43 results about "Uncertainty estimate" patented technology

In various examples, systems and methods for uncertainty estimation for object detection in autonomous and semi-autonomous systems and applications are provided. The systems and methods may use data from one or more sensors (e.g., camera(s) and / or LiDAR sensor(s) to generate a representation of features surrounding a machine. A model may be used to generate probabilities of objects being present in the representation of features and uncertainty estimates corresponding to the object presence probabilities. The uncertainty estimates may be used to identify scenes that are significantly different from the training data, detect errors in the bounding shapes for objects, and / or highlight areas where object detections may have been missed. The systems and methods may also be used to auto-label scenes associated with the representation of features, and the auto-labeled scenes may be used for training purposes.

The invention provides a method and system for on-line determination of molten steel carbon content in converter steelmaking, and relates to the technical field of converter steelmaking, the method comprises the following steps: obtaining multi-source process data, the multi-source process data comprising exhaust gas components, furnace mouth flame multispectrum and process parameters; performing time synchronization and feature extraction on the multi-source process data to generate a feature vector; inputting the feature vector into a carbon content prediction model to obtain a carbon content prediction value and an uncertainty estimation value thereof; the molten pool temperature is obtained, and based on the molten pool temperature, the waste gas components and the technological parameters, a theoretical carbon content value is calculated through a thermodynamic carbon content calculation model; and carrying out weighted fusion on the carbon content predicted value and the theoretical carbon content value to generate a fused carbon content as a target carbon content. According to the method and system for online determination of the carbon content of the molten steel in converter steelmaking, the precision and reliability of online determination of the carbon content of the molten steel can be effectively improved, and powerful support is provided for intelligent production of converter steelmaking.

The invention discloses a storage tank measurement data processing method, computer equipment and a storage medium, and relates to the technical field of measurement modeling, and the method comprises the steps: collecting multi-physics field operation parameter data, carrying out the filtering, interpolation and abnormity elimination, and obtaining a multi-dimensional original data set; establishing a digital twin model based on the data set and calculating an initial parameter set; collecting real-time observation data, and executing residual optimization inversion to obtain a correction parameter set; performing uncertainty analysis and confidence weighted fusion by using the correction parameter set to obtain a fusion measurement parameter set; constructing a multi-objective cost function and updating the weight by adopting an exponential gradient evolution algorithm to obtain feed-forward control intensity; and in combination with the correction parameter set and the feedforward control intensity, carrying out cross-storage-tank robust aggregation and gating judgment to obtain new digital twinborn model parameters. According to the method, self-adaptive fusion and model closed-loop evolution of multi-source metering information are realized through confidence weighted fusion and an exponential gradient evolution weight optimization mechanism and by executing cross-storage-tank robust aggregation.

The invention discloses a damping adjusting method, device and equipment of an anti-snake-shaped shock absorber and a storage medium, belongs to the field of train shock absorption, and aims to consider that other disturbed quantities except the anti-snake-shaped shock absorber can be regarded as a whole (sum uncertainty) in relative head shaking angle information of a bogie. Therefore, the method comprises the following steps: firstly, determining a dynamic differential relational expression of the relative head shaking angle information of the bogie about the damping coefficient of the anti-serpentine damper and the sum uncertainty, and then determining a sum uncertainty estimator (used for outputting a sum uncertainty estimated value) of the bogie; according to the method, the target damping coefficient of the anti-snakelike damper can be determined based on the dynamic differential relation on the basis of the total uncertainty estimated value and the measured value of the relative head shaking angle information, so that the damping coefficient is dynamically adjusted, the method can adapt to complex and changeable operation conditions, and the transverse stability and riding comfort of a train are improved.

The invention provides a wave-rider aircraft aerodynamic data uncertainty analysis method and device, and the method comprises the steps: obtaining the numerical calculation aerodynamic data of a wave-rider aircraft; factors influencing the uncertainty of the numerical calculation pneumatic data are determined; carrying out uncertainty analysis on the basis of pneumatic data, calculating model uncertainty by adopting a range method, designing an orthogonal test according to factors, calculating numerical value and input parameter uncertainty, and carrying out significance analysis on factor influence and interaction according to the orthogonal test; and calculating the total uncertainty according to the model uncertainty, the numerical value and the input parameter uncertainty, and verifying the reliability of the total uncertainty by adopting new sample data. According to the method, the uncertainty of the pneumatic data is comprehensively and effectively obtained, meanwhile, the influence factors of the uncertainty of the numerical calculation pneumatic data and the significance of the interaction are visually obtained, and support is provided for the reliability of the numerical calculation pneumatic data.

