Patents
Literature
Hiro is an intelligent assistant for R&D personnel, combined with Patent DNA, to facilitate innovative research.
Hiro

38 results about "Discretization error" patented technology

In numerical analysis, computational physics, and simulation, discretization error is the error resulting from the fact that a function of a continuous variable is represented in the computer by a finite number of evaluations, for example, on a lattice. Discretization error can usually be reduced by using a more finely spaced lattice, with an increased computational cost.

Information processing method and apparatus

An input coordinate sequence is acquired by sampling a handwritten input pattern at predetermined intervals, and a pattern expressed by this input coordinate sequence is approximated by coupling a plurality of line segments to attain line segment conversion. Adjacent angle data φ[i] is generated based on the directions of the respective line segments. At this time, the segment line length along line segments of all the line segments is divided by a predetermined value at equal intervals, and the angles obtained from the directions of the line segments at respective division positions are defined as φ[i]. This φ[i] is compared with a standard pattern (adjacent angle distribution data) prepared in advance to obtain a matching level. In this way, more accurate pattern matching for a handwritten input, which is approximately invariant to affine transformation and can reduce the influence of discretization errors can be implemented.
Owner:CANON KK

Variable Discretization Method For Flow Simulation On Complex Geological Models

A variable discretization method for general multiphase flow simulation in a producing hydrocarbon reservoir. For subsurface regions for which a regular or Voronoi computational mesh is suitable, a finite difference / finite volume method (“FDM”) is used to discretize numerical solution of the differential equations governing fluid flow (101). For subsurface regions with more complex geometries, a finite element method (“FEM”) is used. The invention combines FDM and FEM in a single computational framework (102). Mathematical coupling at interfaces between different discretization regions is accomplished by decomposing individual phase velocity into an averaged component and a correction term. The averaged velocity component may be determined from pressure and averaged capillary pressure and other properties based on the discretization method employed, while the velocity correction term may be computed using a multipoint flux approximation type method, which may be reduced to two-point flux approximation for simple grid and permeability fields.
Owner:EXXONMOBIL UPSTREAM RES CO

Methodology and system for determining numerical errors in pixel-based imaging simulation in designing lithographic masks

A method is provided for designing a mask that includes the use of a pixel-based simulation of a lithographic process model, in which test structures are designed for determining numerical and discretization errors associated with the pixel grid as opposed to other model inaccuracies. The test structure has a plurality of rows of the same sequence of features, but each row is offset from other rows along an x-direction by a multiple of a minimum step size, such as used in modifying masks during optical proximity correction. The images for each row are simulated with a lithographic model that uses the selected pixel-grid size and the differences between row images are compared. If the differences between rows exceed or violate a predetermined criterion, the pixel grid size may be modified to minimize discretization and / or numerical errors due to the choice of pixel grid size.
Owner:GLOBALFOUNDRIES INC

Digital filter design method and device, digital filter design program, and digital filter

A digital filter is designed by combining unit filters (L10′, H10′) having a predetermined asymmetric numerical sequence as filter coefficients (H1 to H3). Thus, it is possible to automatically obtain a desired digital filter coefficient only by combining the unit filter. Moreover, a symmetric numerical sequence {−1, 0, 9, 16, 9, 0, −1} / 32 is divided at the center into two parts and one of them is used as the asymmetric filter coefficients (H1 to H3). This reduces the number of taps required for the digital filter designed, eliminates use of a window function, and prevents generation of a discretization error in the filter characteristic obtained.
Owner:NSC CO LTD

Continuous attribute discretization method based on Canopy clustering and BIRCH hierarchical clustering

The invention relates to a continuous attribute discretization method based on Canopy clustering and BIRCH hierarchical clustering. Compared with the prior art, the method solves the shortcomings of poor discretization effect and unreasonable discretization. The method includes the following steps: utilizing the Canopy clustering to achieve continuous attribute data initial clustering; utilizing the BIRCH hierarchical clustering to conduct secondary clustering with the clustering center as the sample; finding the nearest neighbors of the clustering centers of the dimension of any sample of the break point set to serve as the discrete micro adjustment basis and achieve continuous attribute discretization. The method can achieve the discretization method of the high-dimensional large-data samples, reduces the number of continuous attribute values, reduces dependency on storage space, enables the discretizated data to be regular and simplified and convenient to understand, use and explain, and expands the application range.
Owner:ANHUI KELI INFORMATION IND

