82 results about "Uncertainty estimation" patented technology

A method for uncertainty estimation for nonlinear inverse problems includes obtaining an inverse model of spatial distribution of a physical property of subsurface formations. A set of possible models of spatial distribution is obtained based on the measurements. A set of model parameters is obtained. The number of model parameters is reduced by covariance free compression transform. Upper and lower limits of a value of the physical property are mapped to orthogonalspace. A model polytope including a geometric region of feasible models is defined. At least one of random and geometric sampling of the model polytope is performed in a reduced-dimensional space to generate an equi-feasible ensemble of models. The reduced-dimensional space includes an approximated hypercube. Probable model samples are evaluated based on data misfits from among an equi-feasible model ensemble determined by forward numerical simulation. Final uncertainties are determined from the equivalent model ensemble and the final uncertainties are displayed in at least one map.

Methods and systems for performing active learning for defect classifiers are provided. One system includes one or more computer subsystems configured for performing active learning for training a defect classifier. The active learning includes applying an acquisition function to data points for the specimen. The acquisition function selects one or more of the data points based on uncertainty estimations associated with the data points. The active learning also includes acquiring labels for the selected one or more data points and generating a set of labeled data that includes the selected one or more data points and the acquired labels. The computer subsystem(s) are also configured for training the defect classifier using the set of labeled data. The defect classifier is configured for classifying defects detected on the specimen using the images generated by the imaging subsystem.

Equipment uptime is getting increasingly important across different industries which seek for new ways of increasing equipment availability. Detecting faults in the system by condition based maintenance (CBM) is not enough, because at the time of fault occurrence, the spare parts might not available or the needed resources (maintainers) are busy. Therefore, prediction failures and estimation of remaining useful life can be necessary. Moreover, not only predictions but also uncertainty in the predictions is critical for decision making. Example implementations described herein are directed to tuning parameters of deep learning network architecture by developing a mechanism to optimize for accuracy and uncertainty simultaneously, thereby achieving better asset availability, maintenance planning and decision making.

A condition assessment and end-of-life prediction system that includes a virtual condition assessment instrument and a virtual end-of-life prediction instrument. The virtual condition assessment instrument measures the condition of the equipment and includes a data capture subsystem for sampling a set of analog signals and converting them into digital signals, a model-based component to estimate disturbances and predict an expected response, a signal-based component to process output from the model-based component, a classification component to process output from the signal-based component, a fuzzy logic expert component to combine information from the classification component and the model-based component in order to assess the condition of the equipment, and a condition assessment panel to display the condition of the equipment. The a virtual end-of-life prediction instrument predicts the equipment end-of-life and includes a condition prediction end-of-life prediction component to analyze information from the virtual condition assessment instrument to predict condition and end-of life, a prediction condition and end-of-life uncertainty estimation component to estimate the uncertainty of the condition and end-of-life prediction, and an end-of-life panel for displaying the condition and end-of-life prediction and uncertainty.

The invention provides a lithium ion battery remaining life direct prediction method based on probability integration, relates to the technical field of lithium ion battery remaining life prediction and aims at solving the problem that a traditional MONESN method is unstable and lack of remaining life uncertainty expression. The method comprises the steps of firstly measuring the maximum capacity of a lithium ion battery in easy circulating period; adopting N MONESN models to predict the lithium ion battery remaining life and obtain N prediction results; performing uncertainty estimation and integration on the prediction results so as to obtain a lithium ion battery remaining life prediction result based on probability integration. The lithium ion battery remaining life direct prediction method fully plays the strong non-linear prediction capacity of the MONESN models and effectively solves the problem of instability of a traditional MONESN algorithm. In addition, uncertainty expression and management are achieved. The lithium ion battery remaining life direct prediction method is suitable for lithium ion battery remaining life prediction under the condition that the capacity can be directly measured and obtained.