A computer-implemented method utilizing a continuous discrete recurrent Kalman network, wherein the method includes receiving, at an encoder, an input from one or more sensors, wherein the input includes one or more time series data associating data at one or more points in time; outputting, to a Kalman filter, a latent observation and uncertainty estimate in response to the input at the encoder; determining a latent state prior and latent state posterior utilizing the Kalman filter; and outputting, via a decoder, a filtered observation utilizing at least the latent state posterior.

The application provides a rotary kiln energy consumption optimization method and system combined with visual recognition, comprising: preprocessing kiln operation data, and establishing a unified fusion feature representation; adopting a stacked long short-term memory network to process long-term dependence relationship, and simultaneously processing local time sequence mode through a multi-layer expansion convolution, and weighting and fusing the two kinds of features through a self-attention mechanism; in order to improve prediction reliability, training a plurality of basic models with different initializations and architectures, and applying a Bayesian model average technology to obtain a point estimate value, a prediction interval and an uncertainty estimate; based on the prediction model, constructing a graph structure representation of a kiln process parameter space, calculating a predicted coal consumption value of each parameter combination, thereby identifying an optimal process parameter combination, and realizing kiln energy efficiency optimization. The application improves coal consumption prediction accuracy and process parameter optimization efficiency.

The invention relates to an input parameter influence analysis method and system for nuclear power plant accident consequence evaluation, a medium and equipment. The method comprises the following steps: acquiring parameters required for evaluation; screening and setting parameters required by evaluation to generate a parameter information table; performing random sampling based on the input parameters to generate a random sampling matrix; calculating by using a random sampling matrix to obtain an accident consequence evaluation result; analyzing uncertainty and sensitivity based on the accident consequence evaluation result, and obtaining an uncertainty analysis result and a sensitivity analysis result of the input parameter; performing comprehensive analysis based on the uncertainty analysis result and the sensitivity analysis result to obtain a comprehensive analysis result; and determining an influence result of the input parameters according to the comprehensive analysis result. The uncertainty and sensitivity of the input parameters are analyzed, the influence degree of each parameter on the accident consequence evaluation result is effectively quantified, it is ensured that the nuclear power plant can make an accurate decision in time when a potential accident occurs, and the public and environment safety is effectively protected.

The invention relates to a belt weigher weighing precision optimization method based on uncertainty analysis. The method comprises the steps that multi-source reference data are acquired and synchronously collected, and standard weight information of a truck scale, real-time data information of a belt weigher and environmental parameter information of the same material batch are synchronously acquired; according to the standard weight information of the truck scale, the real-time data information of the belt weigher and the environmental parameter information, A-class uncertainty is calculated, a B-class uncertainty three-dimensional decomposition calculation model is constructed to calculate B-class uncertainty, and the uncertainty is synthesized to calculate the total uncertainty; component contribution analysis is carried out according to the error size of each component; gradient error two-dimensional control, wherein the gradient error two-dimensional control comprises hardware leveling, algorithm compensation, vibration control and temperature compensation; the comparison test is repeated after optimization, if the measurement error is not smaller than the expected threshold value, the gradient error two-dimensional adjustment control step is repeated until the measurement error is smaller than the expected threshold value, and the method has the advantages of achieving accurate targeted optimization of all error sources and the like.

This application describes a method for automatically labeling data samples with uncertainty quantification. The proposed method comprises the steps of simultaneously feeding an input data sample into an online object detection module with uncertainty estimation and an off-board object detector with uncertainty estimation; comparing the resulting data from both the online classification and uncertainty estimation module and the off-board classification and uncertainty estimation module; and deciding whether to discard the input data sample, send it for human labeling, or archive it along with the classification and uncertainty estimation from the off-board object detection module with uncertainty estimation.These uncertainty estimates are then used in the training phase of subsequent object detectors to weight different samples differently, with human-labeled samples and automatically labeled samples with high confidence receiving more weight.

