173 results about "Metacomputing" patented technology

A system and process for performing an exponentially weighted moving average on streaming data to establish a moving average bit rate of data units is presented. In general, the system or process computes, on a unit-by-unit basis, the product of the moving average bit rate computed for a data unit immediately prior to a unit under consideration and a first fractional weighting factor, added to the product of the instantaneous bit rate of the data unit under consideration and a second fractional weighting factor, wherein at least one fractional weighting factor is not a constant but instead based on the time between data units.

A simulation method and a simulation system for a construction crane are disclosed. The simulation system includes an input device, a processing device, and a display device. The input device is used for inputting an instruction. Furthermore, the processing device includes a computation unit, a collision detection unit, a storage unit, and a graphic unit. The computation unit is used for computing the position and the direction of each part of the construction crane and the suspension parts. The collision detection model is used for detecting whether each part of the construction crane and the suspension parts will be in collision. Besides, the data obtained from the simulation method performed in the simulation system is saved in the storage unit. The graphic unit displays the 3D dynamics images of the construction crane and the suspension parts on the display device, corresponding to the data obtained from the computation unit.

The invention discloses a locomotive semi-physical simulation device, system and method. The device comprises a real-time simulation unit, a computer and an electrical signal conversion unit. The computer converts a control instruction for simulating an upper-layer network control system into a bus communication protocol, and then sends the bus communication protocol to a communication conversion unit of the electrical signal conversion unit. The communication conversion unit sends the control instruction to an external physical controller, the physical controller outputs a control signal to a controlled object model according to the control instruction, the control signal is sent to a real-time simulation unit via the electrical signal conversion unit, the real-time simulation unit calculates a output result of a state signal of the controlled object model according to the control signal, the output result is sent to the physical controller via the electrical signal conversion unit, and the physical controller sends the state signal of the controlled object model back to the computer via the communication conversion unit. The technical problems that simulation integrity and precision of an existing system are not high, function verification cannot be accurately performed, the modification period is long, and efficiency and performability are low are solved.

A hierarchical latent structure setting unit 81 sets a hierarchical latent structure that is a structure in which latent variables are represented by a tree structure and components representing probability models are located at nodes of a lowest level of the tree structure. A variational probability computation unit 82 computes a variational probability of a path latent variable that is a latent variable included in a path linking a root node to a target node in the hierarchical latent structure. A component optimization unit 83 optimizes each of the components for the computed variational probability. A gating function optimization unit 84 optimizes a gating function model that is a model for determining a branch direction according to the multivariate data in a node of the hierarchical latent structure, on the basis of the variational probability of the latent variable in the node.

Disclosed is a non-linear process fault detection method based on two-dimensional dynamic kernel principal component analysis, belonging to the fault detection technical field; the method comprises the following steps: the first step is to determine the sampling parameters, namely, to judge the execution process, determine the sampling parameters, select the data parameters which affect the fault, and then judge whether to carry through training or testing; the second step is training, that is, to collect the data of normal work, pick up the non-linear principal component of the training data through two-dimensional dynamic kernel principal component analysis, calculate the square prediction error of the training data and determine the control limit; the third step is testing, that is, to collect the online observation data, pick up the non-linear principal component of the online observation data through two-dimensional dynamic kernel principal component analysis, calculate the square prediction error of the real-time online observation data, compare the control limit of the square prediction error of the real-time online observation data with the control limit of the square prediction error of the training data, display and gives an alarm if the control limit of the former exceeds that of the latter. The invention can timely detect the fault in the production process and reduce losses in the industrial production process.