In an exemplary embodiment, a method is disclosed for developing an N-phasic finite element model for performing fully coupled analyses of multi-phase compositional fluid flow and heat flow in nonlinearly deforming porous solid media with time-dependent failure. The method can include formulating a finite element model of the behavior of a coupled solid-fluid formation, having complex geometry and behavior, and applying the model to derive the response of the formation in the form of one or more displacement fields for the solid phases and one or more pressure fields for the fluid phases in a zone of interest in a formation. In an exemplary embodiment, a system is disclosed for estimating the uncertainties in the derived displacement and pressure field solutions for the response of the fully coupled solid-fluid phases.

The invention provides a reinforcement learning deep searching method based on bootstrap DAQN (deep Q network), mainly comprising: bootstrap DQN, deep searching and environmental background, wherein the bootstrap DQN includes a bootstrap sample and a bootstrap DQN unit; the deep searching includes deep search test and bootstrap DQN drive deep searching; the environmental background includes generating online bootstrap DQN and bootstrap DQN drive. The bootstrap DQN is a practical reinforcement learning algorithm combining deep learning and deep searching, it is proved that bootstrapping may create effective uncertain estimation on a deep neural network and may also be extended to a large-scale parallel system, information is ranked in multiple time steps, and sample diversity is guaranteed; the bootstrap DQN acts as an effective reinforcement learning algorithm in a complex environment to process mass data in parallel, calculation cost is low, learning efficiency is high, and the method has excellent performance.

The invention discloses a comprehensive analysis method for spatial and temporal distribution of environment variables, and the method comprises the steps: calculating a test variation function valueat each spatial and temporal lag distance, and carrying out the fitting of a theoretical variation function model; combining the spatial and temporal sampling point data, carrying out the spatial andtemporal Kriging interpolation, and estimating the geographic attribute value of an unmeasured spatial and temporal position; building a quantitative relation between environment variable values and aregional position, forming a trend of the spatial and temporal distribution of environment variable values, and obtaining a prediction result based on the spatial and temporal Kriging; proposing various types of spatial and temporal uncertainty estimation methods; and finally estimating the spatial and temporal distribution of environment variables visually. The method gives full consideration tothe spatial and temporal structural property and continuity of the environment variables, achieves the full simulation and analysis of the spatial and temporal distribution characteristics of the environment variables from many aspects: modeling, prediction, trend analysis, uncertainty analysis and the space and time, and provides a basis for the spatial and temporal decision making and auxiliaryanalysis for regional environment estimation and related departments.

A radiation analysis system / method that automatically optimizes the efficiency calibration of a counting system based on benchmark data and variable parameters associated with radiation source / sensor / environment (RSSE) combinations is disclosed. The system / method bifurcates RSSE context (SSEC) model parameters into WELL-KNOWN (fixed) parameters (WNP) and NOT-WELL-KNOWN (variable) parameters (NWP). The NWP have associated lower / upper limit values (LULV) and a shape distribution (LUSD) describing NWP characteristics. SSEC models are evaluated using randomized statistical NWP variations or by using smart routines that perform a focused search within the LULV / LUSD to generate model calibration values (MCV) and calibration uncertainty values (UCV) describing the overall SSEC efficiencies. Sensor measurements using the MCV / UCV generate a measurement value and uncertainty estimation value. An exemplary embodiment optimizes geometry models of radiation sources by benchmarking with respect to measurement data from spectroscopy detectors and / or dose rate detectors.

The invention relates to an unmanned-ship speed and uncertainty estimation system and a design method. According to the system, an echo state network can be applied to speed estimation of an unmannedship, the echo state network is utilized to approximate model uncertainty and environment disturbance to enable the system to obtain target speed observation values, and also approximate unknown dynamics generated by the uncertainty of model parameters, non-modeling of fluid dynamics, external interference caused by wind waves and ocean currents and the like, and the state observation problem containing the model uncertainty and the unknown environment disturbance is effectively solved. Introduction of the echo state network overcomes the problems of slow convergence, proneness of falling intolocal minimums, complicated training processes and the like brought by traditional neural networks based on a learning algorithm of gradient descending. According to the system, the neural network with a low-frequency learning link is adopted to approximate system uncertainty, high-frequency oscillation which may be caused by a high-gain learning rate is effectively filtered out, and steady stateobservation on a system with unknown dynamics is realized.

