71 results about "Covariate" patented technology

In accordance with the principles of the present invention, an apparatus, simulation method, and system for modeling loss in a term structured financial portfolio are provided. An historical date range, time unit specification, maturity duration, evaluation horizon, random effects specification, and set of portfolio covariates are selected. Historical data is then segmented into infinitely many cumulative loss curves according to a selected covariate predictive of risk. The s-shaped curves are modeled according to a nonlinear kernel. Nonlinear kernel parameters are regressed against time units up to the maturity duration and against selected portfolio covariates. The final regression equations represent the central moment models necessary for prior distribution specification in the hierarchical Bayes model to follow. Once the hierarchical Bayes model is executed, the finite samples generated by a Metropolis-Hastings within Gibbs sampling routine enable the inference of net dollar loss estimation and corresponding variance. In turn, the posterior distributions enable the risk analysis corresponding to lifetime loss estimates for routine risk management, the valuation of derivative financial instruments, risk-based pricing for secondary markets or new debt obligations, optimal holdings, and regulatory capital requirements. Posterior distributions and analytical results are dynamically processed and shared with other computers in a global network configuration.

A method and system for accurately predicting the remaining useful life of devices and device components based on rigorous statistical analysis data to reduce service costs by implementing condition-based maintenance is disclosed. One such rigorous statistical model is the general degradation path model. The General Path Model is used to generate simulated data that shares similar data characteristics of historical field failure data. This generated data can be used in a reliability study based on Monte Carlo techniques for RUL prediction. The study can be used to investigate the effects of influential factors such as suspension percentage and heavy-tailed behavior. The remaining useful life prediction is based on both the fixed-time predictors (such as the market segment) and the time-dependent covariates (e.g., dark decay, printing rate, etc.). The Random Forest Model can also be used for accurately predicting remaining useful life based on both fixed-time predictors and time-dependent covariates, which are both contained in the field data of a component. A relationship between component failure and amount of degradation makes it possible to use degradation models for accurate predictions of failure time.

Dependent variables believed to contribute to a likelihood of falling are analyzed using a latent class analysis. The dependent variables are biomedical factors, which may include, for example, arthritis, high blood pressure, diabetes, foot disorders, Parkinson's Disease, stroke, eye disorder, limb disorder, or proprioceptive disorder. Data pertaining to the biomedical factors is gathered from a population of individuals at risk of falling. Covariate data, including for example age and the number of prescriptions taken, is further analyzed against latent class data. For a particular group of at risk individuals, a set of five classes produced useful results broadly corresponding to groups representing individuals who have: good health; a range of diseases; Parkinson's Disease; arthritis; and high blood pressure. A probability of falling is determined, relative to the group of individuals with good health.

A method of predicting a response relationship between elements of two sets includes: specifying a dyadic response matrix; specifying covariates that measure additional dyadic relationships; specifying a number of row clusters and a number of column clusters for clustering the rows and columns of the response matrix; specifying a rank for cluster factors that model average interactions between row clusters and column clusters by products of cluster factors; and determining prediction parameters for predicting responses between elements of the first set and the second set by improving a likelihood value that relates the prediction parameters to the response matrix, the covariates, the observation weights, the row clusters and the column clusters. Determining the prediction parameters includes: updating the prediction parameters for fixed assignments of row clusters and column clusters, and updating assignments for row clusters and column clusters for fixed prediction parameters.

The invention discloses a gait recognition method based on a depth neural network, and the method comprises the steps: carrying out the preprocessing of an original gait video image, extracting the gait Gaussian image of the original gait video image, and dividing a sample data set into a training set and a testing set at a proportion of 5: 1 according to a designed rule; building an eight-layer convolution neural network model consistent with an AlexNet structure, modifying the number of neurons of the last layer of the model, and enabling the model to adapt to a gait classification task; Carrying out the initialization of the parameters of a trained AlexNet model through the former seven layers of the model, and carrying out the random initialization of the last layer of the model; training the convolution neural network through the training set, and enabling the convolution neural network to effectively complete the gait recognition. The method is higher in robustness, can achieve the more effective recognition of the identity of a person under the condition of various types of concomitant variables, and can remarkably reduce the calculation resources and time of a training model.

