation set residual error as a target function; constructing a physical information neural network; constructing a composite loss function based on the control equation set residual sum of squares, and performing end-to-end training on the physical information neural network to obtain a trained physical information neural network; taking the trained physical information neural network as an approximate solver, and for any given thermodynamic system input working condition, predicting a group of approximate solutions as initial predicted values through one-time forward propagation; and inputting the initial predicted value into a traditional numerical solver for accurate solving to obtain a to-be-solved state variable. According to the method, a feasible technical path is provided for applications with strict requirements on the analysis and calculation speed of the thermodynamic system, such as real-time simulation, online optimization and digital twinning.

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 invention discloses an online monitoring alarm network system based on a lightning protection environment, and relates to the technical field of lightning protection monitoring, and the system comprises a data collection module which is disposed at an offshore wind power plant and is used for collecting multi-dimensional data including salt mist concentration, wind speed, lightning parameters and lightning protection equipment state parameters; the data processing module is in communication connection with the data acquisition module and is used for receiving, fusing and analyzing the multi-dimensional data, performing scoring calculation by constructing a quantitative scoring model and endowing each dimensional parameter with a weight, and outputting a quantitative risk level according to a total score interval judgment mode; the early warning and linkage module is in communication connection with the data processing module and is used for executing graded early warning and linkage control based on the risk grades; the data processing module is further used for calculating the health state index of the SPD and dynamically predicting the remaining effective life of the SPD based on the health state index and the real-time environment parameters. The system has the functions of self-adaption to environment change and lightning protection monitoring and early warning.

The application provides a ship-vehicle-person virtual image establishment method and system, which comprises the following steps: obtaining ship, personnel and vehicle data as source data for preprocessing, calculating and mining labels according to the preprocessed source data, and producing ship, personnel and vehicle labels; defining hierarchical labels; defining group labels; extracting basic attributes of the ship, personnel and vehicle; defining personalized labels, adding personalized labels to the ship, personnel and vehicle according to their basic attributes and respective source data; defining correlation relationship labels among the ship, personnel and vehicle; creating a label table in a database, adding the calculated labels into the label table, and then performing correlation analysis on the respective source data of the ship, personnel and vehicle and the label table according to business requirements to obtain index data; and providing the index data to a ship-vehicle-person virtual image system for drawing charts. The application can improve traffic operation efficiency, safety and risk prediction.

The utility model belongs to the field of torque measurement, and particularly relates to a ball screw friction torque measuring device which comprises a base assembly, a fixing plate and a measuring assembly. A sliding groove is formed in the top of a base of the base assembly, and four mounting blocks distributed in a rectangular shape are arranged on the outer surface. A motor and an electric push rod are installed on the fixing plate. The motor transmits power through a coupler. The measuring assembly comprises a connecting block, a first clamping plate, a second clamping plate and a long bolt, the first clamping plate and the second clamping plate are embedded into the first torque sensor and the second torque sensor respectively, the bottom of the connecting block is in sliding fit with the sliding groove, and the output end of the electric push rod is fixedly connected with the connecting block. The electric push rod drives the measuring assembly to move on the sliding groove, clamping and position adjusting of the ball screws of different specifications are achieved, the double torque sensors are used for synchronously collecting friction torque data, and the measuring efficiency and reliability of the friction torque of the ball screws can be effectively improved.

The invention discloses a pathological image segmentation method and system based on coevolution generation type difficult sample mining, and the method comprises the steps: constructing a mask synthesis engine guided by biological information, and generating a cell nucleus mask through introducing a cell affinity matrix and structure prior based on a graph theory; establishing a segmentation-oriented adversarial renderer, and aligning the generated image with a real image in a feature space by using a multi-layer feature discriminator; implementing a closed-loop co-evolution strategy, dynamically identifying vulnerability categories by using performance feedback of the segmentation model, and guiding a generator to carry out adaptive difficult sample mining; and obtaining a cell nucleus segmentation result through alternate mutual promotion of the generator and the segmentation model and regression fine tuning of real data. According to the method, the problems that in existing small sample learning, generated data lacks biological rationality and visual fidelity cannot be converted into segmentation performance are solved, and the segmentation precision and generalization ability of the model are remarkably improved under the condition of extremely few labeled data.

