79 results about "Triangular decomposition" patented technology

The invention discloses a three-dimensional visual information acquisition method based on two-dimensional and three-dimensional video camera fusion. According to the method, a two-dimensional video camera and a three-dimensional video camera form a composite video camera to respectively image a scene synchronously; by establishing a matched mapping model of three-dimensional video camera depth images and two-dimensional video camera images, matched mapping points and mapping areas of the pixels of the three-dimensional camera video depth images in the two-dimensional video camera images are obtained; the mapping areas in the two-dimensional video camera images are processed through triangular decomposition, and a depth curved surface interpolation function of a triangulation area is established on the basis of depth information of adjacent mapping points; depth information of image pixels in the triangulation area is calculated through interpolation, and then depth images corresponding to high-resolution two-dimensional video camera images can be acquired. According to the method, the two-dimensional video camera image information and the three-dimensional video camera depth information are kept to the maximum degree, and the method has the advantages of being high in precision, little in information loss and the like and can be widely applied to the fields of industrial visual measurement, visual assembly, robot visual navigation and the like.

The embodiments of the present invention disclose a method and an apparatus for detecting signals of a MIMO system, and relate to a MIMO technology. The method includes: performing triangular decomposition on a channel matrix of the MIMO system, and decomposing the MIMO system into multiple MIMO subsystems; detecting the receiving signal vector of the first MIMO subsystem, and obtaining the candidate detection vector list of the first MIMO subsystem; detecting remaining MIMO subsystems consecutively in a SIC mode, and generating candidate detection vector lists of the remaining MIMO subsystems; and combining all the lists, and marking decision or performing calculation to obtain the transmitting signal vector of the MIMO system.

The invention discloses a low-complexity precoding method for a downlink multi-user multiple-input multiple-output (MIMO) system. The method is characterized by comprising the following steps of: setting a global channel matrix, calculating a pseudo-inverse matrix of the global channel matrix, performing row decomposition on the global channel matrix, performing column decomposition on the pseudo-inverse matrix, performing orthogonal triangular decomposition on a block pseudo-inverse matrix of a user m, calculating an equivalent channel matrix of the user m, constructing a Hermitian matrix by using the equivalent channel matrix of the user m, performing LDL<H> decomposition on the Hermitian matrix, and calculating a precoding matrix and a receiving matrix of the user m according to L and D. The method has the advantages that by the null-space calculation of the pseudo-inverse matrix of the global channel matrix and the orthogonal triangular decomposition of a block, complex matrix singular value decomposition is avoided, and channel block diagonalization is quickly realized; and in addition, an equivalent channel of each single user is subjected to Hermitian matrix decomposition to finish the design of the precoding matrix, so that a calculation amount is effectively reduced under the condition of no loss of bit error rate performance.

The invention discloses a parallel LLL high-dimensional ambiguity decorrelation algorithm. The algorithm includes: firstly, the calculating efficiency of the LLL algorithm aiming to high-dimensional ambiguity decorrelation is improved and the ability of high-dimensional ambiguity decorrelation is enhanced by mixed adoption of Cholesky lower triangular LTL decomposition and an upper triangular UTU decomposition; secondly, in order to obtain a Z transformation matrix with high decorrelation ability, in each QR decomposition transformation process, a transformation coefficient matrix needs to obtain a smaller integer value so that before each time of lower triangular decomposition, row vectors of an ambiguity covariance matrix are in an ascending order according to the inner products, and column vectors of the matrix are arranged in a descending order according to the inner products before the upper triangular decomposition, in this way, the Z transformation decorrelation performance is better; and finally, the rounding operation of the algorithm in an orthogonal transformation process is shifted to the operation process of the Z matrix, error accumulation caused by repeated rounding operation in the algorithm iteration process can be avoided, the problem of algorithm divergence is solved, and the calculating efficiency and stability of the parallel LLL algorithm can be further improved.

The invention discloses a method for identifying modal parameters of an ocean platform structure. The method comprises the following steps of 1, carrying out directional processing on input environmental excitation response data of the ocean platform structure; 2, constructing a Hankel matrix by utilizing the structural environment excitation response data, carrying out orthogonal triangular decomposition on the Hankel matrix to obtain a projection matrix, then carrying out singular value decomposition on the projection matrix, and solving a vibration system state space equation by utilizing aleast square method and the like; and finally, extracting structural modal information from the system matrix A and the output matrix C in the state-space equation of the vibration system; 3, performing clustering analysis based on density division on the modal result set by utilizing a density-based clustering algorithm; and 4, obtaining real modal parameters, and analyzing a result of the realmodal parameters. According to the method, the real mode of the ocean platform structure can be automatically extracted from the environmental response signal. Therefore, most of noise points are eliminated. Moreover, the health condition of the platform can be evaluated according to a modal parameter identification result.

