88 results about "Temporal models" patented technology

Systems and techniques are provided for blocked sensor detection and notification. A signal may be received from a sensor. The sensor may be determined to be producing anomalous output based on checking the signal from the sensor against a signal from a second sensor on the same sensor device as the sensor, checking the signal from the sensor against signals from sensors on additional sensor devices, and checking the signal from the sensor against a slow temporal model for the sensor. In response to the determination that the sensor is producing anomalous output, a notification may be generated indicating the sensor is producing anomalous output, and a confidence level in the sensor may be degraded.

A system and method for dialog state tracking employ an ontology in which a set of values are indexed by slot. A segment of a dialog is processed to detect mentions. Candidate slot values are extracted from the ontology, based on the detected mentions. The candidate slot values are ranked. A dialog state is updated, based on the ranking of the candidate slot values, which may be conditioned on the output of a temporal model, which predicts whether the value of the slot has been instantiated, modified, or is unchanged.

An optical flow estimation process based on a spatio-temporal model with varying coefficients multiplying a set of basis functions at each pixel. The benefit of over-parameterization becomes evident in the smoothness term, which instead of directly penalizing for changes in the optic flow, accumulates a cost of deviating from the assumed optic flow model. The optical flow field is represented by a general space-time model comprising a selected set of basis functions. The optical flow parameters are computed at each pixel in terms of coefficients of the basis functions. The model is thus highly over-parameterized, and regularization is applied at the level of the coefficients, rather than the model itself. As a result, the actual optical flow in the group of images is represented more accurately than in methods that are known in the art.

The invention provides an SPH (smoothed particle hydrodynamics) algorithm-based simulation method and an SPH (smoothed particle hydrodynamics) algorithm-based simulation system of the process of breaking dam by flood. The simulation method comprises the following steps: A. acquiring on-site geographic spatial information data; B. establishing a dynamical model of the process of breaking dam by flood based on the geographic spatial information data obtained in the step A; C. establishing a geographic entity model according to the geographic spatial information data obtained in the step A; D. analyzing the dynamical model obtained in the step B into an SPH calculation method; E. initializing the geographic entity model obtained in the step C to be hydrodynamics particles and boundary particles for the SPH calculation; F. circularly calculating based on the SPH algorithm; G. carrying out the spatial-temporal process modeling on the calculated value result obtained in the step F to obtain a 3D spatial-temporal model and a database of the process of breaking the dam by flood; and H. dynamically visualizing the 3D spatial-temporal model and the database of the process of breaking dam by flood obtained in the step G. By using the SPH method in the geographic process simulation, the authenticity of the simulation result is effectively improved.

The invention discloses a regional distributed photovoltaic output prediction method based on a satellite cloud picture, and the method comprises the steps: 1, carrying out the dividing of a regional distributed photovoltaic system according to a grid; 2, carrying out the data recording; 3, carrying out the processing of historical data; 4, carrying out the correction of the power data of a power clear sky model; 5, carrying out the calculation of a cloud index; 6, carrying out the normalization processing of different time lag scale data of output power of each grid, corrected clear sky model power, sun spatial-temporal model data, air quality data, weather forecast information and satellite cloud picture cloud index operator data, and building an Elman neural network model through a random training method; 7, inputting the latest calculated input variable parameters into the above models, and enabling the prediction data of each grid to be superposed to serve as the regional photovoltaic output prediction data. The method is higher in prediction precision, and is higher in efficiency.

Technique for early detection of localized exposure to an agent active on a biological population include collecting time series for each data type of multiple different data types. The data types are relevant for detecting exposure to the agent. For each data type multiple time series are collected for corresponding multiple locations associated with the data type. Measures of anomalous conditions are generated at the locations for each of the different data types. The measures of anomalous conditions are based on the time series and a temporal model for each data type. Cluster analysis is performed on the measures of anomalous conditions to determine an estimated location, and an estimated extent, of effects from the agent. The techniques allow a surveillance system to avoid diluting the signal of a localized outbreak over too large and area or consuming excessive resources in computing replicas for a matched filter detector.