The invention discloses an unstructured data report filling method and system based on a large language model, and belongs to the technical field of artificial intelligence, and the method comprises the steps: preprocessing unstructured data, and generating preprocessed unstructured text data; based on the preprocessed unstructured text data, constructing and executing a semantic structured extraction chain; field mapping and report template binding are carried out based on the constructed semantic structured extraction chain, and a field mapping rule is established; based on a field mapping rule, automatically generating and filling a report, and automatically generating the report according to the extracted structured data; and interpretable proofreading and manual auditing are carried out on the automatically generated report. According to the method, implicit information can be accurately extracted, and semantic inference and association are supported; automatic report filling is achieved, and the manual arrangement workload is remarkably reduced; and the output is explainable, and each field can be positioned to the original text, so that the auditing credibility is improved.

The application relates to the technical field of electric power inspection, and discloses a method and system for power grid inspection and identification of an unmanned aerial vehicle, which comprises the following steps: acquiring channel path data, tower position data and inspection task configuration data; generating a channel inspection main flight path and a tower fine inspection sub-flight path; establishing a channel inspection mode parameter set and a tower fine inspection mode parameter set, and setting a mode switching trigger area; controlling the unmanned aerial vehicle to execute the channel inspection mode along the channel inspection main flight path; after completing tower inspection, executing key component collection integrity judgment, and executing supplementary shooting control when there are uncovered key components or unqualified images; then switching back to the channel inspection mode to continue flying; and identifying and analyzing the channel inspection data and the tower fine inspection data, and performing unified spatial marking and task association output. The application realizes integrated execution of channel inspection and tower fine inspection, reduces repeated flight, and improves inspection data integrity.

The invention discloses a lithium ion battery health state estimation method based on physical information Transform. The method comprises the following steps: acquiring voltage and current time sequence data of a lithium ion battery under a pulse charging condition; feature parameters are extracted from the time series data, and a training data set containing the feature parameters and corresponding health state labels is constructed; and constructing a physical information neural network model based on a Transform architecture, inputting the training data set into the physical information neural network model for training, processing pulse charging data of the lithium ion battery to be tested by using the trained physical information neural network model, and outputting a battery health state estimation result. According to the method, a framework which is high in principle and can be popularized is established, and an important contribution is made for constructing a battery health prediction model which is higher in interpretability, better in data efficiency and higher in physical credibility.

The application discloses a six-degree-of-freedom object pose estimation system, method and medium in an industrial scene, and the system comprises a multi-modal data preprocessing module, a shared encoder feature extraction module, a contrast learning and attention fusion module, a pose regression and rotation decoupling module and a simulation-driven data enhancement module, the multi-modal data preprocessing module is used for receiving and processing original RGB images and depth maps, and through filtering denoising and coordinate projection transformation based on camera parameters, an RGB-D data pair of pixel-level spatial alignment is output; the shared encoder feature extraction module is used for extracting features from the aligned RGB-D data pair, and adaptively fusing through a channel attention mechanism to output a shared feature fused with appearance and geometric information. The application can improve the estimation accuracy, the ADD index is improved by 12.8% and the rotation error is reduced by 21.4% under a 70% occlusion or reflection scene, and the application has strong robustness to depth noise, occlusion and reflection and strong practicability.