The invention relates to an HARDI compressed sensing super-resolution reconstruction method based on deep dictionary learning. The method comprises the steps of preprocessing an acquired high-angle diffusion image to obtain training data, establishing a deep network model which is used for dictionary learning and comprises a plurality of layers of dictionaries, training the constructed deep network model through the training data; extracting orthogonal vectors in sequence by adopting orthogonal triangular decomposition to serve as an initial dictionary; the method comprises the following stepsof: solving a last layer of learning dictionary; adding a sparsity constraint item to carry out sparse representation, collecting data of which the density is far lower than that of original data astest data, obtaining a sparse representation coefficient based on the test data, and finally generating a reconstructed three-dimensional diffusion magnetic resonance image related to a human body through a direction distribution function obtained through radial integration. According to the method, the sampling data amount required for reconstructing the diffusion magnetic resonance image with the same resolution is smaller. And the data sampling speed is higher. And the nerve fiber reconstruction capability is better.

The invention discloses a self-interference suppression method based on block diagonalization and triangular decomposition. The implementation steps are as follows: at first, obtaining a channel matrix of each user, the block diagonalization algorithm is used to obtain the first layer precoding matrix of each user, multiplying the first layer precoding matrix of each user by the channel matrix ofthe corresponding user, the second layer precoding matrix of each user is obtained by triangular decomposition and singular value decomposition of the equivalent channel matrix of each user. Finally,the conjugate transpose matrix of the left unitary matrix of the lower triangular matrix of each user is used as the reception decoding matrix of each user. The second layer precoding matrix of each user is obtained by triangular decomposition and singular value decomposition. The invention can eliminate the interference between users, restrain the interference of multiple antennas of the users themselves, improve the transmission rate of the system users and improve the transmission performance of the system downlink.

The invention provides a method for quickly solving a nodal impedance matrix of a power system, and relates to the field of analytical computation of the power system. The method mainly comprises the following steps of inputting data of a nodal admittance matrix Y; establishing an augmented matrix B by the nodal admittance matrix Y and an identity matrix E together; normalizing the augmented matrix B and carrying out a Gauss-Jordan elimination method on the augmented matrix B for n times; obtaining an inverse matrix Z. At present, traditional methods for solving the nodal impedance matrix comprise an LDU (Logic Data Unit) triangular decomposition method and a Gauss elimination method, and compared with the two traditional methods, the novel method for quickly solving the nodal impedance matrix by utilizing the Gauss-Jordan elimination method, provided by the invention, has the advantages that the principle is simple and easy to understand, the computation time is reduced, the programming is convenient, and the like; compared with the traditional LDU triangular decomposition method and the Gauss elimination method, by utilizing the method for verifying systems such as an IEEE-57 node, an IEEE-118 node and an IEEE-300 node, the computation speeds can be respectively increased by about 25%-50%.

The invention discloses a method for rapid simulation of a non-stationary random course based on characteristic orthogonal decomposition. With regard to a multi-point non-stationary course which has arbitrary space distribution and is related to time variation, an EPSD matrix is composed triangularly at first, then POD is used to denote each time-frequency coupled element of the triangular matrix by the sum of products of multiple time functions and multiple frequency functions, and finally an FFT technology is used for rapid simulation; and with regard to a multi-point non-stationary course which has arbitrary space distribution and is related with time invariability, a related matrix is decomposed triangularly at first, then POD is used to denote time-frequency coupled elements of a diagonal matrix composed of auto spectra by the sum of products of multiple time functions and multiple frequency functions and finally a simulation formula is optimized, so that the FFT technology is only used for a few of times for efficient simulation. According to the invention, the POD is used, fitting accuracy is high, and simulated samples are accurate; simulation efficiency is greatly increased through triangular decomposition and optimal use of the FFT technology; and problems existing in current non-stationary simulation methods that applications are limited and efficiency is low can be solved.