The invention discloses an object scale self-adaption tracking method based on a spatial-temporal model, and the method comprises the following steps: beginning a video, reading in a first frame of image, manually assigning the rectangle position of a tracked object; then based on context airspace, initializing the spatial-temporal model and a multiscale history object template library; next reading in a next frame of image, building the spatial-temporal model by iteration, calculating a confidence map, and estimating an object center position; then according to the history object template library, judging a template optimal scale, determining the rectangle position of the object, completing tracking of the current frame of object, and updating scale parameters of the spatial-temporal model and the multiscale history object template library; finally detecting whether the video is finished or not, continuously reading in a next frame if the video is not finished, or completing tracking. By adopting the object scale self-adaption tracking method, the object appearance scale change is effectively coped with under the conditions of illumination change, partial blocking and swift moving, and the robust tracking is realized.

A system and method for processing acquired images to develop useful classifications of subjects such as occupants of a vehicle preferably employs a hierarchical and probabilistic structure, such as a Bayesian Network to analyze acquired images and produce a meaningful classification. The structure preferably includes set of analyzers, a set of Scenario analyzers and a set of Temporal models which are arranged in three respective hierarchical layers. Each respective analyzer operates on the acquired image and, in some circumstances, feedback from the Scenario analyzers, to produce an output representing the probability that a feature that the respective analyzer is concerned with is present in the acquired image. Each respective Scenario analyzer receives output probabilities from at least one of the analyzers and, in some circumstances, feedback from the Temporal Models, to produce an output indicating the probability that a scenario that the respective Scenario analyzer is concerned with, is the scenario captured in the acquired image. Each respective Scenario analyzer can also provide feedback inputs to one or more analyzers to alter their operation. Finally, each respective Temporal Model receives and operates on the output from at least one Scenario analyzer to produce a probability that a classification with which the Temporal Model is concerned is represented by the acquired image. Each respective Temporal Model can also provide feedback inputs to one or more Scenario analyzers to alter their operation. The structure processes the classification probabilities output from the Temporal Models to produce a classification for the acquired image.

The invention provides an information extraction method and system based on a joint training model, and belongs to the technical field of natural language processing and deep learning, in order to solve the problems that in an existing information extraction technology, a large amount of manpower and time are consumed, and the flexibility of a model is not high, and error transmission is caused, and information extraction is incomplete. The information extraction method comprises the steps: labeling corpora, and obtaining training corpora containing labeling information; sampling the trainingcorpus; converting each character in the sampled corpus into a word vector; inputting the word vector into two deep learning models based on different neural networks for joint training, and iteratively updating neural network parameters of a joint model to obtain a trained information extraction joint model; and inputting a to-be-extracted text into the information extraction joint model, and extracting triple information containing a head entity, a tail entity and an entity relationship.

The invention discloses a smart power grid line loss detection method and system. The smart power grid line loss detection method includes the steps of obtaining monitoring data of a smart power grid, classifying monitoring values according to serial numbers of monitoring devices to obtain a plurality of monitoring sets, extracting the monitoring values within corresponding set time periods in the monitoring sets through a time coordination mechanism according to sampling time of the monitoring values, computing and outputting line loss values corresponding to the monitoring sets according to the monitoring values within the corresponding set time periods in the monitoring sets and a preset spatial-temporal model, enabling the lowest common multiples of sampling periods corresponding to all the monitoring values in the corresponding monitoring sets to serve as duration of the set time periods to extract the monitoring values, and synchronously computing the line losses through the extracted monitoring values within the same set time periods to guarantee the computing accuracy. The monitoring data of the smart power grid are obtained, coherent processing is carried out on the monitoring data, then the line losses are computed, the practical line loss condition can be truly reflected, and compared with a traditional smart power grid line loss detection method, the computing accuracy is improved, and the detection error is reduced.