C-arm tomography imaging method using semi-accurate filtered back-projection

The invention discloses a tomography imaging method suitable for C-arm imaging equipment. The tomography imaging method is characterized by having shift invariant features, wherein the semi-accurate reconstruction is performed on cone-beam projection data obtained by scanning a circular arc track. The circular arc track is a short scanning track, namely a scanning track of pi and a fan angle, and is also an ultra-short scanning track, namely a track smaller than the short scanning track. The method comprises the following steps: weighing of partial derivatives of projection data; one-dimensional Hilbert transform in a horizontal or non-horizontal direction; and back projection of the circular arc track. Compared with FDK type-approximate reconstruction algorithms, the C-arm tomography imaging method using semi-accurate filtered back-projection has the advantages of small reconstruction accuracy error, small discretization error and the like.
Owner:UNIV OF ELECTRONIC SCI & TECH OF CHINA

Discrete optimizing method for magnetic resonance image-forming superconducting magnet design

The invention relates to an optimization design method for the discretization of magnetic resonance imaging (MRI) superconducting magnets, which belongs to a superconducting magnet design technology and is characterized in that an objective function is set by a weighting method and indexes of magnetic field intensity, uniformity of magnetic field, stray field range, thread consumption, fabrication error, and the like are taken into comprehensive consideration. The use of discrete variables as optimization variables can avoid rounding and discretization errors occurring in common methods. By considering the robustness of errors in design objectives, the optimization design of the robustness is realized. The use of the design method for designing the superconducting magnets can take a plurality of design indexes into consideration in a synthesized manner, thus having the advantages of convenient processing and manufacturing and strong error robustness.
Owner:TSINGHUA UNIV

Voltage automatic calibration device

The invention discloses a voltage automatic calibration device, which includes a high precision voltage reference source, multichannel transfer relays, a voltage acquisition module and a single chip control unit, wherein one end of each of the channel transfer relays is respectively connected with the high precision voltage reference source, the other end of each of the channel transfer relays is connected with each of voltage acquisition channels of the voltage acquisition module respectively, the single chip control unit is connected with a control end of each of the channel transfer relays and is connected with the voltage acquisition module through a controller area network (CAN) bus, thereby obtaining channel acquisition voltage of the voltage acquisition module and finishing the calculation of voltage calibration value. The single chip control unit performs voltage calibration on the voltage acquisition module through the CAN bus. Without manual operation, the voltage automatic calibration device avoids battery voltage discretization errors and personal errors, and can guarantee that voltage acquisition precision is promoted from 10 m V to 2 m V.
Owner:力高(山东)新能源技术股份有限公司

Magnetic disk drive with feedback control

A feedforward control system using a polynomial formula not affected by discretization error is realized in a magnetic disk drive. In one embodiment, the magnetic disk drive is intended to move a head for performing at least one of recording and playback to and from a magnetic recording medium to a target position. The feedforward input has a seek control system computed by a polynominal formula of a time function. The coefficients of the polynomial formula have been derived from boundary conditions including the characteristics of a zero-order hold.
Owner:WESTERN DIGITAL TECH INC

Cycloidal gear tooth curve variable step size discretization method

The invention discloses a cycloidal gear tooth curve variable step size discretization method, which comprises the following steps of S1, setting a cycloidal gear tooth curve parameter; S2, deriving out a cycloidal gear tooth curve equation; S3, deriving out a formula of the curvature radius Rho of the practical tooth curve of a cycloidal gear; S4, deriving out an equation of phi relevant to a pressure angle alpha; S5, deriving out an inflexion equation of the cycloidal gear tooth curve; S6, giving the curve linear interpolation initial step size discretization method. The discretization errors and given errors are compared for automatically regulating the closing step size, so that the maximum step size smaller than the closing errors can be obtained; on the premise of meeting the precision requirements, the number of cutting feeding times is reduced; the processing efficiency is improved. The cycloidal gear tooth curve variable step size discretization method provided by the invention has the advantage that the good integral index is realized in the aspects of closing precision, discretization efficiency and calculation speed.
Owner:CRRC QISHUYAN INSTITUTE CO LTD

Digital filter design method and device, digital filter design program, digital filter

A numerical string consisting of a ratio of “−1, m, −1” or “1, m, 1” is subjected to a predetermined moving average calculation n times. A numerical string thus obtained is used as filter coefficients of a basic filter and at least one basic filter is combined in an arbitrary way for cascade connection, thereby calculating the filter coefficients of the digital filter to be obtained. This significantly reduces the number of taps and the number of multipliers used as compared to the conventional FIR filter. Moreover, by using the numerical strings “−1, m, −1” and “1, m, 1” so that the filter impulse response becomes a finite-base function, it is possible to obtain a preferable frequency characteristic having no discretization error and having a great attenuation amount out of band.
Owner:NSC CO LTD

