32 results about "Stochastic partial differential equation" patented technology

A fast and robust segmentation model for piecewise smooth images is provided. Local statistics in an energy formulation are provided as a functional. The shape gradient of this new functional gives a contour evolution controlled by local averaging of image intensities inside and outside the contour. Fast computation is realized by expressing terms as the result of convolutions implemented via recursive filters. Results are similar to the general Mumford-Shah model but realized faster without having to solve a Poisson partial differential equation at each iteration. Examples are provided. A system to implement segmentation methods is also provided.

The invention belongs to the field of numerical simulation technology of ECR ion sources, and particularly relates to an MPM hybrid algorithm applied to numerical simulation of an ECR ion source. Thealgorithm is suitable for use in an ECR ion source structure. The MPM hybrid algorithm of simulation of the ECR ion source is established through combining an MAGY theory and PIC/MCC simulation algorithms, a time-varying electromagnetic field is described by the MAGY theory, self-consistent interaction of charged particles and the electromagnetic field is described by the PIC algorithm, and inter-particle collision processes are described by the MCC algorithm. A complex and complete solving process which originally needs to be carried out on Maxwell's equations is enabled to be simplified to solving on a set of coupled one-dimensional partial differential equations about mode amplitudes, relatively larger time step length can also be taken due to that compared with high-frequency cycles, changes of the mode amplitudes are slower, and computational complexity and a computational amount are greatly reduced. In addition, an electromagnetic model is adopted, and thus compared with adoptingan electrostatic model, the algorithm can more accurately describe an actual physical process.

The invention provides a stochastic partial differential equation based wind speed fluctuation characteristic modeling method, comprising steps of: according to sample wind speed data, constructing a stochastic partial differential equation about wind speed variation, and obtaining a typical wind speed variation sequence by solving the stochastic partial differential equation; then, by fitting a probability distribution function of the sample wind speed, randomly generating a wind speed sequence, and using the solved typical wind speed variation sequence to reconstruct the random wind speed sequence according to the principle of least square method, thereby obtaining a final wind speed model; finally, verifying correctness of the modeling method through comparison with a measured wind speed sequence. The method provided by the invention can effectively depict the real wind speed fluctuation characteristics, and overcome the problem that the existing wind speed model can only consider the probability distribution of wind speed from one aspect; the wind speed model established by using the method provided by the invention can not only reflect the conditions of the probability distribution of wind speed, but also show the fluctuation characteristic of the wind speed within a required study period.

The invention discloses an aviation centrifugal pump blade profile optimization design method. Under a Matlab platform, a five-point four-time bezier curve and a linear function are adopted to controlthe blade profile circumference angle distribution and the product superposition change rule, and internal flow field numerical simulation is carried out on 15 sets of design results through a software UG and Fluent combined batch processing method. On the basis of obtaining a calculation space boundary condition through curved surface interpolation, numerical solution is carried out on the dual-harmonic partial differential equation by adopting a central difference format, and a hypercurved surface performance agent model is sestaboished. Global optimization is carried out on the design variable based on an artificial fish swarm algorithm by taking the highest efficiency as an objective function. The result shows that the hypersurface agent model based on the double harmonic equation canensure that the test value and the predicted value are completely consistent, the wake effect in the optimized centrifugal impeller flow is obviously weakened, and the hydraulic efficiency is improved by 5.4% compared with that of a prototype pump.

A guidance, navigation and control (GNC) system guides an aircraft based on a special partial differential equation (PDE). The system does not depend on a predetermined desired flight path. To overcome effects of unpredictable wind speed and direction, systems and methods effectively repeatedly determine the worst landing the wind can cause and issue flight control commands that minimize among these worst outcomes. A function C that satisfies the PDE calculates smallest miss distance a guidance system can bring about by appropriately steering an aircraft, given the aircraft's current heading, current location relative to a target and current remaining amount of time to fly. The system repeatedly determines at least a component of a current gradient of the function C. A value of the gradient component is used to select an appropriate flight control command, such as turn left or turn right.

The invention provides a method for designing a flexible mechanical arm disturbance observer based on a partial differential equation. The method includes the four steps that firstly, dynamic modeling of a flexible mechanical arm is conducted; secondly, the disturbance observer is designed; thirdly, stability of the disturbance observer is verified; fourthly, design is finished. According to the method, firstly the Hamilton principle is used, so that a PDE model of a whole system is obtained; then, based on the model, the reasonable disturbance observer is designed so that external unknown disturbance can be estimated; finally, a proper Lyapunov function is designed, so that the designed observer is analyzed and the stability of the observer is verified.