The present disclosure relates to methods and devices for building spatio-temporal neural networks and making predictions using them. According to an exemplary embodiment of the present disclosure, the method for constructing a spatio-temporal neural network may include: clustering a spatio-temporal sequence data set to divide the spatio-temporal sequence data set into a plurality of sub-regions in a spatial region; determining the the spatio-temporal correlation of each sub-area in the plurality of sub-areas; and constructing the spatio-temporal neural network based on the spatio-temporal correlation of each sub-area. According to the present disclosure, a heterogeneous spatiotemporal neural network is constructed by considering the heterogeneity of spatiotemporal sequence data. Using this method, the spatiotemporal model can be constructed more reasonably, and can improve the prediction accuracy of spatiotemporal sequence data while improving the fitting performance of training data.

A method includes obtaining (402) operating data associated with operation of a cross-directional industrial process controlled by at least one model-based process controller (106, 204). The method also includes, during a training period (502a, 502b), performing (406) closed-loop model identification with a first portion of the operating data to identify multiple sets of first spatial and temporalmodels. The method further includes identifying (408) clusters (604) associated with parameter values of the first spatial and temporal models. The method also includes, during a testing period (504a, 504b), performing (410) closed-loop model identification with a second portion of the operating data to identify second spatial and temporal models. The method further includes determining (412) whether at least one parameter value of at least one of the second spatial and temporal models falls outside at least one of the clusters. In addition, the method includes, in response to such a determination (414), detecting that a mismatch exists between actual and modeled behaviors of the industrial process.

Gaps in proficiencies may be identified within an enterprise. Understanding gaps in the existing workforce may help inform training, hiring, and firing decisions to ensure successful completion of the upcoming projects and deadlines. Using a multi-level model for each proficiency that accounts for enterprise needs as well as hiring, retraining, and the like, a relationship between proficiencies, projects, and employees over time may be generated as a multi-dimensional temporal model. The temporal model may be simulated to forecast gaps in proficiencies of the employed workforce. Recommendations regarding retraining, hiring, and termination can be made to help users remedy the deficiencies. Additionally, the proficiencies most valuable to the enterprise may be determined using a catalog of proficiencies to cluster the proficiencies into proficiency clusters for each job or job category and the proficiencies scored. Employees and candidates may be scored using the clusters to inform hiring, firing, and retraining decisions.

A computerized spatial-temporal regulation method for accurate spatial-temporal model estimation receives a spatial temporal sequence containing object confidence mask. A spatial-temporal weight regulation is performed to generate weight sequence output. A weighted model estimation is performed using the spatial temporal sequence and the weight sequence to generate at least one model parameter output. An iterative weight update is performed to generate weight sequence output. A weighted model estimation is performed to generate estimation result output. A stopping criteria is checked and the next iteration iterative weight update and weighted model estimation is performed until the stopping criteria is met. A model estimation is performed to generate model parameter output. An outlier data identification is performed to generate outlier data output. A spatial-temporal data integrity check is performed and the outlier data is disqualified.

The invention provides a configuration scheme and a task scheduling method for multiple elevators of a three-dimensional warehouse , relates to the technical field of logistics and the field of industrial automation, and aims to solve the problem that no scheduling optimization method suitable for the multiple elevators of the three-dimensional warehouse exists in the technical field of logisticsat present. The method comprises the steps that based on a time model of a four-way shuttle vehicle and elevators in a three-dimensional library established under an XYZ three-axis model, tasks of themultiple elevators are reasonably distributed and dispatched and optimized, and the model is solved and verified through a cyclic completion method. The reliability of the method is verified based ona simulation model of MATLAB, and a configuration optimization and scheduling method for multiple elevators in the current industry is supplemented.

The invention discloses a method, a system, a computer device and a storage medium for establishing a composition model. The method comprises the following steps: obtaining a music data set of a MIDIformat and converting the music data set of the MIDI format into a piano key axis; and converting the music data set of the MIDI format into a piano key axis; cleaning up the data of piano key axis after format conversion; establishing the interference track model and the composing track model by using the generated countermeasure network, and establishing the mixed track model by combining the interference track model and the composing track model; dividing the generator into a time structure generator Gtemp and a music bar generator Gbar, and constructing the temporal correlation between themusic bars to form a temporal model by the time structure generator Gtemp and the music bar generator Gbar; combining the hybrid track model and the time model to form a multi-track symphony composition model. The above composing method satisfies people's requirements for diversification of musical changes and is natural.