Low-carrier-wave-ratio dead-beat control system and method for permanent magnet synchronous motor

The invention discloses a low-carrier-wave-ratio dead-beat control system and method for a permanent magnet synchronous motor. The system comprises the permanent magnet synchronous motor, an encoder,an abc-dq coordinate conversion unit, a rotating speed outer ring PI controller, a 1 / 2 moment current estimation unit, a current dead-beat controller, a dq-alpha beta coordinate conversion unit, an SVPWM modulation module and an inverter. In the process of building a current prediction model, the inverter and the motor are regarded as a whole for building the model, the delay action of the inverter is considered, and 1 / 2 moment current estimation is added for correcting a discretization error, so that the built model is more accurate than a traditional current prediction model; and a current error decoupling integral compensation module is added into the dead-beat controller, so that a stator current control offset is greatly improved, the purpose of good dead-beat control under a low-carrier wave ratio is achieved finally, and a result shows that the motor speed and stator current control is good, the motor stator current control offset is relatively small, and the output torque pulsation is also improved.
Owner:CHINA UNIV OF MINING & TECH

Method for preparing a pattern to be printed on a plate or mask by electron beam lithography, corresponding printed circuit design system and computer program

A method for preparing a pattern to be printed on a plate or mask by electron beam lithography comprising the following steps: modelling of the pattern by breaking down this pattern into a set of elementary geometric shapes intended to be printed individually in order to reproduce said pattern and, for each elementary geometric shape of the model; determination of an electrical charge dose to be applied to the electron beam during the individual printing of the elementary shape, this dose being chosen from a discrete set of doses including several non-zero predetermined doses recorded in memory. The set of elementary geometric shapes is a bidimensional paving of identical elementary geometric shapes covering the pattern to be printed. In addition, when the doses to be applied to the elementary geometric shapes are determined, a discretisation error correction is made by dithering.
Owner:COMMISSARIAT A LENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVES +1

Digital filter design method and device, digital filter design program, and digital filter

InactiveCN1706101ASimple designAchieve frequency characteristicsDigital technique networkDigital filter designWindow function
A symmetric unit filter (L10') is designed by combining in series two unit filters (1L10, 2L10) having asymmetric numerical sequences as filter coefficients (H1 to H3 and H4 to H6). Thus, by designing a filter by series connection, it is possible to automatically obtain a desired digital filter coefficient only by combining one type of unit filter (L10') in series. Moreover, a symmetric numerical sequence {-1, 0, 9, 16, 9, 0, -1} / 32 is divided at the center into two parts and one of them is used as the asymmetric filter coefficients (H1 to H3 and H4 to H6). This reduces the number of necessary taps, eliminates use of a window function, and prevents generation of a discretization error in the filter characteristic obtained.
Owner:NEURO SOLUTION CORP

Discretized differentiable neural network search method based on entropy loss function

The invention discloses a discretized differentiable neural network search method based on an entropy loss function. According to the method, a new loss term is designed into a constraint loss term suitable for different target network structure configurations based on an entropy function according to the characteristics of system entropy minimization driving system element (weight) sparsity and discretization so as to reduce discretization errors. According to the discretized differentiable neural network search method based on the entropy loss function, a discretized friendly target networkstructure is obtained through one-time search, and discretization precision loss existing in an existing search algorithm is greatly reduced; parameters of an entropy function-based structural constraint loss function may be modified to be adapted to search for arbitrarily configured network structures.
Owner:UNIVERSITY OF CHINESE ACADEMY OF SCIENCES

Tunnel resistivity modeling method and system based on hybrid grid

The invention provides a tunnel resistivity modeling method and system based on a hybrid grid. The method comprises the following steps: determining the physical size, the actual shape, the power supply and acquisition electrode position and the possible unfavorable geological area of a tunnel to be modeled; establishing a finite element geometric model according to the actual shape and size of the tunnel; carrying out mesh generation on the finite element geometric model, carrying out non-structural tetrahedral mesh local encryption subdivision on power supply and acquisition electrode positions and possible unfavorable geological areas, and carrying out mesh generation on other uniform surrounding rock media by adopting an irregular hexahedron to form an optimized model; quality parameters of all units of the optimized model are checked, iterative optimization is carried out on the shapes of the units, so that transition between the units of different sizes and different types is uniform until all the units meet the quality requirements, and the method can well fit a complex structure type and is high in practicability. The discretization error of the model is reduced to a great extent; and the forward modeling precision of an actual tunnel complex environment resistivity method is improved.
Owner:SHANDONG UNIV