The invention provides a double-flow vehicle-mounted pedestrian and vehicle prediction method based on boundary frame and distance prediction. The method comprises the main contents: pedestrian trajectory prediction, Bayesian modeling, recurrent neural network (RNN) encoder-decoder, distance prediction, and training and reasoning. The process comprises that a distance prediction flow is used for predicting a most possible vehicle distance sequence, and the boundary frame flow is composed of a Bayesian RNN encoder-decoder structure and is used for predicting the attitude distribution on a pedestrian trajectory and capturing cognition and arbitrary uncertainty; since the prediction flow of the distance prediction method is used for estimating a prediction point, the prediction flow is trained by minimizing the mean square error of a training set; and the Bayesian boundary frame prediction flow is trained by estimating and minimizing the KL divergence approximate to weight distribution. The double-flow system structure including the pedestrian boundary frame prediction and the vehicle distance prediction is adopted, the time required for prediction is greatly shortened, and the prediction accuracy of the model is also significantly improved by the uncertainty estimation.

The invention discloses a method for estimating uncertainty of indexes of integrity of a sedimentary cover of a carbon dioxide geological sequestration site. The method comprises the following steps of determining influence indexes of the integrity of the sedimentary cover; determining a failure mode; performing calculation to obtain pore pressure in the failure mode; performing five-level tornado analysis on the influence indexes of the integrity of the sedimentary cover; deleting divisors with influence factors of zero to obtain first-order removal indexes; determining a quadratic polynomial regression equation according to the first-order removal indexes; and optimizing the quadratic polynomial regression equation and the first-order removal indexes according to a linear coefficient of the quadratic polynomial regression equation and sensitiveness of a quadratic term coefficient to obtain the optimal carbon dioxide geological sequestration condition of the site. According to characteristics of the structure of the sedimentary cover, influence indexes of the integrity of the sedimentary cover without faults and influence indexes of the integrity of the sedimentary cover with faults are respectively selected. Key influence indexes can be extracted effectively. An optical designing scheme of the key influence indexes of the integrity of the sedimentary cover of the carbon dioxide geological sequestration site is obtained.

A mobile robot configured to move on a ground. The mobile robot including a contact angle estimation unit estimating contact angles between wheels of the mobile robot and the ground and uncertainties associated with the contact angles, a traction force estimation unit estimating traction forces applied to the wheels and traction force uncertainties, a normal force estimation unit estimating normal forces applied to the wheels and normal force uncertainties, a friction coefficient estimation unit estimating friction coefficients between the wheels and the ground, a friction coefficient uncertainty estimation unit estimating friction coefficient uncertainties, and a controller determining the maximum friction coefficient from among the friction coefficients such that the maximum friction coefficient has an uncertainty less than a threshold and at a point of time when the torque applied to each of the wheels changes from an increasing state to a decreasing state, among the estimated friction coefficients.

The invention belongs to the technical field of man-machine conversation, particularly relates to a man-machine hybrid response method, system and device, and aims to solve the problem that online learning cannot be realized by an existing man-machine response method. The method comprises the following steps of: encoding a current conversation context Ct to obtain a first representation vector E (Ct); Based on a candidate reply statement under the conversation task, and carrying out encoding to obtain a second representation vector E (Ct) , based on the first representation vector E (Ct), wherein the second representation vector E (Ct) is a positive integer greater than or equal to 1. a second representation vector (shown in the specification) obtains the confidence coefficient of the candidate reply statement capable of correctly replying to the question of the user through an uncertainty estimation method, and the confidence coefficient of the candidate reply statement capable of correctly replying to the question of the user is obtained through the uncertainty estimation method; and if the confidence coefficient is greater than a set threshold value, selecting a candidate replystatement corresponding to the confidence coefficient to carry out response output, otherwise, obtaining the reply statement input by the human-computer interaction equipment or the selected candidatereply statement to carry out response output, and carrying out parameter optimization in the above steps based on all dialogue statements obtained after response output. According to the method, it is guaranteed that the output response statement has enough confidence, and online learning and updating of the dialogue model are achieved.