The invention belongs to the field of power system analysis and computing and discloses a method for solving a node impedance matrix of an electric system on the basis of a Gaussian elimination method of a sparse symmetric matrix technology. The method mainly comprises the following steps that a node admittance matrix Y is formed; the matrix Y and a matrix En form an augmented matrix Bn=[YEn]; elimination is carried out on the matrix Bn according to the spare symmetry to obtain Bn(n-1)'=[Y(n-1)'En(n-1)']; according to Y(n-1)'Zn=En(n-1)', sparseness and symmetry, elements above and on the left of a diagonal element Znn of a matrix Zn are solved; a matrix Y(k-1)' is obtained according to the Y(n-1)'; elements above and on the left of a diagonal element Zkk of the matrix Zk are obtained according to Y(k-1)'Zk=Ek(k-1)', sparseness and symmetry. By the utilization of the symmetric sparseness, all invalid computation of the previous generation process is avoided, and computation of about 50% of nonzero elements is reduced; by the utilization of the characteristics of the E matrix element structure and the sparseness of upper triangle elements, the elements of the matrix Zk are obtained in a back substitution mode according to a symmetry mode, and back substitution computation is greatly accelerated. The method can check IEEE-30, -57 and -118 node systems and the like, and the computation speed for the IEEE-118 node system can be improved by 96-97% compared with a traditional Gaussian elimination method and an LDU triangular decomposition method.

The invention relates to a method for extracting a mesh graph from a vector graph and performing filling through quick computing. The method comprises the following steps of reading input GDS vector graph information by a control module and analyzing a corresponding layer of a GDS file; performing conversion defined in the GDS vector graph file on a vector point, and forming absolute coordinates of the vector point of the GDS vector graph file; meshing the vector graph by the control module according to a view field size and a lens magnification rate of a collection module, and according to the absolute coordinates of the vector point, performing division and clipping on a polygon where the vector point is located according to coordinates of meshes; and obtaining polygon coordinates of meshes of a stripe once by the control module, and sending the polygon coordinates to a GPU for performing parallel filling. According to the method, a universal GPU computing card is adopted for realizing conversion from the vector graph to a bitmap; a plurality of GPU computing cards can be adopted for cooperative processing according to system demands; and triangular decomposition of the polygon does not need to be carried out in the processing process, so that the processing performance and production capacity of a system are improved.

The invention discloses a large grid Thevenin equivalent parameter online identification method based on LU decomposition, comprising: based on the single state section data of a large grid, directlyusing a node voltage equation to obtain an open-circuit voltage at a node to be solved as a Thevenin equivalent potential and then obtaining all Thevenin equivalent parameters; when a generator node is reactive out of limit, changing the type of a PV node in order that the PV node becomes a PQ node, and quantifying and calculating the equivalent parameters; in the process of solving the open-circuit voltage of different nodes, because a constructed linear node voltage equation coefficient matrix is a constant matrix, obtaining an equation solution by using LU triangular decomposition through fast back substitution.

The invention provides a laminated matrix decomposition-based low frequency electromagnetic property simulation method. Firstly, the problem of low frequency collapse of the conventional electric field integral equation method is eliminated by employing an augmented electric field integral equation method; a low frequency multi-level fast multi-pole method is adopted to establish a laminated matrix expression for the coefficient matrix of the augmented electric field integral equation; the laminated matrix is further compressed by using a double compression method to remove redundant information; finally, a laminated matrix format algorithm is adopted to perform upper-lower triangular decomposition on the compressed laminated matrix. The laminated matrix decomposition can reduce calculating complexity and memory consumption to be almost linear; a precondition technology is constructed for an iteration solution and a direct solution is constructed. The method has the advantages of high solving speed, low memory consumption and precision controllability and can be applied to various kinds of low frequency electromagnetic property analysis.

The invention discloses a method for calculating short-circuit current of a power network with a high-temperature superconducting current limiter, comprising the steps of establishing a short-circuitcurrent calculation method with a high-temperature superconducting current limiter and a high-temperature superconducting current limiter; Short-circuit current calculation model of FCL; inputting grid data and fault information into the model; simplifying the original network; constructing the admittance matrix of the simplified network; triangulating the admittance matrix; solving the voltage vector by the forward-backward substitution method; recalculating the impedance of the nonlinear branch according to the voltage vector, and updating the admittance matrix; calculating the sum of the currents of adjacent branches of the fault node, and further calculating the current deviation; judging whether the current deviation is less than a preset threshold value, and if so, completing the calculation; injecting the three-sequence currents of the fault node and the bus at both ends of the current limiter into the linear network, and solving the other bus voltages and branch currents. The technical proposal provided by the invention has the advantages of simple calculation, accurate result, strong practicability and the like, and can provide the basis for the safety and stability analysis of the electric network containing the high-temperature superconducting current limiter, the relay protection setting calculation and the like.