A multiply-accumulate (MAC) operation in a deep neural network (DNN) consists of multiplying each input signal to a node by a respective numerical weight data and summing the products. Using ternary values for the input signals and weight data reduces memory and processing resources significantly. By representing ternary values in two-bit binary form, MAC operations can be replaced with logic operations (e.g., XNOR, popcount) implemented in logic circuits integrated into individual memory array elements in which the numerical weight data are stored. In this regard, a ternary computation circuit (TCC) includes a memory circuit integrated with a logic circuit. A memory array including TCCs performs a plurality of parallel operations (e.g., column or row elements) and determines a popcount. A TCC array in which logic circuits in columns or rows employ a single read-enable signal can reduce routing complexity and congestion of a metal layer in a semiconductor device.

The embodiment of the invention provides a data processing method and device, and relates to the field of communication. The method comprises the steps that a first network data analysis network element determines identification information of one or more sub-models related to a target model; for the network element corresponding to the identification information of the sub-model, the first network data analysis network element sends the identification information of the sub-model to the network element; and the first network data analysis network element receives an influence value for reflecting the sub-model on the target model from the network element, and further performs training to obtain the target model. According to the embodiment of the invention, under the condition that the plurality of network elements do not exchange the original data, the target model can be trained by using the output result which is calculated by the plurality of network elements and is used for reflecting the influence value of the local model in the network elements on the target model, so that data leakage is avoided, and the more comprehensive and accurate target model can be obtained; therefore, accurate data collection, data analysis, training and other functions can be provided.

The invention provides an AMF redirection method based on slice priority, a newly added management network element, computer equipment and a storage medium, and the method comprises the steps that thenewly added management network element obtains a first AMF selectable set allocated to a terminal by a core network at a network side when the terminal is migrated from a first base station to a second base station; a second AMF selectable set is obtained on the terminal side, the second AMF selectable set comprises a plurality of AMFs corresponding to the S-NSSAIs connected before terminal migration, and the AMFs are grouped and sorted according to the comprehensive priorities of the S-NSSAIs corresponding to the AMFs; obtaining the performance of each AMF in the first AMF selectable set andthe second AMF selectable set; and according to the first AMF selectable set, the second AMF selectable set and the performance of each AMF in the first AMF selectable set and the second AMF selectable set, selecting one AMF as a redirected target AMF. According to the technical scheme, the optimal AMF can be selected as the redirected AMF, and the slicing quality is guaranteed.

The invention losslessly encodes multidimensional data such as a color lookup table in which neighboring elements in a space have high correlations, by a simple arrangement at a high compression ratio. To this end, a multidimensional data input unit inputs data of an X-Y plane in an order of Z = 0, 1, 2,... from a lookup table expressed by three-dimensional X-, Y-, and Z-coordinates. An inter-plane difference generation unit calculates differences D between elements of two neighboring planes, and outputs the calculation result as two-dimensional inter-plane differences. A prediction error generation unit considers data Di of interest in the inter-plane differences D as an element of two-dimensional data, and outputs a difference e from a predicted value p, which is obtained with reference to already encoded data, to a prediction error encoding unit. The prediction error encoding unit generates and outputs a codeword from the inputted difference e.

The invention discloses a parameterized computational analysis system for a lubricating oil film of an oil film bearing and an operating method of the parameterized computational analysis system. The invention belongs to the technical field of designing of the oil film bearing. A foreground human-computer interface comprises a welcoming interface module, an account management module, a parameter input module, a parameter guide module, a working condition selection module, an analysis flow module and a clock and state prompt module. A background computation program comprises a parameter initialization module, an Ansys parametric design language (APDL) command module, a common command language (CCL) command module, a compound file explorer (CFX) finite element computation module, a different-working-condition program module and an analysis file input / output module. The operating method comprises the following steps of: (1) installing an Ansys in a D disk; (2) starting the parameterized computational analysis system and logging in the system; (3) inputting corresponding parameters; (4) generating an APDL file; (5) starting the Ansys in the background, and generating a model; (6) generating a CCL file; (7) activating a CFX, and carrying out CFX pre-processing; (8) importing an analysis file, and directly solving; (9) carrying out CFX post-processing, checking an analysis result, and outputting an analysis report; and (10) logging out of the system.