This specification provides an image detection method and apparatus, a target classification model, a computer-readable storage medium, an electronic device, and a computer program product. The method includes: first, determining a target classification model, which is trained using training samples from a first data domain, meaning it is suitable for detecting images to be tested in the first data domain. Based on the model, determining a first uncertainty estimate corresponding to a first target sample set from the first data domain, and determining a second uncertainty estimate corresponding to a second target sample set from a second data domain (another data domain). Further, based on the first and second uncertainty estimates, determining a feature offset, which characterizes the feature changes between the second and first data domains. Finally, based on the target classification model and the feature offset, determining the detection result for the image to be tested in the second data domain.

The present application relates to the field of geological disaster prevention, and particularly relates to a rainfall monitoring data driven small watershed geological disaster intelligent early warning method, comprising: obtaining historical rainfall data and geological disaster historical record data of a target small watershed for preprocessing; matching geological disaster events to corresponding rainfall events to obtain a rainfall event set in which geological disasters occur and serving as positive samples, and a rainfall event set in which no geological disasters occur and serving as negative samples; using the positive samples and the negative samples to construct a data set for learning, training and testing of a machine learning model; constructing an array of small watershed geological disaster intelligent early warning machine learning models and training, verifying and testing the array using the data set; constructing a target small watershed geological disaster occurrence probability calculation model and an uncertainty analysis model under the influence of a rainfall event, calculating the target small watershed geological disaster occurrence probability and uncertainty under the rainfall event, and performing early warning; and the present application improves the reliability and accuracy of small watershed geological disaster early warning results.

A system and method for distributed fiber-optic sensing (DFOS) event classification utilizing an energy-based generative artificial intelligence model. The method leverages a joint energy-based model (JEM) to ingest both human-annotated labeled sensing data and abundant unlabeled ambient sensing data collected from a DFOS interrogator. The joint energy-based model models the data distribution of high-dimensional spatial-temporal sensing signals, enabling semi-supervised classification that improves generalization, provides calibrated uncertainty estimates, and avoids domain-specific data augmentation constraints. The system further provides a test-time refinement mechanism utilizing Stochastic Gradient Langevin Dynamics (SGLD) updates in the input data space to remove the effects of sensor noise and recover discriminative classification accuracy upon deployment in pre-existing telecom cable networks.

This invention relates to the field of drug performance evaluation technology, and more particularly to a drug performance evaluation method and system based on chemical spatial clustering. The method includes: extracting structural features of known drug molecules from a calibration dataset and calculating chemical spatial distances; clustering the calibration dataset using a density-aware clustering strategy, performing statistical calibration within each chemical similarity cluster, and constructing a chemical spatial hierarchical calibration framework; using a graph neural network ensemble model to predict the drug performance of the drug molecule under test, obtaining prediction results and uncertainty estimates, determining the weights of each chemical similarity cluster based on structural features and the chemical spatial hierarchical calibration framework, and generating confidence intervals; adjusting the sensitivity of the confidence intervals according to the safety level corresponding to the drug performance under test, and outputting the prediction results and confidence intervals after a credibility assessment. This invention provides statistically reliable quantitative information for differentiated decision-making in early-stage drug development.

A software tool for predicting a candidate recipe for a food product using: (a) a predictor model trained to output, for a given candidate recipe passed as input to the predictor model, (i) a predicted value for at least one target variable and (ii) a predicted value for a given subset of evaluation variables, and (b) a generator model for: (1) training a base prediction model configured to output (i) a predicted value for at least one target variable of a space of a possible recipe and (ii) an uncertainty estimate for the predicted value, and (2) selecting a candidate recipe from the space of possible recipes based on (i) a balance between a predicted value output by the base prediction model and the uncertainty estimate and (ii) a set of constraints.