The invention discloses a GPU (Graphics Processing Unit) acceleration method for dynamic response analysis of a thin plate gridless Galerkin structure. The method comprises the following steps: reading sheet data into a host memory through a CPU (Central Processing Unit), calculating relevant data required by parallel processing of a GPU (Graphics Processing Unit) according to input data, and copying the obtained data into a GPU global memory; setting the number of thread blocks and the number of threads on the GPU, performing parallel acceleration calculation on shape function values corresponding to the nodes, then establishing a one-to-one mapping mode of the thread blocks and the cross nodes of the GPU, and performing parallel acceleration assembly on an overall stiffness matrix and amass matrix of the thin plate; correcting the overall stiffness matrix on the GPU according to boundary conditions, then obtaining a damping matrix and an effective stiffness matrix through the mass matrix and the corrected stiffness matrix, and performing triangular decomposition on the effective stiffness matrix; and carrying out dynamic response analysis on the thin plate in a GPU by adopting aNewmark method, and finally outputting a result of solving the displacement, speed and acceleration. According to the method, the solving efficiency of dynamic response analysis is greatly improved.

The invention discloses a method for parallel detection of a multiaerial system based on a breadth-first search. The method is characterized in that a processing system is connected with a preprocessing module through input, the output end of the preprocessing module is connected with the input end of a tree-searching module, the output end of the tree-searching module is connected with the input end of an output module, and the output end of the output module is system output; new conversion from a complex-number detection model to a real-number detection model is introduced, the particular characteristics of an upper triangular matrix obtained through orthogonal upper triangular decomposition conducted on a lower real-number channel matrix in the real-time detection model is used, and the method that two layers are calculated at the same time is used for accelerating operation, so that the parallel degree of an algorithm is improved; the subtree multiplex method is used for reducing operand, the instruction cycles are reduced, the operation speed of the algorithm is increased, and the throughput rate of coding is increased. Compared with an existing list Kbest detection method, the method can guarantee that the performance is not decreased, and the needed execution cycles are averagely reduced by 35% in the multiaerial system detection process.

The invention relates to an electric power system node impedance matrix Z solving method based on CU triangular decomposition and belongs to the field of electric power system analysis and computing. The method mainly includes the steps of forming a node admittance matrix Y; performing quick CU triangular decomposition on the matrix Y; solving a matrix Wk through CWk=Ek; solving non-diagonal elements of a diagonal element Zkk and the diagonal element above the diagonal element Zkk of a matrix Zk through UZk=Wk; solving non-diagonal elements on the left of the diagonal element Zkk in accordance with the symmetry; writing data of a matrix Z to a data file. According to the electric power system node impedance matrix Z solving method based on CU triangular decomposition, a CU triangular decomposition method higher in calculation efficiency than a LDU triangular decomposition method is used for solving elements of the matrix Z, the symmetrical relationship between elements of a matrix Y and elements of a matrix C and a matrix U is used for quickly forming a composite matrix of a factor matrix C and a factor matrix U, and the symmetry of the matrix Z and the calculation order of elements of the matrix Zk is used for omitting back substitution solving of lower triangular elements of the matrix Z for achieving direct solving, so that the speed of formation of the matrix Z is greatly improved. Compared with the traditional LDU triangular decomposition method, the electric power system node impedance matrix Z solving method based on CU triangular decomposition has the calculation speed increased by about 33% when the method is used for IEEE-30, IEEE-57 and IEEE-118 node system checking calculation.

The invention discloses a short-circuit current calculation method based on a dynamic phasor time domain method. The short-circuit current calculation method comprises the steps of establishing an element dynamic phasor model suitable for short-circuit current calculation of a three-phase alternating current system; converting the dynamic phasor model into an algebraic equation by using a trapezoidal integration method to obtain a conductance matrix of a fault node; triangular decomposition is carried out on the conductance matrix, a formula obtained after triangular decomposition is used forcarrying out current calculation on a faulted circuit, and the problem that the calculation precision is insufficient when a short-circuit current standard simplified practical calculation method generates short circuit at the positions near a long-distance power transmission line and a series compensation capacitor is solved.

The invention provides an LR triangular decomposition method based on a symmetric sparse matrix technology and non-zero element random storage, and belongs to the field of power system analysis and calculation. The method mainly comprises the following steps: writing Y-matrix Y (n, d) data file data stored according to a random sequence into an LR (n, d) array; obtaining a reciprocal of the diagonal element rii in the ith row, and assigning elements such as lij in the fourth column to the fifth column in the ith row of non-diagonal element group at a time according to the column proportion according to the elements such as rij in the first column to the third column in the ith row of non-diagonal element group; determining a calculation element r and a storage mode thereof according to thediagonal element, the non-zero cross element and the non-zero elimination element; calculating an element r step by step by using a quadrangle rule; solving diagonal elements zkk and elements above zkk in the Z'k matrix according to an RZ'k = E'k equation; obtaining elements on the left of zkk according to symmetry; and writing the Z matrix data into a data file. For an IEEE-300 node system, compared with a traditional LR decomposition method, the method has the advantages that the data file reading time, the LR decomposition time and the LR decomposition + back substitution time are about 5.75%, 2.16% and 6.83% and are all reduced along with increase of the system scale.