The invention discloses an intensity finite element computing method for a wind generating set lifting tool. The method comprises the steps that firstly, an actual lifting three-dimensional model is built in Solidworks software, then, model simplification is performed in a Workbench, and a finite element computing model is built in a Hypermesh; finally, assembling, parameter setting, loading computing and post-processing are performed in ANSYS, stress of a bolt and a lifting base is analyzed, and the security is checked. Finite element computing is adopted, the influence of the rigidity of thelifting tool on stress distribution and the stress state of the local structure of the lifting tool are considered, and the welding seam with the most risky stress is subjected to intensity analysis;in addition, the influence caused by slippage on bolt stress and hole forming in the binding face portion caused by unbalance loading can be sufficiently considered. Accordingly, the method can moreaccurately and visually reflect the actual stress state in the lifting base and bolt lifting process, and therefore the reliability and safety of the lifting operation are improved.

The invention relates to a method and a system for autonomously compensating an inclined angle of a base without leveling. The method comprises the following steps of establishing a transmitting coordinate system, and calculating coordinates of an object under the transmitting coordinate system; establishing a horizontal coordinate system, and transforming the coordinates of the object under the transmitting coordinate system into coordinates of the object under the horizontal coordinate system according to the inclined angle of the base; resolving firing data under the horizontal coordinate system according to the coordinates of the object under the horizontal coordinate system; calculating and transforming the firing data under the horizontal coordinate system into firing data under the transmitting coordinate system. The system is provided with an inclined angle sensor fixed on the base and a control computer connected with the inclined angle sensor through a communication bus. According to the method and the system, regulating operation of equipment on an unprepared position is avoided, preparation time of unfolding the equipment is shortened, and quick response ability of the equipment is improved.

An approximate computation unit computes an approximate of a determinant of a Hessian matrix relating to observed data represented as a matrix. A variational probability computation unit computes a variational probability of a latent variable using the approximate of the determinant. A latent state removal unit removes a latent state based on a variational distribution. A parameter optimization unit optimizes a parameter for a criterion value that is defined as a lower bound of an approximate obtained by Laplace-approximating a marginal log-likelihood function with respect to an estimator for a complete variable, and computes the criterion value. A convergence determination unit determines whether or not the criterion value has converged.

The invention discloses an unstructured data analysis method for a three-dimensional inspection model of three-dimensional modeling software, and aims to provide an unstructured data analysis method which can improve the compatibility and applicability of a system and can be successfully identified or used. The method is realized through the following technical scheme: according to the internal relation of the three-dimensional modeling software, a three-dimensional vector finite element calculation process based on the three-dimensional modeling unstructured grid is established; when the datais stored, assembling nodes in the assembling digital-analog structure are stored in a design end product data management system in a structure presentation form; the system serves as a manufacturingend PDM system of an application layer, an assembly part document is regenerated according to the position coordinate matrix recorded in the database; and the data is imported into a manufacturing end PDM system to construct an unstructured three-dimensional model data analyzer, geometric data and non-geometric data in the three-dimensional model data and required non-geometric data labels are analyzed, and all xml information is converted into identifiable unstructured three-dimensional model data.

The invention discloses a self-adaptive degree-of-freedom electromagnetic temperature multi-physical field coupling analysis method. The method comprises the steps of establishing an electrical equipment geometric model and initial unstructured grid discretization; establishing an electromagnetic temperature multi-physical field weak coupling numerical model; analyzing finite element calculation format derivation based on an electromagnetic temperature multi-physical-field weak coupling numerical model of an unstructured grid unit; calculating and solving an electromagnetic temperature multi-physical field in the electrician equipment, and carrying out error analysis on a numerical solution; and adjusting the degree of freedom of each physical field, and solving again according to the adjustment result until the precision of the numerical solution meets the requirement to finish analysis. According to the method, independent and flexible adjustment of the freedom degrees of two physical fields is achieved on one set of grids, different requirements of the physical fields for discretization are met with small computing resources, grid sparsity and subsequent complex operation of actual operation are avoided, a grid mapping function and errors possibly introduced by the grid mapping function are prevented from being used, and the calculation efficiency of electromagnetic temperature coupling analysis calculation is effectively improved.