The invention discloses a covariant initial value determination method for the optimal landing track design and belongs to the field of deep space exploration. The covariant initial value determination method includes building a small body fixing coordinate system and corresponding detector landing kinetic equations; converting the optimal design problem of the small body landing track into the optimal control problem and the corresponding two-point boundary value problem, and defining the problems as the problem 1; approximating the problem 1, and defining the approximated problem 1 as the problem 2; solving the covariant initial values lambda<r2(t0)> and lambda<v2(t0)> of the problem 2, setting the the covariant initial values lambda<r2(t0)> and lambda<v2(t0)> of the problem 2 as iterative initial values of the covariant initial values lambda<r1(t0)> and lambda<v1(t0)> of the problem 1, determining the iterative initial value of the covariant initial value lambda<m1(t0)> of the problem 1 according to the setting of the iterative initial values, namely lambda<r2(t0)> and lambda<v2(t0)>, of the covariant initial values lambda<r1(t0)> and lambda<v1(t0)> of the problem 1, thereby achieving the iterative initial value setting of the covariant initial values for the optima landing track design. The defect that it is difficult to solve the two-point boundary value problem due to improper initial value setting can be avoided by application of the method.

A failure rate calculation method for a relay protection device in consideration of covariates belongs to the technical field of protection of electric power systems. The method comprises the following steps of: determining the function relationship between relay protection failure rate distribution parameters and the covariates by using a multiplication model; obtaining failure rate distribution functions under the Weibull distribution condition by the function relationship between the parameters and the covariates; and obtaining unknown parameters of the distribution functions through the formula (referring to the Specification) in order to calculate the failure rate of the relay protection device, wherein m and Eta in the formula respectively represent Weibull distribution shape parameters and scale parameters. According to the failure rate calculation method for the relay protection device in consideration of the covariates disclosed by the invention, the failure characteristics of the relay protection device are accurately simulated in consideration of the intrinsic characteristics of the relay protection and the external working conditions thereof, so that the covariates and failure rate calculation method for the relay protection is built. And the method is helpful to find out the principal factors which influence the reliability indexes of the method, and provides the guidance to improve the relay protection reliability.

A method and apparatus to determine: (a) the likelihood and timing for a sales opportunity to become a sale based on analytical models that incorporate the history of sales stage evolution and other covariates; and (b) the expected number of sales from invisible opportunities prior to a target date. Additionally, the method and apparatus is configured to predict an expected amount of revenue and/or an amount of resources given a current sales history.

The invention discloses a method for predicting the service life of a key system of rail transit vehicles. The method comprises the following steps of firstly, determining the key system of urban rail vehicles, and bringing in a proportional hazard model to carry out service life prediction on the key system; secondly, screening historical overhauling data of the key system to obtain concomitant variable factors, and processing concomitant variables by applying a principal component analysis method so as to obtain a concomitant variable matrix; thirdly, solving a parameter estimation value in the proportional hazard model by adopting a maximum likelihood estimation method and a Newton iteration method; finally, calculating the predicted service life of the key system of the rail transit vehicles by utilizing the obtained proportional hazard model. According to the method disclosed by the invention, the influence of the concomitant variable factors on the service life of the key system of the rail transit vehicles is considered, and the defect that the service life prediction is inaccurate as only system failure is considered in traditional service life prediction is overcome.