An LSI design method according to the present invention is to estimate a timing uncertainty in an early stage of design for each item of which an influence on timing is uncertain among respective items requiring consideration relating to establishment of timing; and define a timing margin in each design stage by using the timing uncertainty estimation result depending on whether or not an influence of the each item on timing has been determined, followed by proceeding with the design in the respective design stages accordingly. As such, according to the present invention, a timing uncertainty is estimated in an early stage of LSI design, followed by proceeding with the design by using the timing uncertainty as required.

A method of determining an uncertainty estimate of an estimated velocity of an object includes, determining the uncertainty with respect to a first estimated coefficient and a second estimated coefficient of the velocity profile equation of the object. The first estimated coefficient being assigned to a first spatial dimension of the estimated velocity and the second estimated coefficient being assigned to a second spatial dimension of the estimated velocity. The velocity profile equation represents the estimated velocity in dependence of the first estimated coefficient and the second estimated coefficient. The method also includes determining the uncertainty with respect to an angular velocity of the object, a first coordinate of the object in the second spatial dimension, and a second coordinate of the object in the first spatial dimension.

A system and method for subsurface characterization including depth and structural uncertainty estimation is disclosed. In one embodiment, the method may include determining a detectability threshold for moveout in a seismic data gather based on the seismic data and computing a depth uncertainty function, wherein the depth uncertainty function represents an error estimate that is used to analyze an interpretation of the seismic data. In another embodiment, the method may include receiving a depth uncertainty volume and at least one interpreted horizon from seismic data, extracting a depth uncertainty cage for each of the interpreted horizons based on the depth uncertainty volume, and simulating multiple realizations for each of the interpreted horizons, constrained by the depth uncertainty cage. The multiple realizations may be used for analyzing changes to geometrical or structural properties of the at least one interpreted horizon. The changes may be plotted as at least one distribution and may be used to make P10, P50 and P90 estimates.

The invention discloses an electric energy substitution scheme prediction method and device based on a Gaussian regression combination prediction model, and the method comprises the steps: receiving prediction targets of different energy substitution schemes of an energy substitution conversion enterprise, and obtaining frequent item sets of various types of targets; building a Gaussian regressioncombined prediction model by adopting a Gaussian regression process; carrying out clustering prediction on the frequent item sets of various targets of different energy substitution schemes, and carrying out linear combination on prediction target values to obtain an annual cost value prediction target value; based on the principle that the annual cost values are equal, obtaining the boundary electricity price of the electric energy substitution scheme, calculating uncertainty estimation of the electric energy substitution scheme, and obtaining a prediction result of the electric energy substitution scheme; and performing data feedback based on the prediction result, comparing the data with actual data received by a related business application platform of the power system, adjusting parameters of the Gaussian regression combination prediction model, and performing electric energy replacement scheme prediction by adopting the Gaussian regression combination prediction model after parameter adjustment.

The invention discloses a rolling bearing friction moment parameter uncertainty estimation method. The method comprises the following steps of: firstly measuring a rolling bearing friction moment to obtain a rolling bearing friction moment time sequence, expressing the rolling bearing friction moment time sequence by a data sequence, assuming the data sequence as a normal distribution with unknown mean values and known variances, and calculating the mean values and variances of data in the data sequence; carrying out robust processing on the data sequence to obtain prior experience functions, and calculating the mean values and variances of the prior experience functions; and finally calculating a rolling bearing friction moment Bayes estimation interval under a certain confidence degree. According to the method disclosed by the invention, two optimum estimations are organically fused and complement each other's advantages; a data robust processing method which is not only capable of reflecting data position features, but also capable of reflecting total data attitude, and test data processed by the method is used as prior information of Bayes estimation; and a Bayes method is used for estimating the performance of mechanical products.