Techniques for localizing a vehicle in real time using dynamic uncertainty estimates are presented. The techniques include obtaining a terrain image captured by the vehicle; passing the terrain image to a trained evidential deep learning neural network subsystem, from which a dynamic uncertainty value and a first feature vector are obtained in real time; for each of a plurality of candidate terrain locations, comparing the first feature vector to a respective second feature vector representative of a candidate terrain location, from which a respective similarity score is obtained; for at least one of the plurality of candidate terrain locations, updating in real time, by a recursive Bayesian estimator, a respective location weight based on the dynamic uncertainty value and the respective similarity score; estimating, in real time, a location of the vehicle based on the plurality of location weights; and providing the location of the vehicle.

The invention relates to the technical field of road engineering material detection, in particular to an asphalt mixture density uncertainty analysis method and system.The method comprises the steps that an asphalt mixture detection piece is manufactured, an experimental scheme is designed, and then a test data set of the asphalt mixture detection piece is obtained; selecting an error factor in density uncertainty analysis according to the test data and the detection piece, and obtaining an uncertainty analysis result of the factor; constructing an error factor credibility evaluation model, and screening the error factors in combination with test data to obtain credible error factors; and obtaining an uncertainty comprehensive analysis result of the asphalt mixture density based on the credible error factor and the uncertainty analysis result, and realizing accurate analysis of the asphalt mixture density uncertainty. According to the method, credibility analysis and screening are carried out on error factors, the uncertainty condition of the density of the asphalt mixture can be reflected more truly, and the accuracy of an analysis result of the density of the asphalt mixture is improved.

This invention discloses a tracking control method for a flexible space robot arm based on a credibility-gated interval type-II fuzzy neural network, belonging to the field of space robot control. To address the technical problem of insufficient tracking accuracy in existing space robot arm tracking control methods, this invention establishes a dynamic model of the flexible space robot arm system considering lumped uncertainties. It then uses a credibility-gated interval type-II fuzzy neural network to approximate the lumped uncertainties in the dynamic model online, obtaining an estimate of the lumped uncertainty. The lumped uncertainty estimate, network compensation term, robust saturation term, and flexible modal active suppression term are jointly introduced into the nominal control law, simultaneously achieving high-precision joint tracking, floating base coupling compensation, and residual vibration suppression of flexible links for the floating-based flexible space robot arm under complex trajectory conditions. This method is primarily used for tracking control of flexible space robots.

The invention discloses a large-load scratching instrument calibration method based on multi-dimensional parameter uncertainty analysis, and relates to the technical field of material performance detection equipment calibration, and the method comprises the following specific steps: firstly, constructing an integrated system containing a normal force calibration module, a horizontal displacement calibration module, a friction force calibration module and a pressure head angle calibration module, and equipping related data equipment; preparing before calibration, mounting components, fixing materials and a pressure head, clearing a mark, presetting parameters and measuring times; then implementing a multi-parameter collaborative test, and collecting storage data; processing the data and carrying out uncertainty analysis; finally, determining a calibration result, giving a report if the calibration result reaches the standard, and adjusting and remeasuring if the calibration result does not reach the By constructing the integrated calibration system, multi-parameter collaborative testing is realized, full-process factors are controlled, calibration normalization and data integrity are improved, uncertainty analysis is carried out by adopting comprehensive evaluation, a judgment feedback mechanism is established, targeted adjustment and optimization are carried out, calibration precision and stability are effectively improved, the system is adaptive to multiple scenes, and the calibration efficiency is improved. Guarantee is provided for accurate operation of equipment, and the practical popularization value is enhanced.

The invention discloses a distributed task execution time prediction method based on a heterogeneous graph neural network, and the method comprises the steps: obtaining multi-source information, constructing a heterogeneous graph which comprises various types of nodes and various relationships, and employing a relationship perception feature learning mechanism and a type perception neighborhood sampling mode, different relationships and importance of neighbor nodes are distinguished, thereby generating high-order representations of tasks and resources. On the basis, the total execution time is decomposed into two parts of queuing waiting time and actual execution time, and the final execution time is predicted and synthesized based on the resource dynamic state and the task-resource interaction characteristics. Monte Carlo Dropout is further utilized to obtain uncertainty estimation, confidence levels are divided according to prediction variance, a prediction strategy is adaptively adjusted, and the robustness of the model under the condition of high load or insufficient data is improved. According to the method, the execution time prediction precision and reliability in the distributed scheduling environment can be remarkably improved.

