76 results about "Uncertainty quantification" patented technology

A method for forecasting production from a hydrocarbon producing reservoir, the method includes defining an objective function and characteristics of a history-matched model of a reservoir and acceptable error E. At least one geological realization of the reservoir is created representing a probable geological setting. For each geological realization, a global optimization technique is used to perform history matching in a series of iterative steps to obtain acceptable models. Production of the reservoir is forecasted based upon simulation runs of the respective models.

The present invention relates to systems, devices and methods for acquiring, measuring, monitoring, processing and analyzing physiological signals. More particularly, the present invention relates to using physiological signals to determine a subject's response to various conditions, variables or constraints. Still more particularly, the present invention relates to monitoring the subject's external body motion and / or environmental factors and determining the amount of pain a subject is suffering as a result of the motion and factors. Still more particularly, the present invention relates to a system, device and methods of quantifying a subject's pain to provide an objective measurement of the subject's pain. The present invention further relates to establishing and improving pain management protocols and therapy or treatment for the subject's pain based on the quantified pain measurement.

A system and method for providing multivariate climate change forecasting are provided that obtain, from one or more climate model datasets, simulated historical and future climate model data, and from one or more climate observational datasets, historical observed climate data. A statistical distribution, using a Bayesian model, is provided of extremes or climate indices for one or more variable climate features using the simulated climate model data and the observed climate data. One or more metrics are determined, including a prediction of a future climate variable for a determined future time period, a confidence bound of the prediction of the future climate variable for the determined future time period, and a prediction bound for the future climate variable for the determined future time period. The metrics can be transmitted to a variety of applications in a variety of formats.

A method for forecasting production from a hydrocarbon producing reservoir, the method includes defining an objective function and characteristics of a history-matched model of a reservoir and acceptable error E. At least one geological realization of the reservoir is created representing a probable geological setting. For each geological realization, a global optimization technique is used to perform history matching in a series of iterative steps to obtain acceptable models. Production of the reservoir is forecasted based upon simulation runs of the respective models.

A method for probabilistic fatigue life prediction using nondestructive testing data considering uncertainties from nondestructive examination (NDE) data and fatigue model parameters. The method utilizes uncertainty quantification models for detection, sizing, fatigue model parameters and inputs. A probability of detection model is developed based on a log-linear model coupling an actual flaw size with a nondestructive examination (NDE) reported size. A distribution of the actual flaw size is derived for both NDE data without flaw indications and NDE data with flaw indications by using probabilistic modeling and Bayes theorem. A turbine rotor example with real world NDE inspection data is presented to demonstrate the overall methodology.

An integrated and comprehensive method and system is disclosed for management and optimization of innovation and associated processes under uncertainty. A first embodiment of the invention consists of a data mining and clustering module to compare a new innovation submission with existing internal and external entries and databases, identify similarities and group similar entries together. A second embodiment of the invention is directed towards an intelligent machine learning module to learn from the available data of previous innovation projects and provide estimates of outputs or target values for new innovation submissions or entries. In a third embodiment of the invention, an uncertainty quantification method and system is introduced to handle uncertain inputs of innovation entries and provide probabilistic estimates of outputs by generating a plurality of solutions and scenarios. In a fourth embodiment of the invention, a multiobjective optimization module is used to simultaneously optimize multiple competing objectives.

The invention discloses a heating-power system joint economic scheduling method by using heating network energy storage features. Influences of heating network energy storage on a thermoelectric unitare analyzed; constraints when the heating network keeps safe and stable operation after the thermoelectric unit changes a heating supply extraction flow are determined; according to specific reasonsfor wind curtailment generation by a power system containing a high proportion of thermoelectric units, a heating-power system two-stage scheduling policy considering the heating network energy storage is made; the system safety risk cost caused by wind power uncertainty is calculated; and a heating-power system joint economic scheduling model in consideration of heating network heat storage and wind power uncertainty cost and with the minimum wind curtailment amount as a target is built; and a fast particle swarm optimization algorithm is adopted to solve the model. The method of the invention can improve the peaking capability of the thermoelectric unit, the two-stage thermoelectric unit scheduling policy is adopted according to the specific wind curtailment reasons of the system in theheating supply stage, and system wind curtailment is thus reduced; and during the thermoelectric unit scheduling process, the wind power uncertainty is quantified as the risk cost and is added to a scheduling objective function, and the system operation cost and the risk can be coordinated.