The invention belongs to the technical field of optical detection and image processing, proposes a new method for values to solve the partial differential equation, so that the partial differential equation filtering effect is improved. The directional partial differential equation filtering method for striped projection images includes the steps of 1. inputting a stripe projection image u with noise; 2. discretizing a stripe projection image u; 3. calculating the included angles Theta i, j between the strip direction and the x-axis direction; 4. performing adoption to obtain the time step Delta t; 5. performing adoption to obtain the iteration frequency Nc; 6. for the each iteration, obtaining the first-order partial derivative and the second-order partial derivative of each pixel of the image u; 7. obtaining the value solution (as shown in the description) of each pixel of the image u; 8. repeating step 6 and 7 until the set largest iteration frequency Nc is reached to stop the iteration, the value solution (as shown in the description) at this time is a filtering image. The directional partial differential equation filtering method is applied to image processing.

The invention provides an elastic soft robot kinematical modeling method based on a constant curvature assumption, and belongs to the field of robot kinematical modeling. According to the method, by means of the Kirchhoff bar theory and the constant curvature assumption, firstly, equations of force, displacement and control quantities corresponding to artificial muscles implanted into a flexible material of an elastic soft robot are established; then the actual length of the elastic soft robot at the working temperature is calculated, and mechanical parameters of the flexible material are usedfor calculation; and finally, the section inertia moment about a neutral surface is calculated, the curvature radius and the angle of the elastic soft robot at the working temperature are calculated,the true shape of the elastic soft robot at the working temperature is obtained, and kinematical modeling is completed. According to the method, the modeling process is simple and convenient, the influences of elastic deformation of the elastic soft robot are considered, numerical calculation of partial differential equations is not involved, and the relationship between the shape of the elasticsoft robot and the parameters of the artificial muscle in the elastic soft robot can be rapidly established.

A system and method are provided for computing partial differential equations in a hardware graphics pipeline. Initially, input is received in a hardware graphics pipeline. Next, the input is processed to generate a solution to a partial differential equation utilizing the hardware graphics pipeline.

Provided is a gait contour extraction method. A level set function is expressed in a spatial-temporal discrete grid. First, an SDF (Sign Distance Function) is constructed, and initialization for zeroing is carried out; then, the numerical solution of a partial differential equation is solved; and finally, re-initialization is carried out through a curvilinear sign distance function, whether the numerical value is stable or not is judged, iteration is stopped if the numerical value is stable, or the process is repeated. The gait contour extraction method has a good localization effect for image boundaries. The image edge and regional information are combined, and the definition and distinctiveness of the gait contour are improved.

The invention belongs to the technical field of a modeling method, and discloses a wind speed fluctuation characteristic modeling method based on stochastic partial differential equations (SPDE). A wind farm modeling method is proposed to establish a wind farm stochastic model, the wind farm operating characteristics are simulated, in order to achieve predictable and controllable goal services of large scale wind power generation, through simulation comparison between a wind farm equivalent model and a detailed model, the rationality of the modeling method is verified, and a wind farm model considering random fluctuations in the wind speed and direction in the study of wind farm dynamic characteristics and the influence thereof on a power grid. The invention can adjust the predictable angles of fluctuation wind speed and direction randomly, and improves the wind speed fluctuation modeling method.

The invention discloses an image denoising method based on non-downsampling wavelet transform and an improved four-order partial differential equation, and belongs to the field of image processing. The image denoising method comprises the steps of constructing a novel convexity-preserving diffusion function and establishing an improved four-order partial differential denoising model on the basis of the convexity-preserving diffusion function, wherein the corresponding energy function has a globally unique minimum value solution; combining the gaussian filtering with a laplacian operator to overcome the defect that a Y-K model is too sensitive to noise; applying the improved four-order partial differential denoising model to a non-downsampling wavelet transform high-frequency sub-band to suppress the generation of false edges and massive effects. Experiment comparison shows that the image denoising method can be used for removing image noise while retaining the image texture and other detail information.

A method for designing the profile of blade used for turbine-type fluid compressing machinery includes such steps as simplifying the flow procedure controlled by N-S partial differential equation to become the flow procedure controlled by ordinary differential equation, creating state equation, normalizing the equation to become relative optimal control mode, using the installation exit angle of blade as control variable, and iterative calculating to the 3D geometric model of blade and said ordinary differential equation.

The invention provides a non-parameterized image restoration method based on a partial differential equation and a BM3D (Block-matching and 3D filtering) and belongs to the technical field of image processing. In a conventional BM3D improved method, a parameter estimation is obtained by use of an existing noise parameter estimation method, and is then introduced into an original BM3D method, however, the result is unstable. In the image restoration method, a denoising process and a noise variance estimation process of the conventional BM3D improved are improved by use of the partial differential equation, the method is characterized by closely fusing with pretreatment of the partial differential equation, and the improved BM3D method is enabled to have high precision and stability. The non-parameterized image restoration method based on the partial differential equation and the BM3D can be applied to various fields of image processing.