Continuous attribute discretization method based on Chi2 statistics

The invention relates to a continuous attribute discretization method based on Chi2 statistics, which belongs to the field of data mining, and is characterized in that Chi2 series algorithm is analyzed, a discretization standard is re-defined, and continuous attributes are discretized more reasonably and more effectively; and a difference sequence interval combination method serves as a standard, so that all adjacent areas have fair combination opportunities, thereby well solving the problem of unfairness. In addition, the imprecise Eji value in Chi2 statistics is analyzed, and two improving schemes are presented. The continuous attribute discretization method based on Chi2 statistics has the advantages and the benefits that the high efficiency in extracting information from original data, the discretization process is more precise, and higher accuracy is realized in machine learning.
Owner:DALIAN UNIV OF TECH

Method and device for calculating Hausdorff distance approximate value between Bezier curves

The invention discloses a method and a device for calculating a Hausdorff distance approximate value between Bezier curves. The method comprises the following steps of: discretizing an input Bezier curve c1 and an input Bezier curve c2 by using a recursion diving method, wherein the discretization result is that two fold lines are formed, and a discretization error can be controlled; calculating a unidirectional Hausdorff distance between the two fold lines by using a pruning strategy on the basis of the characters of line segment set Voronoi images and the characters of the Hausdorff distance; and using the unidirectional Hausdorff distance as an approximate value of the Hausdorff distance between the Bezier curve c1 and the Bezier curve c2. The device comprises a Bezier curve discretization module, a line segment-fold line unidirectional Hausdorff distance pruning module, and a fold line-fold line unidirectional Hausdorff distance pruning module. The method and the device have the effects of easy implementation, controllable approximate errors and high calculating speed.
Owner:ENERGY RES INST OF JIANGXI ACAD OF SCI +2

Biaxial rotation modulation initial fine alignment method

ActiveCN112284419AReduce time requirements for initial alignmentHigh precisionMeasurement devicesGyroscopeComputation complexity
The invention provides a biaxial rotation modulation initial fine alignment method. The method comprises the steps: acquiring the latitude and elevation information of the position where the carrier is located, completing the coarse alignment process of the system, and executing the fine alignment process based on the navigation calculation result. The method mainly solves the problems that in thefine alignment process, due to the fact that constant zero offset of an inertial device in an error model is not completely observable, and the convergence speed is low. By introducing the double-shaft rotation modulation mechanism, an error model can be improved to meet completely observable conditions, so the precision and reliability of an alignment result are ensured. By shortening the discretization time and increasing the length of the observation sequence in unit time, richer observation information is obtained, the convergence time of the equivalent sky-oriented gyroscope constant zero offset estimation sequence is greatly shortened, and the rapidity of initial alignment is improved. By optimizing the updating method of the discretization error model, the problem that the calculation complexity is increased due to the fact that the filtering period is shortened is solved. In combination with the three parts, the patent provides a rapid and high-precision biaxial rotation modulation initial fine alignment method with high engineering practicability.
Owner:AIR FORCE UNIV PLA

Digital filter design method and device, digital filter design program, and digital filter

A symmetric unit filter (L10″) is designed by combining in series two unit filters (1L10, 2L10) having asymmetric numerical sequences as filter coefficients (H1 to H3 and H4 to H6). Thus, by designing a filter by series connection, it is possible to automatically obtain a desired digital filter coefficient only by combining one type of unit filter (L10″) in series. Moreover, a symmetric numerical sequence {−1, 0, 9, 16, 9, 0, −1} / 32 is divided at the center into two parts and one of them is used as the asymmetric filter coefficients (H1 to H3 and H4 to H6). This reduces the number of necessary taps, eliminates use of a window function, and prevents generation of a discretization error in the filter characteristic obtained.
Owner:NEURO SOLUTION CORP

Analog filter

An analog filter includes a first arithmetic operation section 2-1 having a plurality of sets of processing circuit being cascade connected, each processing circuit having an S / H circuit of plural stages for holding a ΔΣ-modulated signal and an analog adder for adding the input and output signals of the S / H circuit, in which the number of stages of the S / H circuits 11-1, 14-1, 17-1 and 20-1 decreases toward the end of cascade connection, and a second arithmetic operation section 2-2 configured in the same way, which are cascade connected. By using such an analog filter, over-sampling and convolution of a ΔΣ-modulated signal are conducted so that the envelope of the filter output may be a quadratic curve of finite carrier that converges to zero at finite sampling points to prevent phase distortion of an LPF and a discretization error due to a conventional function. Compared with a conventional circuit for over-sampling and convolution, the number of stages of the S / H circuits and the number of adders are small.
Owner:NSC CO LTD