The invention relates to the field of computer vision, adopts a deep learning framework, and particularly relates to a pedestrian re-identification method based on uncertainty optimization, which comprises the following steps of 1) using a twin network structure to respectively use two original images belonging to the same or different pedestrians as the input of two isomorphic networks; 2) usingan inception network and a Dropout layer superposition mode to realize the Bayesian convolutional neural network with uncertainty optimization as a feature extraction network; 3) calculating binary classification loss and multi-classification loss of respective networks according to feature output of the twin network, and superposing the binary classification loss and the multi-classification lossfor back propagation and parameter optimization of the networks; 4) inputting the pedestrian image to be identified and all comparison images into the trained model, and extracting image features; 5)obtaining a final distance between the pedestrian image to be identified and the comparison image by using a Euclidean distance calculation formula; and 6) performing sorting according to the featuresimilarity distance to obtain a comparison image matching sorting corresponding to the to-be-identified pedestrian. Compared with the prior art, the method has the advantages of high accuracy, high robustness, rapidness, simplicity, convenience and the like under the conditions of all samples and few samples.

Embodiments described herein provide a system for facilitating a training system for a device. During operation, the system determines a system model for the device that can be based on empirical data of the device. The empirical data is obtained based on experiments performed on the device. The system then generates, from the system model, synthetic data that represents behavior of the device under a failure. The system determines uncertainty associated with the synthetic data and, from the uncertainty, determines a set of prediction parameters using an uncertainty quantification model. The system generates training data from the synthetic data based on the set of prediction parameters and learns a set of learned parameters associated with the device by using a machine-learning-based classifier on the training data.

The invention provides a method for analyzing large break accident of a nuclear power plant. The method comprises the following steps: S1, building a power plant model for catching major physical phenomenon and transient state of real large break loss of coolant accident; S2, analyzing and determining key parameters of the major physical phenomenon and the transient state; S3, classifying the keyparameters, wherein the key parameters are classified into at least conservative assumption parameters, sampling parameters and penalty parameters; S4, setting the most severe condition assumption forthe conservative assumption parameters, and quantitatively analyzing uncertainty of the sampling parameters and the penalty parameters so as to obtain the target parameter value at the set level; S5,selecting a penalty model from the penalty parameters, and performing penalty, so as to realize that the target parameter value obtained through the penalty parameters obtained by penalty treatment includes the target parameter value at the set level. With the adoption of the method in the embodiment, the sampling calculation of a method combining the best estimation and the uncertainty analyzingis reduced, and the conservation margin of deterministic realistic method is decreased.

The invention discloses a wetland early warning method based on artificial intelligence and a random self-adaptive threshold, and the method builds a random self-adaptive parameter learning module through employing a loop iteration principle, and enables an early warning system to have a self-adaptive adjustment function. Numerous uncertainties faced by a wetland water system are considered, an uncertainty quantification theory (random) is applied to construction of a random adaptive threshold, concepts such as a normal threshold parameter probability density function (PDF) and a normal threshold parameter cumulative distribution density function (CDF) are used for quantifying random uncertainty, and an artificial intelligence algorithm is used for ensuring the reaction efficiency of real-time early warning. Residual analysis of the prediction data and the observation data is used as index data for monitoring the wetland ecological state, so that the early warning system is more sensitive; the introduction of Bayesian reduces the false alarm rate, and introduces the iterative early warning rate. Therefore, the accuracy of wetland health early warning is comprehensively improved.

The invention provides a CFD model confirmation method based on a multi-credibility deep neural network and application thereof. The uncertainty of key parameters is quantified by using the multi-credibility deep neural network, so that the calculation amount is reduced. According to the method, the uncertainty of the key parameters is quantified by using the multi-credibility deep neural network, so that the calculation amount is reduced, the CFD result is measured under multiple working conditions by using the model measurement method based on the distance method, the influence of the uncertainty on CFD output can be quickly evaluated, so that quick model confirmation is realized, and the technical bottleneck of large computational complexity of CFD uncertainty quantification at present is broken through. According to the method, a parameter checking strategy based on high-quality small samples is established, parameters with cognitive uncertainty in the CFD model are rapidly and effectively corrected, key parameters with cognitive uncertainty can be rapidly and effectively corrected, and therefore the CFD simulation model with high fidelity is constructed.

Systems and methods are provided for constructing and evaluating a story of interest. The system includes a plurality of decision algorithms. Each decision algorithm is operative to quantify at least one category of uncertainty associated with the story of interest as a set of at least one uncertainty parameter. An uncertainty management component is operative to reconcile the sets of uncertainty parameters from the plurality of decision algorithms as to produce a global uncertainty parameter for each of the plurality of uncertainty categories for the story of interest.

Embodiments are directed to identifying and implementing hydrocarbon production opportunities including recompletion opportunities, new vertical drill target opportunities, and horizontal or deviated well target opportunities. Computer systems access multi-disciplinary data, perform data validation and pre-processing on the data, identify field development opportunities by identifying candidates, forecasting production at those candidates, generating an uncertainty quantification, and vetting and validating the data. The computer systems then list the viable well target opportunities including recompletion opportunities, vertical new drill target opportunities and deviated and horizontal target opportunities.