The invention relates to a meshless method seepage free surface solving method based on a radial base point interpolation method. The meshless method seepage free surface solving method comprises thefollowing steps: 1) determining a control equation; 2) determining boundary conditions; 3) determining an RPIM control equation; solving a partial differential equation by adopting RPIM, dispersing aproblem domain into a series of field nodes, introducing an integral rigidity matrix into a background grid in order to facilitate integration and assembly, integral points existing in the backgroundgrid, each integral point having a support domain, and interpolating the field nodes in the support domains to obtain information on the integral points. According to the method, the seepage free surface can be solved conveniently and quickly, the precision is high, the working process can be simplified, and engineering practice can be better guided.

The invention discloses a gait contour extraction method, and relates to the technical field of gait contour extraction. The extraction method includes the steps that a horizontal set function is shown in a space-time discrete grid; an SDF function is firstly constructed; initialization for zero returning is accordingly carried out; a numerical solution of a partial differential equation is solved; reinitialization is carried out through a curvilinear symbol distance function; whether a numerical value is stable or not is judged; if iteration is stopped, the steps are repeatedly executed. The gait contour extraction method has the good localization effect on image boundaries, the image edges and regional information are combined, and the definition and the distinctiveness of the gait contour are improved.

Provided is a simulated fighting method. A level set function is expressed in a spatial-temporal discrete grid. First, an SDF (Sign Distance Function) is constructed, and initialization for zeroing is carried out; then, the numerical solution of a partial differential equation is solved; and finally, re-initialization is carried out through a curvilinear sign distance function, whether the numerical value is stable or not is judged, iteration is stopped if the numerical value is stable, or the process is repeated. The simulated fighting method has a good localization effect for image boundaries. The image edge and regional information are combined, and the definition and distinctiveness of the gait contour are improved.

Provided is a gait contour simulation method. A level set function is expressed in a spatial-temporal discrete grid. First, an SDF (Sign Distance Function) is constructed, and initialization for zeroing is carried out; then, the numerical solution of a partial differential equation is solved; and finally, re-initialization is carried out through a curvilinear sign distance function, whether the numerical value is stable or not is judged, iteration is stopped if the numerical value is stable, or the process is repeated. A new variational level set model combining the image edge and regional information is adopted. The gait contour simulation method has a good localization effect for image boundaries. The image edge and regional information are combined, and the definition and distinctiveness of the gait contour are improved.

The invention discloses a self-adaptive threshold image denoising method based on a four-order partial differential equation, and the method comprises the steps: S1, inputting an original sample imageu, and adding noise to the original sample image u to obtain a noise-containing image u0; S2, constructing a denoising model based on a four-order partial differential equation: wherein delta is a Laplacian operator, k (k > 0) is a threshold value and used for judging the characteristics of the image; S3, selecting the threshold value k (k > 0) by adopting a self-adaptive threshold value method;S4, discretizing the denoising model by adopting a finite difference method, and solving the denoising model by adopting an iteration method to obtain a denoised image. According to the method, multiplicative noise can be effectively removed, meanwhile, image edge detail information is reserved, and compared with a classical denoising model, the denoising effect is better.

The present invention relates to a dynamic model solution method for a one-dimensional plug flow reactor with the feedstock of gas. The method comprises: starting from a mechanism model of a plug flowreactor (ideal tubular reactor), discretizing partial differential equations for describing the dynamics model into algebraic equations by using a finite difference method; converting the concentration in the model equation into a function of the temperature and gas molar flow by using a gas state equation; and by using simultaneous equations, solving to obtain distribution of the temperature andcomposition of the plug flow reactor with the space and time. The method disclosed by the present invention can be used for the calculation method for the ideal tubular reactor with the feedstock component of the gas phase, and the assumption that the gas volume flow must be set to a constant parameter in the previous solution process is overcome, so that the dynamic model solution result of thereactor is more accurate.

The invention discloses a continuous reinforcement learning system and method based on a stochastic differential equation. The system comprises an action strategy generator APG, an environment state estimator ESE, a value estimator VE, a memory storage module MS and an external environment EE. The method comprises the following specific steps: an action strategy generator APG, an environment stateestimator ESE and a value estimator VE are initialized; an action strategy generator APG calculates an output action value increment delta ak; the external environment EE outputs a next action valueak + 1, a next environment state value sk + 1 and a current reward value Rk and stores the values in a memory storage module MS; the environment state estimator ESE updates the environment state parameter set theta p and predicts a future environment state estimated value s'k; the VE optimizer updates the Q function network and predicts a future reward estimation value R'k; and the APG optimizer updates the action value parameter set theta v. The method is based on a stochastic differential equation as a basic model, continuity of action control can be achieved, variance of the training process can be controlled, and actions can be selected by predicting changes of the environment so as to achieve better environment interaction.