Convolutional neural network architecture search method based on differentiable sampler and progressive learning

PendingCN114239795AOptimizing Probability Distribution FunctionsReduce discretization errorCharacter and pattern recognitionNeural architecturesEngineeringArtificial intelligence
Aiming at the limitation of the prior art, the invention provides a convolutional neural network architecture searching method based on a differentiable sampler and progressive learning, and the method comprises the steps: directly carrying out the sampling optimization of a constructed super network through employing the differentiable sampler; an optimization target of architecture search can be converted from optimization of a super network to search for an optimal probability distribution function, and the expectation of a loss function of a sub-network under the probability distribution is minimized; according to the method, the probability distribution function of the sub-network can be optimized by evaluating the performance of the sub-network, so that discretization errors are reduced. Meanwhile, due to the fact that a progressive learning strategy is adopted, stable search can be carried out in the search space which is increased in an index level, and therefore a more complex convolutional neural network architecture can be obtained.
Owner:GUANGDONG UNIV OF TECH

Analog filter

An analog filter comprising a first arithmetic operation section 2-1 consisting of a plurality of sets of processing circuit being cascade connected, each processing circuit having an S / H circuit of plural stages for holding a DELTA SIGMA -modulated signal and an analog adder for adding the input and output signals of the S / H circuit, in which the number of stages of the S / H circuits 11-1, 14-1, 17-1 and 20-1 decreases toward the end of cascade connection, and a second arithmetic operation section 2-2 configured in the same way, which are cascade connected. By using such an analog filter, over-sampling and convolution of a DELTA SIGMA -modulated signal are conducted so that the envelope of the filter output may be a quadratic curve of finite carrier that converges to zero at finite sampling points to prevent phase distortion of an LPF and a discretization error due to a conventional function. Compared with a conventional circuit for over-sampling and convolution, the number of stages of the S / H circuits and the number of adders are small.
Owner:酒井康江

Method for determining corrected rotational speed signal and electric motor device

The invention relates to a method and an electric motor arrangement for determining a corrected rotational speed signal. In one step of the method, a periodic rotational angle signal is detected which has a periodic rotational angle error which is dependent on the rotational angle. Converting the rotation angle signal into a digital original speed signal (16), due to the rotation angle error, the digital original speed signal has an interference ripple, and the cycle duration of the interference ripple corresponds to the cycle duration of the rotation angle error . The digital raw rotational speed signal (16) comprises a plurality of sampled values ​​(07), which are spaced apart in time by the duration of the sampling period, the half-period duration of the disturbance ripple of the raw rotational speed signal (16) being determined. Averaging at least one pair of sampling values ​​(07) of the original rotational speed signal (16), the time interval of the at least one pair of sampling values ​​aside from the time discretization error is the interference ripple of the original rotational speed signal (16) The known half-cycle duration. The average value is obtained by averaging. A correction value is formed that is dependent on the sampling period duration and the frequency of the disturbance ripple to account for discretization errors. The corrected rotational speed signal is formed from the difference between the mean value and the corrected value. The correction signal varies in a sawtooth fashion as a function of the difference between the sampling period duration and the period duration of the disturbance ripple, the maximum amplitude of the correction signal also increasing with increasing frequency of the disturbance ripple.
Owner:SCHAEFFLER TECH AG & CO KG

Sensor grid discretization error evaluation method based on ANSYS software

PendingCN113283133ASolving Discretization Error Estimation ProblemsOvercome the disadvantage of only being able to evaluate relative errorsDesign optimisation/simulationResourcesAlgorithmSensor grid
The invention belongs to the technical field of sensors and computers, and particularly relates to a sensor grid discretization error evaluation method based on ANSYS software. According to the method, firstly, a sensor model is established, grids are divided, then ANSYS software grid quality self-evaluation and error self-evaluation are carried out, and after self-evaluation, high-precision grid quality and an accurate simulation result are obtained by inserting convergence control and setting convergence parameters; and a final grid discretization error is calculated by adopting a Richardson extrapolation formula and an error calculation formula, and whether the error can meet requirements or not is judged, and if not, it is continued to set convergence parameters. According to the sensor grid discretization error evaluation method provided by the invention, the evaluation process of the sensor grid discretization error is described in detail, the grid quality of a sensor simulation model is improved, the grid discretization error of the sensor is solved, the accuracy of a simulation result is improved, and ANSYS error self-evaluation can only obtain a relative error.
Owner:杜玉玲 +1
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products