The invention relates to an image uncertainty prediction method and device, equipment and a storage medium. The image uncertainty prediction method comprises the steps: obtaining a training image set,marking each image in the training image set, and obtaining a marked image set; initializing a preset deep learning model, wherein the preset deep learning model comprises a distributed sampling network and an image segmentation network; training the preset deep learning model based on the training image set and the annotation image set to obtain a first prediction model; extracting a second prediction model from the first prediction model; and obtaining a target image, and performing image uncertainty prediction on the target image through the second prediction model. The invention providesan uncertainty prediction method, capable of obtaining diversified Monte Carlo samples through one-time forward calculation, and enabling uncertainty estimation to be more accurate while reducing thecalculated amount.

The invention relates to a vehicle trajectory prediction method based on uncertainty estimation. The method comprises the steps of collecting the pose information and local semantic map information of surrounding vehicles in real time, and obtaining the historical pose information of the vehicles, according to the collected vehicle position information, in combination with a high-precision map, a lane connection relationship and a traffic rule, determining all candidate lanes of a future trajectory end point, evaluating the uncertainty of the historical pose of the vehicle according to the pose of the vehicle and the local semantic map, converting the historical pose of the vehicle to the coordinate system of each lane, conducting feature coding in combination with information such as lane directions, and predicting the probability of a vehicle driving end point on each candidate lane, and predicting probability distribution of a future driving route of the target vehicle according to the feature coding. Compared with the prior art, the method solves the problems that in the prior art, input vehicle historical pose uncertainty is neglected, and trajectory multi-mode modeling is incomplete, an accurate and reliable information source can be provided for downstream decision planning of automatic driving, and risks are reduced.

A purification method for preparing a methamphetamine standard substance used for forensic scientific drug detection is disclosed. The method includes (1) detecting methamphetamine purities of seized methamphetamine samples, and selecting the methamphetamine sample the methamphetamine purity of which is not less than 70 wt% as a raw material for preparing the methamphetamine standard substance; and (2) preparing the methamphetamine standard substance by utilizing high performance liquid chromatography. The obtained methamphetamine has high purity and can be directly used as a standard substance. The methamphetamine obtained by the method is validated by nuclear magnetic resonance, liquid chromatography-tandem mass spectrometry and infrared spectroscopy analysis, the purity value of the methamphetamine obtained by the method is determined by liquid chromatography and gas chromatography, impurities not responding to chromatography are measured, and according to standard substance development requirements, stability, uniformity, the fixed value and total uncertainty estimation all meet related regulations, thus meeting excepted indexes.

The invention relates to a hybrid simulation method based lithium battery mechanical strength probability model. The model is established through the following steps: (1) a step of establishment of alithium battery mechanical strength boundary condition model, namely a step of determining boundary conditions of mechanical strength response based on materials of a lithium battery; (2) a step of establishment of a lithium battery finite element model, namely a step of establishing the lithium battery finite element model according to the mechanical strength boundary condition model and a displacement-mechanical strength relation curve; and (3) a step of establishment of a lithium battery mechanical strength estimation hybrid simulation model, namely a step of establishing the hybrid simulation model by introducing an artificial neural network (ANN) and an uncertainty estimation theory (MUET). the beneficial effects of the lithium battery mechanical strength probability model are as follows: the model combines the artificial neural network and the matrix based uncertainty estimation theory and evaluates the internal mechanical strength of the lithium battery by using the lithium battery finite element model, and the probability model and research results help lithium battery manufacturers to improve the application safety of electric vehicles and traffic road safety.