The invention provides a time sequence Bayesian compression sampling and signal decompression reconstruction method and a data loss recovery method. The method comprises the following steps: selectinga measurement matrix of signal compression sampling; designing a basic vector matrix, modeling the change sparsity of a signal and an adjacent time period thereof, solving the Bayesian probability ofa decompressed reconstructed signal in each time period, quickly optimizing and estimating hyper-parameters, diagnosing the reconstruction precision based on posteriori uncertain quantization, recovering the signal loss in the health monitoring wireless sensor, and the like. According to the method, a hierarchical sparse Bayesian learning modeling and solving algorithm is adopted, and the methodhas unique advantages in the aspects of embedding two sparse features of a signal and the time-varying sparse features of the signal, rapid solving of hyper-parameters, signal reconstruction robustness and signal reconstruction uncertainty quantification under a high compression rate and the like, and has good noise robustness.

A method for statistical inference and confirmation of product performance change and life high confidence small samples includes such steps as (1) establishing a statistical distribution model of product performance change; (2) carrying out parameter estimation and uncertainty quantification of the product performance change distribution model; (3) establishing a product storage life distribution model; (4) carrying out statistical inference of quantitative changes of product performance and storage life. The invention can reduce the required number of sample groups to one group or two groups under various given confidence levels, and the required sample size data of each group can be reduced to five, so that the confidence level of the prescribed reliability evaluation result is not significantly lower than the prescribed confidence level.

The invention provides a parameter uncertainty analysis method and system for a nuclear power reactor system, and the method comprises the steps: building a system evaluation model according to a system design value through employing an optimal estimation program, and carrying out the steady-state debugging of the system evaluation model; carrying out sensitivity analysis on the input parameter / physical model by utilizing a system evaluation model to obtain a key input parameter / key physical model, and obtaining a target change range of the key input parameter / key physical model according to characteristics of the key input parameter / key physical model; constructing a target parameter sample point set according to a target change range of the key input parameter / key physical model, and performing uncertainty analysis on the target parameter sample point set by using a system evaluation model to obtain an uncertainty quantification result of the key input parameter / key physical model, according to the method, the excessive conservative allowance of the nuclear power reactor system is effectively released, and the economical efficiency of the nuclear power reactor system is improved to the maximum extent.

The invention relates to a field line coupling uncertainty quantification and global sensitivity calculation method. An adaptive hyperbolic truncation scheme is combined with a minimum angle regression method to carry out sparse processing on a generalized chaotic polynomial, an agent model is established to analyze the uncertainty problem of the radiation sensitivity of a multi-conductor transmission line, and the global sensitivity of an input variable is analyzed by a global sensitivity analysis method. According to the scheme, on the premise that the calculation precision is guaranteed, related statistical characteristic parameters and the like in the uncertainty problem of the radiation sensitivity of the multi-conductor transmission line can be rapidly and accurately calculated, andthe scheme is effective when the total sensitivity index and the first-order sensitivity index of the radiation sensitivity of the multi-conductor transmission line are calculated.

The invention discloses a high-confidence coefficient statistical inference and confirmation method for limited quantity product performance failure rates. The method comprises the following steps of:1, carrying out small sample sampling to obtain performance data; 2, estimating a limited quantity product performance parameter mean value and a standard deviation and quantifying uncertainties: (1)determining reliabilities (as shown in the specification) of binomial distributions, (2) determining relative uncertainties estimated by the reliabilities (as shown in the specification) of the binomial distributions, and (3) estimating performance parameter values and quantifying uncertainties of the performance parameter values; and 3, estimating a limited quantity product performance failure rate (number). The method is an effective method capable of reducing the product performance checking and testing number, correctly judging whether limited quantity product performance failure rates (numbers) achieve stipulated requirements or not and ensuring that the judging results have relatively high credibility in production, acceptance, storage and using of current limited quantity products.

The invention relates to a face recognition method based on an uncertainty quantization probability convolutional neural network. The method comprises a training stage and an identification stage. Theprobabilistic convolutional neural network is trained through samples with known categories. The extraction of face features is realized by a learning process of a probability convolutional neural network. The description of the face features is represented by the connection weight; testing the trained probability convolutional neural network by using a training sample, and determining a classification threshold. A to-be-identified sample is input into the probability convolutional neural network. An output vector of the probability convolutional neural network is calculated. A maximum component is compared with a classification threshold value to give an identification result. Uncertainty quantitative analysis is made on the identification result, and a mean value and variance estimationof the identification result are provided. The probability convolutional neural network system is adopted. The explicit feature extraction process is avoided through the difference extraction layer and the feature mapping layer. The feature which greatly contributes to the construction of the sample space is implicitly obtained from the sample. Compared with the prior art, the recognition rate and the anti-interference performance are higher.