The invention discloses a method and a device for presenting multimedia information. The method comprises the following steps: receiving a four-dimensional spatial-temporal model used for presenting representation information, wherein the four-dimensional spatial-temporal model has an attribute which can present the representation information along with time changes in a digital form; carrying out decoding processing on the four-dimensional spatial-temporal model to obtain a decoded four-dimensional spatial-temporal model; and according to the decoded four-dimensional spatial-temporal model, presenting the representation information presented by the four-dimensional spatial-temporal model. In the scheme, the four-dimensional spatial-temporal model has the attribute which can present the representation information along with time changes in a digital form, so that the problem of time delay of the presented representation information is solved to a certain degree, and therefore, the scheme solves the defect of the time delay in the prior art to a certain degree.

The invention discloses a method and system for unified management of AI models based on a distributed file system. Based on the distributed file system, a model iteration management module is additionally arranged to extract preset model file information; wherein the model file information comprises information such as a model name, a model version, model creation time, whether a model is onlinepublic and whether the model is dirty, an AI model record is newly added in the metadata table, and the model is stored in a model warehouse according to a preset model storage path to construct an AImodel management system formed by combining the metadata table and the model warehouse. The newly-added model reading module analyzes data input by a user, extracts model information, matches the model information with records in a metadata table, extracts metadata items in the table, checks whether a model is online or not, and if the model is online, extracts complete target model information including a dirty model state from nodes of the distributed file system according to metadata and returns the complete target model information to the user. A user can use the model in real time and can also optimize the model and upload the model again to facilitate optimization and updating of the model.

The invention discloses a position transition prediction method based on job trajectory data. Influencing factors of user position transition are discovered and the prediction method for the user position transition is researched on a model for analyzing the user position transition. The method comprises the steps of extracting resumes from a database and obtaining history job trajectory data of users; discovering user position transition spatial and temporal characteristics by carrying out statistics, classification and analysis on massive user job trajectory data, and establishing a user position transition spatial and temporal model; extracting and quantifying the influencing factors of the user position transition from the user position transition spatial and temporal model; training a position transition prediction model through a decision tree algorithm through combination of the position transition data of the users according to definition and quantification of the influencing factors of the user position transition; and on the basis of the position transition prediction model obtained in the S4, predicting the positions after the user position transition according to the current positions of the users. According to the method, the influencing factors of the user position transition are taken into full consideration, and the transition accuracy is predicted well.

According to the present invention, a method for a travel time predicting apparatus to predict a travel time by using a travel time prediction model, including generating a first prediction model for prediction of a travel time of a vehicle that will pass through a specific area by using first traffic data that are related to a travel speed at the specific area; generating a second prediction model by modifying the first prediction model that relates to the travel speed at the specific area and is different from the first traffic data; and predicting a travel time of the vehicle that will pass through the specific area by using the second prediction model can be provided so that a travel time at which a specific vehicle passes a specific area can be more precisely predicted compared to a conventional art and predicted travel time information can be promptly and precisely provided to the user.

Data is asynchronously provided to a participant of a conversation. A temporal model is maintained for each of a plurality of client devices associated with the conversation. A temporal model describes a state of an associated client device. Data for sending to a client device is identified based on conversation context information describing a context of a conversation participant associated with the client device. A determination of whether to send the identified data to the client device is made based on the temporal model associated with the client device. The temporal model associated with the client device is updated based on the determination.