The invention discloses a method for calculating a load amplification factor in system dynamic analysis. The method comprises the steps of establishing a system model dynamic analysis model by applying a finite element calculation program,through a mechanical analysis result of a nominal model, determining key input and output parameters by combining contact relationships between the components; using the distribution interval and the uniform distribution hypothesis of key input parameters; establishing a probability density function for describing key input parameters of the system model; obtaining a probability density function and a distribution interval thereof; applying a Latin hypercube sampling algorithm to perform numerical sampling on key input parameters of the system and verifythe rationality of the key input parameters; creating a key input-output big database through finite element transient calculation to make a result analysis of the database, acquiring the ratio of anupper limit value of a 99% confidence interval to a nominal model calculation value and quantitatively giving the dynamic response influence degree, namely a load amplification factor, of input uncertainty on response.

The invention discloses an intermittent industrial process online quality prediction method based on JY-MKPLS. According to the method, two-dimensional input matrixes X and X and two-dimensionaloutput matrixes Y and Y are obtained through three-dimensional input data of a process a and a process b; standardization processing is performed; the matrixes are projected from a low-dimensional original space to a high-dimensional characteristic space F, and kernel matrixes K and K are calculated in the high-dimensional characteristic space; the kernel matrixes K and K are standardized; the JY-MKPLS algorithm is run; load matrixes of K<ai> and K<bi> are calculated; the steps are repeated till A principal elements are extracted; a score matrix T and a load matrix P of an input data matrix K and a score matrix U and a load matrix Q of an output data matrix Y are calculated; batch process quality prediction is performed; latest output data y<new> is obtained on line, anda prediction error beta<n> of a current batch is calculated; a model prediction error is checked; newly generated a-process data is used for replacing the model prediction error; and model updating is performed. Through the method, a new process prediction model with high precision can be quickly established, modeling efficiency and prediction precision can be improved, and the operation cost ofan enterprise is effectively controlled.

In order to provide simple, fast, and reliable verification of the functioning and processing of an automation task in the form of software in a multi-channel safety-oriented automation component (1), the software (SW1) is run in one channel (K1) of the automation component (1) in an active unit (P1) of the hardware of the channel (K1), and first diversity software (SW3) redundant relative to the software (SW1) is run in a verification unit (V1) in this channel (K1), wherein in a processing step (Z1) input data (Ez) associated with the software (SW1) and first output data (Az) computed by the software (SW1) in this processing step (Z1) are temporarily stored in a memory unit (M1), and the diversity software (SW3) in the verification unit (V1) computes second output data (Az′) based on the stored input data (Ez) independently of the processing of the software (SW1) in the active unit (P1), and the second output data (Az′) computed by the diversity software (SW3) is compared with the stored first output data (Az) of the software (SW1) in order to verify the processing.

The invention provides a method for realizing the shortest path of double layers based on a cluster computing unit. The method comprises the following steps: adding node identification information responsible for each cluster computing unit layer in a flooding information packet of the cluster computing unit layer according to a routing protocol expanding the cluster computing unit layer so as to improve the permeability of layer topology of the cluster computing unit layer, calculating the shortest path reaching a target cluster computing unit from a source cluster computing unit, and directly transmitting a path selection request to the target cluster computing unit along the shortest path to avoid blooding of the path selection requests, thereby greatly reducing the problem of network overloading caused by information blooding. Meanwhile, the adopted method of double layer shortest path can greatly save path selecting resources, reduce convergence time of path selection in large scale, and reduce the complexity of path selection. According to the method, the problem of excessive blooding information caused by mass path selection requests, and problems of time convergence difficulty and computing resource waste in path selection in multiple domains can be solved.