The present application relates to biological signal processing technology, and is a non-invasive continuous blood pressure estimation system based on a subject adaptive feature modulation mechanism, an individual prior embedding vector is generated by an individual prior embedding vector generation module; a subject adaptive feature extraction module performs real-time multi-level dynamic modulation on the preprocessed time series physiological signal according to the individual prior embedding vector, and outputs a high-dimensional time series feature map; a sequence processing module performs time series dependence modeling and context refining on the high-dimensional time series feature map, and outputs a time series feature; a cross-modal attention fusion module splices the individual prior embedding vector and the time series feature; the spliced feature vector is input into a multi-head attention fusion module to realize dynamic cross-modal feature weighted fusion and obtain a high-dimensional feature vector; and a blood pressure regression output module maps the high-dimensional feature vector through a full connection layer to regress to obtain an estimated value of systolic pressure and diastolic pressure. The present application solves the problem that the existing static mapping paradigm cannot cope with individual heterogeneity.

The application provides an airborne wind measurement laser radar dynamic adaptive attitude-position collaborative compensation system and method which can dynamically adapt to flight states and accurately couple attitude and position interference. The system comprises a GNSS positioning module (1), a MEMS-IMU attitude sensing module (2), a two-stage coordinate conversion module (3), a dynamic speed solving module (4), an adaptive collaborative compensation module (5) and a laser radar wind measurement module (6). The method comprises the following steps: S1, multi-source data synchronous acquisition, S2, ECEF coordinate system conversion, S3, ENU coordinate system conversion, S4, dynamic speed solving, S5, dynamic weight adaptive compensation, S6, data output and feedback storage. The application is applied to the technical field of wind measurement laser radar.

The application discloses a physical information multi-fidelity neural network three-dimensional cascade flow field reconstruction method based on Fourier characteristics, and relates to the field of aero-engine turbine machinery aerodynamic design. The method comprises the following steps: generating high / low fidelity data sets with boundary consistent and gradient perception sampling; constructing a multi-fidelity neural network containing Fourier characteristic embedding, multi-fidelity collaborative network and RANS equation residual constraint; adopting a two-stage training strategy, first optimizing data fitting loss to learn the flow field mapping relationship, and then jointly improving the conservation with physical loss; finally, realizing fast, high-precision and physically consistent reconstruction of three-dimensional cascade space coordinates to key physical quantities such as velocity, static pressure and temperature. The method significantly reduces the dependence on high-fidelity data, improves the prediction accuracy in high gradient areas, and is suitable for rapid iterative design of aero-engine cascades.

A reinforcement learning system, method, and computer program code for controlling an agent to perform multiple tasks while interacting with an environment. The system learns options, where an option comprises a sequence of primitive actions performed by the agent under the control of an option policy neural network. In implementations, while the agent interacts with the environment, the system discovers options useful for multiple different tasks by a meta-learning reward for training the option policy neural network.

This invention relates to the field of digital psychology and provides an AI-based method and system for crawling dynamic multi-modal psychological test questions. The method includes the following steps: obtaining the URL of the target psychological test page and selecting a set of device configuration parameters that includes at least desktop and mobile versions according to a preset device type library; loading and rendering the target psychological test page according to each configuration parameter in the device configuration parameter set, and crawling the complete document object model tree presented in each simulation environment; using AI to compare multiple document object model trees to obtain unique content nodes between versions, and selecting effective nodes to merge into a unified content tree structure; generating standardized test question data based on the unified content tree structure. This invention achieves efficient, accurate, and automated crawling of dynamic multi-modal psychological test questions by simulating multiple client access environments and introducing an AI-driven intelligent comparison, filtering, and fusion mechanism.

This invention discloses a knowledge-guided method and system for predicting metal-ion-organic interactions, belonging to the fields of environmental separation and computational chemistry. The method constructs a training set through DFT calculations, extracts organic molecule descriptors, and quantifies their prior relationship with binding free energy. It employs a graph attention network to extract molecular features and fuses metal ion features using a cross-attention mechanism. A loss function incorporating prior knowledge is designed, and high-precision prediction is achieved through two-stage training. This invention solves the problem of traditional machine learning's dependence on large amounts of labeled data, maintaining excellent performance even in sparse data scenarios. It is suitable for high-throughput screening of separation agents such as extractants and adsorbents, significantly reducing R&D costs.