The invention relates to uncertainty estimation of object detection in autonomous and semi-autonomous systems and applications. In various examples, systems and methods for uncertainty estimation of object detection in autonomous and semi-autonomous systems and applications are provided. The systems and methods may use data from one or more sensors (e.g., one or more cameras and / or one or more LiDAR sensors) to generate a representation of features around the machine. The model may be used to generate a probability of presence of an object in the feature representation and an uncertainty estimate corresponding to the probability of presence of the object. The uncertainty estimation may be used to identify scenes that are significantly different from the training data, detect errors in the bounding shape of the object, and / or highlight areas where object detection may have been missed. The systems and methods may also be used to automatically annotate scenes associated with the feature representations, and the automatically annotated scenes may be used for training purposes.

Parameters of a probabilistic model comprising Gaussian processes are learned. A system (100) is disclosed for learning a set of parameters of a probabilistic model having layers of multiple Gaussian processes (e.g., deep GPs) from a training data set. The set of parameters includes at least induced locations for the multiple Gaussian processes, and parameters of a probability distribution approximating outputs of the multiple Gaussian processes at the multiple induced locations. The probability distribution comprises a multivariate normal probability distribution having a covariance matrix defined by a Kronecker product of a first matrix indicating similarities between the multiple Gaussian processes and a second matrix indicating similarities between the multiple induced locations. A system (200) is also disclosed for determining one or more samples of an output of the probabilistic model for a given input using the set of parameters, e.g., to determine a mean and / or uncertainty estimate for the probabilistic model.

Several examples provide systems and methods for uncertainty estimation for object detection in autonomous and semi-autonomous systems and applications. These systems and methods can use data from one or more sensors (e.g., one or more cameras and / or one or more LiDAR sensors) to generate a representation of features surrounding a machine. A model can be used to generate probabilities of objects being present in the feature representation and uncertainty estimates corresponding to these probabilities. The uncertainty estimates can be used to identify scenes that differ significantly from the training data, detect errors in the object boundary shapes, and / or highlight areas where object detection may have been missed.The systems and procedures can also be used to automatically label scenes that are assigned to the representation of features, and the automatically labeled scenes can be used for training purposes.

This invention discloses an intelligent imputation method and system for continuous missing values ​​in time series based on context-aware generative adversarial networks. The method includes: preprocessing the original time series to generate a mask matrix; constructing a multi-scale dynamic contextual cue generator, extracting multi-scale features through a parallel temporal convolutional network, and dynamically fusing contextual information using a hierarchical attention mechanism to generate an adaptive cue matrix; constructing a time-aware generator, fusing the missing sequence, mask, noise, and cue matrix to generate imputed values ​​and uncertainty estimates; constructing a multi-scale discriminator; jointly training multiple relevant datasets using a multi-task learning framework, and jointly optimizing the model through adversarial, reconstruction, KL divergence, uncertainty calibration, and multi-task consistency loss; and using the trained model to impute missing data, outputting the complete sequence and the uncertainty at each position. This invention significantly improves the accuracy and generalization ability of continuous missing value imputation and provides reliable confidence assessment.

The invention discloses an industrial park carbon emission metering method based on a carbon emission flow theory and green electricity auxiliary service. The method comprises the following steps: S100, determining an industrial park carbon emission metering model; s200, identifying and analyzing an uncertainty source; s300, establishing an uncertainty mathematical model; s400, evaluating an uncertainty component; s500, the uncertainty is synthesized; and S600, the uncertainty is expanded. According to the method, the influence of direct carbon emission, indirect carbon emission and green electricity auxiliary service in the industrial park on the uncertainty of the carbon emission measurement data is comprehensively considered, and the uncertainty of the total carbon emission data of the industrial park is analyzed and calculated by integrating the uncertainty introduced by each influence factor; according to the method, the carbon emission measurement data is represented by the carbon emission data and the uncertainty, so that the method is more comprehensive and scientific compared with the traditional method that the carbon measurement data only comprises the data but does not have the uncertainty, and the credibility and the accuracy of the data are more intuitively expressed.