The invention discloses an air pollution prediction method based on deep fusion of multi-source space-time big data. The method comprises the following steps: collecting and preprocessing multi-source big data; inverting the meteorological data to obtain high-resolution ground meteorological parameters; carrying out missing inversion and upscaling on aerosol parameters and NO2 remote sensing parameters; traffic variables, land utilization variables, social economy and POI variables and spatial-temporal variation variables are extracted; carrying out space-time fusion on covariable data of various kinds of space-time big data to form a data set with unified scale and space coordinates; inverting high-resolution earth surface parameters of the air pollution concentration; verifying and evaluating the precision; if the standard is reached, outputting a result; and if not, adjusting and circularly training until a reasonable model and prediction are obtained. According to the method, the space-time coverage is large, grid modeling of meteorological data and interpolation of satellite parameters are improved through an advanced optimization technology, high test precision and high generalization are achieved, estimation deviation is reduced through a result verification and circulating modeling mechanism, and the efficiency of practical application is improved.

The invention relates to a proportional hazard rate model method for estimating operation reliability of a tool. Defects of difficulty in obtaining burn-out life data, incomplete matching with actual operation conditions of equipment and incapability of timely prevention and maintenance exist in the prior art. According to the technical scheme provided by the invention, the proportional hazard rate model method for estimating operation reliability of the tool comprises the following operation steps of: 1, using an effective value (RMS (Root Mean Square)) and a peak value (P) in a characteristic signal as covariates in a proportional hazard rate model; 2, calculating model parameters according to a nonlinear equation described in the specification; and 3, obtaining parameter results of maximum likelihood estimation of the tool working in different regressive states by applying a maximum likelihood estimation method of the proportional hazard rate model, and solving a current hazard rate according to state signals of the processing system tool working in different regressive states and finally obtaining the reliability. The method has the advantages of good instantaneity and reliable estimation result.

Techniques for iterative feature extraction using domain knowledge are provided. In one aspect, a method for feature extraction is provided. The method includes the following steps. At least one query to predict at least one future value of a given value series based on a statistical model is received. At least two predictions of the future value are produced fulfilling at least the properties of 1) each being as probable as possible given the statistical model and 2) being mutually divert (in terms of numerical distance measure). A user is queried to select one of the predictions. The user may be queried for textual annotations for the predictions. The annotations may be used to identify additional covariates to create an extended set of covariates. The extended set of covariates may be used to improve the accuracy of the statistical model.

In an embodiment, a process to predict an outcome of a competition includes receiving time-stamped position information of participant(s), the time-stamped position information captured by a telemetry tracking system during the competition. The process includes calculating while the competition is ongoing a covariate parameter for each of one or more participants at a point in time, where each respective covariate parameter is derived from the time-stamped position information of a corresponding participant at the point in time. The process includes predicting the outcome of the competition, as of the point in time, based at least in part on (i) a difference between a calculated competitor strength of the first competitor the second competitor based on historical data associated with the competitors, and (ii) the calculated first covariate parameter(s).

An embodiment of the invention relates to a prediction method and a prediction device. The prediction method comprises the steps of: training a Cox proportional hazard model according to acquired historical technical indexes, and acquiring a rising model, a declining model and a covariate; determining a risk function according to the covariate; determining an accumulative risk function according to the risk function; and calculating a probability of future rise and a probability of future decline according to the accumulative risk function, current technical indexes composed of time of previous rise and time of previous decline, the rising model and the declining model, and conducting a transaction based on the probability of future rise and the probability of future decline. Therefore, the prediction method and the prediction device are higher in precision of prediction results, are conductive to helping investors to make the most reasonable decision, increase profits and reduce the risks.

A gas compressor rotation stall early warning method based on a time expansion convolutional network comprises the following steps: firstly, preprocessing dynamic pressure data of an aero-engine, and dividing a test data set and a training data set from experimental data; secondly, sequentially constructing a time convolutional network module, constructing a Resnet-v network module, constructing a time expansion convolutional network prediction model, and storing an optimal prediction model; finally, performing real-time prediction on the test data: firstly, adjusting the data dimension of the test set according to the input requirement of the time convolutional network prediction model; according to a time sequence, calculating a surge prediction probability of each sample through a time expansion convolutional network prediction model; and calculating the real-time surge probability of a pair of samples containing covariables and not containing covariables through a time expansion convolutional network prediction model, and observing the improvement effect of the covariables on the model prediction effect. The time domain statistical characteristics and the change trend are integrated, and the prediction precision is improved; and the active control performance of the engine can be improved, and certain universality is achieved.