The invention discloses an outbound algorithm for dynamically adjusting the call frequency. The algorithm comprises the following steps: extracting a plurality of historical call data samples in different time periods, and carrying out data marking to determine whether data are connected or not; carrying out secondary processing on the marked data, calculating an internal and external call connecting rate and a call loss rate of the connected data in unit time, and taking the rates as expected input values; constructing a three-layer neural network, inputting the historical call data sample and the expected input value into a neural network, and calculating an actual call value per minute; comparing the actual call value with the expected input value, reversely adjusting and optimizing thecall connecting rate and the call loss rate, inputting the adjusted and optimized call connecting rate and call loss rate into a neural network unit as new expected input values, and calculating an optimal call value per minute; and predicting an outbound machine to obtain an optimal call value per minute, initiating a test type batch outbound call according to a current online seat quantity anda current idle seat quantity, initiating a test type batch outbound call and dynamically adjusting the outbound call frequency according to different project models.

The embodiment of the invention discloses a computing power sharing method based on network scheduling and a related product, which are applied to a distributed computing system, and the distributed computing system comprises an application function network element, a computing cooperation end and a computing demand end. The method comprises the steps that firstly, after an application function network element receives computing power service request information from a computing cooperation end or a computing demand end, computing power service capability inquiry information is sent to a preset target, and the preset target comprises the computing cooperation end and / or the computing demand end; the application function network element receives response information from a preset target and provides a computing power sharing service according to the response information, and finally the application function network element receives notification information which comprises notification information sent by the preset target when the preset target determines that the computing power sharing service is completed; through the method, the efficiency and the convenience of computing power sharing can be improved.

The invention provides a computer aided design system for an operation process of a pulse column extractor. The computer aided design system is characterized by comprising a data input unit, a data inspection unit, a circulation unit, a calculation result inspection unit and an error data management unit, wherein the data input unit is connected with the data inspection unit, the data inspection unit is connected with the circulation unit, the circulation unit is connected with the calculation result inspection unit, and the error data management unit is respectively connected with the data inspection unit and the calculation result inspection unit. Compared with the prior art, the computer aided design system has the advantages that changes of concentrations of corresponding components in the process operation process can be calculated and displayed in real time, errors possibly occurring in abnormal operating conditions can be reported in the process operation process, after system operation is ended, the result is automatically stored in a storage unit, and users can conveniently analyze and process computational results under different conditions.

The invention relates to the technical field of intelligent management of oil production wells, and discloses an oil well full-life-cycle management method and system of a side cloud architecture, and the method comprises the steps: a unique data source collection intelligent controller is used for managing full-service chain data in the full-life-cycle of an oil well in an oil field, and data attributes with an oil well number as a unique identifier are established; collecting and recording all dynamic production and maintenance measure data of the oil well; establishing an oil well data management cloud platform based on the block chain; the intelligent unit calculates to obtain oil well characteristics, oil well data and the characteristics are stored in the cloud platform, and the cloud platform encrypts the stored data; the user terminal sends the unified data identification code to the cloud platform, the cloud platform sends corresponding encrypted data to the user terminal, and the user terminal decrypts the data and displays a statistical chart. According to the oil well full-life-cycle management method, edge calculation of collected data is completed through an intelligent unit, and oil well data collection, calculation, storage and management based on an edge cloud architecture are achieved.