The invention relates to a wind power fluctuation uncertainty estimation method, in particular to a wind electricity uncertainty estimation method based on a wind power fluctuation strength instant model. The problems that a wind power fluctuation instant model adopted in a traditional method is weak in universality and a wind power real-time prediction result is not accurate are solved. The method comprises the steps of obtaining actually-measured wind power data, and using the Mallat wavelet decomposition and reconstruction algorithm as a tool for conducting wavelet decomposition on the actually-measured wind power data; utilizing the Mallat wavelet decomposition algorithm for conducting decomposition and reconstruction on a wind power fluctuation residual error to obtain an instant standard deviation sigma m of a wind power minute-level fluctuation residual error corresponding to the same-period hourly average wind power and an instant standard deviation sigma s of a wind power second-level fluctuation residual error, and obtaining the minute-level wind power fluctuation strength and the second-level wind power fluctuation strength corresponding to the hourly average wind power; conducting fitting on the wind power fluctuation strength modeling; obtaining the final wind power fluctuation strength instant model, and conducting quantitative estimation on the uncertainty of a predication result. The wind electricity uncertainty estimation method is applicable to power grid operation and scheduling.

A system and method for subsurface characterization including depth and structural uncertainty estimation is disclosed. In one embodiment, the method may include determining a detectability threshold for moveout in a seismic data gather based on the seismic data and computing a depth uncertainty function, wherein the depth uncertainty function represents an error estimate that is used to analyze an interpretation of the seismic data. In another embodiment, the method may include receiving a depth uncertainty volume and at least one interpreted horizon from seismic data, extracting a depth uncertainty cage for each of the interpreted horizons based on the depth uncertainty volume, and simulating multiple realizations for each of the interpreted horizons, constrained by the depth uncertainty cage. The multiple realizations may be used for analyzing changes to geometrical or structural properties of the at least one interpreted horizon. The changes may be plotted as at least one distribution and may be used to make P10, P50 and P90 estimates.

The invention provides a field-correlation single-side self-calibration light beam adjustment method for photogrammetry. The method comprises the following steps: a, establishing a linear pinhole imaging model, and auxiliary arranging a field-correlation nonlinear distortion model; b, determining a field-correlation single-side self-calibration light beam adjustment theoretical model; c, establishing an error equation of a field-correlation single-side self-calibration imaging model; d, solving a partial derivative of the field-correlation imaging model in the error equation for an exterior orientation parameter; e, solving a partial derivative of the field-correlation imaging model in the error equation for space coordinates; f, solving a partial derivative of the field-correlation imaging model in the error equation for a single-side interior parameter; g, performing adaptive proportion adjustment on each item in an error equation Jacobian matrix; h, rapidly calculating each item in a normal equation through a portioning mode; i, solving various parameters in self-side self-calibration; j, performing parameter proportion inverse adjustment to eliminating proportion change; and k, calculating weight errors, and carrying out uncertainty estimation of the various parameters in the single-side self-calibration.

The invention relates to a preparation method of a standard sample for inorganic components in urban sludge; the method comprises the steps of acquiring a sample; measuring preliminarily; soaking; mixing well; air-drying; drying by baking; mixing well for the second time, to be specific, mixing well again by means of pile shifting method; grinding, and screening; mixing well for the third time; primarily inspecting uniformity; subjecting the sample passing the primary inspection of uniformity to uniformity inspection, stability inspection, and fixed value and uncertainty estimation again so as to perform fixed value calibration on the sludge sample. According to the standard sample for inorganic matters in urban sludge prepared herein, results of standard recovery for sludge samples A and B are better than the results of both a soil standard sample and a precipitate standard sample. Therefore, the standard sample of sludge prepared by using the method is more suitable for the quality control in monitoring inorganic matters in sludge from urban sewage treatment plants and urban disposed sludge; by preparing the special standard sample of sludge, measurement results for sludge quality control process are more accurate, a sludge treatment process can be accurately guided, and late harmless treatment of sludge is facilitated.