A method of processing magnetic resonance data of a sample under investigation includes the steps of provision of the MR data being collected with an MRI scanner apparatus, and machine learning based data analysis of the MR data by supplying the MR data to an artificial neural network being trained with predetermined training data, wherein at least one image parameter of the sample and additionally at least one uncertainty quantification measure representing a prediction error of the at least one image parameter are provided by output elements of the neural network. Furthermore, a magnetic resonance imaging (MRI) scanner apparatus being adapted for employing the method of processing MR data is described.

The invention provides a test data based health state assessment method for a solid rocket engine. The method specifically comprises the steps of dividing health state grades of the solid rocket engine; determining test parameters and weights; performing unitary quantification processing on test data; correcting a unitary quantification result by utilizing a time correction function; determining a membership degree in a triangular fuzzy function; determining basic credibility; and synthesizing distribution credibility of different health state grades.

The invention discloses an uncertainty modeling and measuring method for remote sensing image features. The method is mainly composed of two parts of uncertainty modeling of remote sensing image features in a geographic space domain and uncertainty modeling of a feature space domain. Then, weighted combination is performed on the geographical space uncertainty and the feature space uncertainty toobtain a comprehensive feature uncertainty index FUI (Feature Unity Index) so as to measure the feature uncertainty of the remote sensing image more accurately and comprehensively. Different expression characteristics of the image characteristic uncertainty under different visual angles are considered, a characteristic uncertainty quantification result of the whole image pixel by pixel can be provided, and manual intervention can be minimized. The method is high in accuracy and adaptive degree, high in calculation efficiency, high in operability, easy to implement and high in expandability ofthe whole model. The uncertainty quantification result has very high indication capability for classification errors, so that the method has very high practical value.

The invention discloses a polynomial chaos-based turbine blade tip gas thermal performance uncertainty quantification system. The system comprises a parameterized geometric modeling module, a sampling point generation module, a polynomial chaos expansion model construction module, a sample point initial field distribution module, a turbine blade tip gas heat performance full-automatic pretreatment module, a multi-core remote asynchronous distributed calculation module, a calculation progress real-time feedback module, a polynomial chaos expansion coefficient solving module, a sensitivity analysis module, a result file sampling and analysis module and the like. According to the turbine blade tip gas thermal performance uncertainty detection system, the number of samples needed for obtaining the turbine blade tip gas thermal performance uncertainty can be greatly reduced, the time needed for calculating the numerical value of a single sample can be shortened, and the system is further suitable for turbine blade tips of any size and form; and the requirements of designers of the turbine cooling system are better met.

The invention discloses a logistic regression-based aircraft flutter analysis and QMU evaluation method, and relates to a quantification technology of uncertainties and margins. The method comprises the following steps of 1) establishing a database of existing flight tests, namely whether the flight test suffers from flutter or not under a certain group of designed parameters, recording flutter as 1 and recording non-flutter as 0; 2) establishing a regression prediction model; 3) fitting a regression coefficient; 4) predicting the model; and 5) performing QMU evaluation. By omitting a complex finite element analysis process, high calculating efficiency and convenient operation are realized, and the method is applicable to the field where it is difficult to perform numerical value simulation on flutter and the like of a hypersonic thermal structure; the analysis result is evaluated by the QMU, and uncertainties of the flutter speed and flutter border are taken into consideration at the same time, so that security criterion for a flutter problem is high in credibility; and flutter probability of an aircraft under a certain designed variable condition can be predicted, so that the method can be used for assisting aircraft design.

A method for probabilistic fatigue life prediction using nondestructive testing data considering uncertainties from nondestructive examination (NDE) data and fatigue model parameters. The method utilizes uncertainty quantification models for detection, sizing, fatigue model parameters and inputs. A probability of detection model is developed based on a log-linear model coupling an actual flaw size with a nondestructive examination (NDE) reported size. A distribution of the actual flaw size is derived for both NDE data without flaw indications and NDE data with flaw indications by using probabilistic modeling and Bayes theorem. A turbine rotor example with real world NDE inspection data is presented to demonstrate the overall methodology.

The invention provides a semantic segmentation network model uncertainty quantification method based on evidence reasoning. The method comprises the following steps: constructing an FCN network model, and training the FCN network model by using a training data set to obtain a trained FCN network model for semantic segmentation of image data; transplanting a D-S evidence theory to the trained FCN network model to obtain a reconstructed FCN network model; and inputting image data to be segmented into the reconstructed FCN network model, outputting a classification result of an image to be segmented by the FCN network model, and calculating an uncertain value of the classification result of each pixel point by using a D-S evidence theory index. Quantitative calculation of the semantic segmentation uncertainty can be effectively completed in a short time, the calculation efficiency is greatly improved, and time and resource cost are saved.