The invention discloses a double-weight spatial interpolation method. The method comprises the following steps: S1, determining an interpolation resolution, a spatial interpolation range and an interpolation attribute field, and dividing grid units; s2, calculating a two-dimensional coordinate of a central point of each to-be-interpolated grid, and screening sampling points meeting an interpolation requirement quantity according to a nearest distance principle; s3, calculating a distance weight and a local spatial position interpolation attribute variation weight of the sampling point, and performing interpolation on the grid point to be interpolated by using the distance and the difference coefficient as double weights; and S4, outputting the obtained interpolation result in a grid form.According to the invention, the spatial self-similarity principle is considered based on the classical interpolation method; in combination with variation characteristics of sampling point attribute values used for interpolation in space, a quantitative expression formula of sampling point variability is designed and serves as a second weight to be applied to interpolation calculation, and the adverse effect of sampling point abnormal values on an interpolation result can be effectively weakened.

The invention relates to the technical field of artificial neural networks, in particular to an asynchronous communication interconnection architecture and a brain-like chip with the architecture, and the asynchronous communication interconnection architecture comprises an in-chip asynchronous communication interconnection architecture, an inter-chip asynchronous communication interconnection architecture, a neuron calculation unit and an on-chip routing unit; the neuron calculation unit and the on-chip routing unit are respectively provided with an independent clock management module in an independent clock domain, the neuron calculation unit and the on-chip routing unit are connected in the same brain chip, and the on-chip routing unit and the adjacent on-chip routing unit are connected through the on-chip asynchronous communication interconnection architecture. The adjacent brain-like chips are connected with each other through the asynchronous communication interconnection architecture between the chips. According to the method, a large number of neuron computing units can be efficiently integrated in the brain-like chip, and meanwhile, the brain-like chip is efficiently cascaded and expanded, so that neuron computing resources of a large scale are obtained.

The invention discloses a universal safety computer for a train control system. The universal safety computer for the train control system comprises a core host, and a radio block center (RBC) man-machine operation display unit, a temporary speed restriction server (TSRS) man-machine operation display unit, a computer based interlocking (CBI) man-machine operation display unit and a remote peripheral device which are in communication with the core host through a secure network. The core control logics of an RBC, a TSRS, CBI and a train control center (TCC) are borne in the core host; and the remote peripheral device is electrically connected with I / O ports of the CBI, an orbital electronic unit and a lineside electronic unit. According to the universal safety computer for the train control system, the core signal function logics of the CBI, the TSRS, the TCC, the RBC and the like, as well as the functions of various scheduling and management tasks, such as the scheduling of tasks of the system, monitoring over the sound condition of the system and the like are integrated, so that different functions are fulfilled at different stages by calling each signal logic.

The invention provides a word vector correction method based on semantic relation constraint and a computing system. The method comprises the following steps: a synonym and antonym constraint set with a symmetric relation and a direct hypernym / hypernym constraint set with an asymmetric relation are extracted as an external knowledge source update word embedding vector from WordNet and Roget semantic dictionaries; and the similarity of the pair of vocabularies can be calculated by randomly inputting two vocabularies of which the similarity needs to be calculated. The vocabulary semantic similarity calculation system comprises an input unit, an initialization unit, a calculation unit and an output unit. According to the method, the existing neural network word embedding vector is updated based on the lexical semantic relationship constraint provided by the external knowledge source so as to be used for lexical semantic similarity calculation. The updating speed and the word embedding vector semantic updating effect are obviously better than those in the prior art, and the accuracy rate in calculating the word semantic similarity is higher.

The invention discloses a method for predicting a curing deformation rebound angle of an L-shaped composite material workpiece, which belongs to the technical field of finite element calculation of composite materials, and is characterized by comprising the following steps of: a, model establishment: constructing a model profile in graphic processing software, and assigning a value to the feature structure variable of a model, importing the molded surface of the model into finite element processing software, and carrying out grid division, boundary constraint and material laying layer setting; b, performing finite element simulation calculation; c, measuring a deformation rebound angle; d, generating a prediction model. According to the method, the geometric model of multiple groups of continuous variables is established, finite element simulation calculation is carried out through feature assignment, and finally the curing deformation springback angle prediction model under any geometric structure parameter is established, so that the curing springback deformation condition of the L-shaped composite material part under any geometric structure parameter can be accurately predicted.