Techniques for iterative feature extraction using domain knowledge are provided. In one aspect, a method for feature extraction is provided. The method includes the following steps. At least one query to predict at least one future value of a given value series based on a statistical model is received. At least two predictions of the future value are produced fulfilling at least the properties of 1) each being as probable as possible given the statistical model and 2) being mutually divert (in terms of numerical distance measure). A user is queried to select one of the predictions. The user may be queried for textual annotations for the predictions. The annotations may be used to identify additional covariates to create an extended set of covariates. The extended set of covariates may be used to improve the accuracy of the statistical model.

The invention provides a system for evaluating reliability of a bogie rolling bearing based on offline state vibration characteristics. The system comprises a jacking device, friction wheels, a sensor, a data collector and a computer, wherein the jacking device is used for jacking wheel pairs of a tested vehicle when the tested vehicle is in an offline state, so that the wheels are separated from rails; the friction wheels are in drive connection with the wheel pairs of the tested vehicle and are used for driving the wheel pairs of the tested vehicle to rotate; the sensor is arranged on a bogie of the tested vehicle and is used for determining vibration acceleration signals of an axial bearing and a traction motor bearing; the data collector is connected with the sensor and collects the collected vibration acceleration signals; the computer is connected with the data collector; the computer builds a proportion-fault rate model by statistical characteristic values, of the vibration acceleration signals as response covariate and reliability and calculates the fault rate and the reliability of the bearing through the proportion-fault rate model.

The invention provides a multi-output gradient lifting tree modeling method for survival risk analysis, which comprises the following steps: firstly, constructing an expression of survival data for establishing a survival prediction model of financial, insurance, medical, traffic or industrial target industries under a model algorithm framework of an optimal gradient lifting tree (XGBoost); then defining and calculating a loss function corresponding to the survival data; then, defining and calculating a first step degree and a second step degree corresponding to the loss function; and finally,inputting the calculated loss function value and the first-order gradient value and the second-order gradient value of the loss function into an XGBoos model algorithm framework at the same time, andperforming automatic training to generate a survival prediction model of the target industry. The modeling method provided by the invention can better represent the relationship between the model covariable and the risk prediction value. The prediction performance and the generalization capability of the model are improved. The prediction performance and the risk distinguishing degree are better,and the application scene is wide.

The invention provides a prediction error extreme value analysis method and device, a computer device and a readable storage medium. The method comprises: acquiring a wind power prediction errors, numerical weather forecast data and historical anemometry data, dividing the wind power prediction errors according to the corresponding relationship between the preset conditions and the wind power prediction errors, to obtain a plurality of blocks, obtaining the block maximum value in each block, selecting a concomitant variable according to the numerical weather forecast data, the historical anemometry data and the maximum value of each block, and obtaining the extreme value distribution of the wind power prediction errors through the concomitant variable and the maximum value of each block. By adopting the method, the extreme values of the day-ahead wind power prediction errors are analyzed, thus the accuracy of extreme value modeling is increased, and the scheduling availability of the day-ahead prediction result is improved.

The invention belongs to the technical field of computer aided design, and discloses a method for predicting a residual life quantile based on left truncate and right censoring data. By establishing aclassical residual life quantile prediction model, the functional relationship between a residual life quantile and a dynamic covariate is constructed by a COX model, and the estimation method of unknown parameters of the model is given by combining the characteristics of the left truncated and right censored data, so the estimation value of the residual life quantile based on dynamic covariate prediction is obtained. The method for predicting the residual life quantile under the left truncated and right censored data of the present invention is relatively simple in realization process, and the COX model is used to establish the functional relationship of the residual life quantile prediction containing the dynamic covariate, and the prediction accuracy of the residual life is improved atthe same time.