Apparatus and method for uncertainty-aware code generation using LLMs. For example, one embodiment of a method comprises: generating, by a large language model (LLM) code generator, a plurality of RTL code blocks based on a design prompt; determining syntactical similarities and semantic similarities between pairs of the RTL code blocks; arranging the RTL code blocks into a plurality of clusters based on a combination of the syntactical similarities and the semantic similarities; generating uncertainty estimates indicating levels of uncertainty associated with one or more clusters of the plurality of clusters; and determining whether to synthesize an RTL output using one or more of the RTL code blocks based on the uncertainty estimates.

The present disclosure relates to methods and systems for training and utilizing a machine-learning model with a Deep Kernel Learning with Gaussian processes (DKL-GP) architecture to handle datasets with missing values. The system can receive a dataset with incomplete data, identify missing values, and process the dataset using the DKL-GP architecture. This can involve generating latent variables, utilizing inducing variables to approximate a Gaussian process, and mapping the latent variables to output predictions with associated uncertainty estimates. The system can optimize model parameters through a training process that leverages Pólya-Gamma data augmentation and Gaussian process inducing points for efficient computation. The trained model can subsequently be used to generate predictions for data records with missing data values, while obviating the need to impute potential values for the missing values, and make decisions based on the predictions.

The application provides an OTA test uncertainty analysis method and device, comprising the following steps: measuring the EIS pattern of a main antenna and the EIS pattern of a diversity antenna; performing biasing on the EIS patterns according to a preset penalty term, to obtain biased EIS patterns; calculating a TIS value CTIS by using a standard TIS calculation method according to the obtained combined pattern Std‑i‑j ; calculating a TIS value CTIS by using a single-point compensation method SPOT‑i‑j ; calculating the difference between the two TIS values; repeating the above calculation process for each set of EIS patterns and preset penalty terms to obtain a preset number of difference samples; taking the statistical distribution parameter value of the difference samples as the uncertainty value introduced by the single-point compensation method in a multi-antenna receiving scenario; determining whether the uncertainty value is within a preset limit value range; if yes, determining that the single-point compensation method meets the accuracy requirement of OTA testing in the multi-antenna receiving scenario; if not, determining that the single-point compensation method does not meet the accuracy requirement of OTA testing in the multi-antenna receiving scenario.

The application provides a leaf area index pixel scale relative true value uncertainty analysis method and device, relates to the technical field of remote sensing product authenticity inspection, and comprises the following steps: a complete link for obtaining a leaf area index pixel scale relative true value of a heterogeneous ground surface is divided into a plurality of consecutive links; the uncertainty sources of each link are traced back; the uncertainty of each uncertainty source in each link is quantified respectively; the uncertainties of the links are synthesized to obtain the total uncertainty of the leaf area index pixel scale relative true value; and the main uncertainty sources affecting the acquisition of the leaf area index pixel scale relative true value are analyzed according to the total uncertainty analysis result, and the uncertainty of a controllable link is traced back. The application can trace the quality source and locate problems in each link in the whole process, reduce the uncertainty of the controllable link, and comprehensively improve the accuracy of remote sensing product authenticity inspection.

The invention relates to the technical field of uncertainty analysis, in particular to a combined cycle power station performance index uncertainty analysis method and system, and the method comprises the steps: obtaining equipment configuration information, equipment operation state parameters and uncertainty source parameter measured values of power generation equipment; determining a topological structure type of the generator set, selecting a thermodynamic model and an uncertainty synthesis formula of the generator set, and extracting an error range and parameter precision of a performance measurement device in the power generation equipment; inputting measured values of equipment configuration information, equipment operation state parameters and uncertainty source parameters into the thermodynamic model, outputting performance indexes of the generator set, and calculating an intermediate thermodynamic state quantity; multi-dimensional uncertainty is calculated; synthesizing the multi-dimensional uncertainty into the combined uncertainty of the generator set through an uncertainty synthesis formula; calculating extended uncertainty based on the joint uncertainty; and generating an uncertainty analysis report. According to the invention, precise quantification and standardized output of unit-level performance index uncertainty can be realized.