The invention discloses an industrial robot stop motion trajectory planning method. According to the method, stop planning is carried out directly for a joint position finally output in each interpolation period, delay from a trajectory interpolation layer to a joint position output layer is avoided, and response is rapider and more timely. Discrete interpolation output joint positions are storedand added with time information at the same time, a cubic spline model is used for converting discrete joint position information into continuous information in a time domain, then a time scaling method is used for stopping planning, it is guaranteed that data output in each period in the stopping process still come from the joint position output through trajectory interpolation, and then it is guaranteed that there is no trajectory deviation in the stopping process. Meanwhile, a simplified quintic polynomial is adopted to plan time, a model is simple, the calculated amount is small, and continuous, smooth and impact-free operation of the whole stop process is guaranteed through the high-order trajectory planning method.

The present invention provides a method for constructing a power GIS temporal model based on a ground state correction model. The method comprises the following steps that: S1: a user reads a specified partition geographic layout from an existing database power grid GIS, and synchronizing the specified partition geographic layout and a main version chain device of a primary connection diagram in the station into a temporal GIS library; S2, the variation of a power grid model along the time axis is divided into several continuous state sections by using the temporal GIS, so that the temporalization of the grid model is realized; S3, a day+1 manner is used to perform incremental updating in the state sections; and S4, the updated data is saved into a power grid model data table. According tothe method for constructing a power GIS temporal model based on a ground state correction model provided by the present invention, the historical data storage and retrieval efficiency can be solved,and redundant data can be greatly reduced.

The invention provides a ship dynamic positioning system nonlinear unbiased prediction control method based on input increment. The method comprises the following steps: establishing a dynamic positioning system state space model considering the influence of low-frequency environment interference caused by wind, waves and flows, and discretizing to obtain a discrete time model; taking the input increment as a state variable, taking the control input as an extended state, establishing an extended state observation model containing the input increment, and estimating an unknown state and the control input by adopting an unscented Kalman filter algorithm; based on a dynamic positioning system nominal model without interference, an unbiased NMPC optimization problem with input constraints is defined, an optimal control input increment sequence in a control time domain is solved, a first element of the sequence acts on the system, and control input at the current moment is obtained. According to the method, the ship position or the ship route of the dynamic positioning ship interfered by the environment can be set in an unbiased tracking mode, and the control precision and robustness ofa dynamic positioning system are improved.

A computerized spatial-temporal regulation method for accurate spatial-temporal model estimation receives a spatial temporal sequence containing object confidence mask. A spatial-temporal weight regulation is performed to generate weight sequence output. A weighted model estimation is performed using the spatial temporal sequence and the weight sequence to generate at least one model parameter output. An iterative weight update is performed to generate weight sequence output. A weighted model estimation is performed to generate estimation result output. A stopping criteria is checked and the next iteration iterative weight update and weighted model estimation is performed until the stopping criteria is met. A model estimation is performed to generate model parameter output. An outlier data identification is performed to generate outlier data output. A spatial-temporal data integrity check is performed and the outlier data is disqualified.

A method for identifying a mask for a patient includes: receiving a plurality of images of a patient's face; analyzing the plurality of images to generate a temporal model of the patient's face, determining a mask for the patient using the temporal model of the patient's face, and identifying the mask to the patient.

The invention discloses a Runge-Kutta type periodic rhythm neural network method capable of resisting periodic noise. The Runge-Kutta type periodic rhythm neural network method comprises the followingsteps that a tail end task is given; inverse kinematics analysis is conducted on the track of a mechanical arm through doble-infex quadric form optimization, so that a quadric form optimization scheme of the weight sum and indicator of the minimum torque and the angular deflection two-norm square of the mechanical arm is converted into a standard quadratic programming problem; the Karush-Kuhn-Tucker optimality condition is solved through a continuous time periodic rhythm neural network, so that a continuous time model is obtained; a discrete periodic rhythm neural network is obtained througha Runge-Kutta method, and a discrete solution of an original quadratic programming problem is obtained through the neural network; and finally, a result is transmitted to a mechanical arm controller,and the mechanical arm is driven to trace the track. The discrete periodic rhythm neural network designed through the method has the capability of inhibiting the periodic noise in the network model; in addition, the influence of an initial error of the mechanical arm on the motion programming can be eliminated; and motion of the mechanism arm is programmed successfully.