The invention discloses a distributed information weighted coherent state filtering method for a sensor network, belonging to sensor information fusion technology, which relates to a distributed nonlinear state estimation problem of the sensor network. Firstly, a square root volume rule is used to obtain the prediction information vector and the square root information matrix of each node to the target state. Then, each node updates local state information by weighted local prediction information and observation information vector, updates local square root information matrix by weighted localprediction information and observation information matrix, and combines matrix triangular decomposition. Finally, the information consistency of the whole network is achieved by using the weighted consistency of the neighboring nodes, and the estimation of the target state is completed. This method improves the convergence rate of the state estimation in the network, enhances the numerical stability of the algorithm, and improves the accuracy of the state estimation of the target under the condition of limited energy.

The invention discloses a method for solving a nodal impedance matrix of an electric system on the basis of a triangular decomposition method of A=LDU, and belongs to the field of analysis and computation of the electric system. The method comprises the main following steps of: forming the nodal impedance matrix Y, and carrying out LDU triangular decomposition on the matrix Y; solving hkk elements by using an equation of DHk=Wk(Ek); solving diagonal elements Zkk and off-diagonal elements above the diagonal elements Zkk of a matrix Zk for UZk=Hk; solving off-diagonal elements at the left of the diagonal element Zkk according to the symmetry; writing data of a matrix Zk to a data file. The method disclosed by the invention has the advantages that the computation of the matrix W is omitted by mainly utilizing the structural characteristics of a matrix Ek of a unit matrix E, the computation sequence of elements in the matrix Zk and the symmetry of the elements of the matrix E, the computation to a matrix H is simplified into computation to diagonal elements hkk, the computation of 50% off-diagonal elements can be omitted for the matrix Z, and the computation speed of the elements of the matrix Z is greatly increased. Compared with the traditional LDU triangular decomposition method, the method disclosed by the invention has the advantage that by carrying out checking on node systems such as IEEE-57, IEEE-118 and IEEE-300, the computation speed can be increased by about 35-45%.

The invention discloses a method of quickly solving a node impedance matrix of an electric power system based on Gaussian elimination method, relating to the field of analytical calculation of the electric power system. The method comprises the following main steps of inputting data of a Y array of a node admittance matrix; forming an augmented matrix Bn=[YEn] by utilizing the last column of En arrays of the Y array and the E array; carrying out the (n-1)th normalized Gaussian elimination on the Bn array to obtain Bn(n-1)'=[Y(n-1) 'En']; solving the Zn array in the Z array of the node admittance matrix according to Y(n-1)'Zn=En'; obtaining all elements at the left side of Znn according to symmetry; solving diagonal elements of the Zk array of (n-1)th column to first column of the Z array and the elements according to Y(k-1)]'Zk=Ek' in a back substitution manner; obtaining all elements at the left side of Zkk according to symmetry. The method has the advantages of simple principle and clear principle, the Z array is solved by the Y array of IEEE-57, IEEE-188 and IEEE-300 node systems, and in comparison with that of an LDU (Logical Data Unit) triangular decomposition method and a method without the normalized Gaussian elimination, the computation speed is greatly improved.

The embodiment of the invention discloses a channel quality indicator determining method and device and a communication device. The method comprises the steps of determining an equivalent channel matrix according to a received channel matrix and a precoding matrix; carrying out noise mean square error normalization on the equivalent channel matrix according to a received channel noise mean square error; carrying out orthogonal triangular decomposition on the normalized equivalent channel matrix and determining a decomposed upper triangular matrix; partitioning the upper triangular matrix according to master diagonal elements and non-master diagonal elements and determining a mutual information (MI) value of each layer; and calculating an average MI value of a code word according to the MI values of the layers and determining a CQI value of the code word based on the average MI value of the code word.

The invention discloses a GIS partial discharge positioning method based on truncated singular value decomposition. The method comprises the following steps of S1, establishing a mathematical model based on time difference of arrival partial discharge positioning; S2, a linear positioning equation set PX=Q being obtained through a spherical surface conversion method; S3, performing centralized pretreatment on the linear positioning equation set PX=Q to obtain a centralized positioning equation set PX = Q'; S4, performing decomposition calculation on the matrix P by applying a principal component triangular decomposition method to obtain an equivalent matrix P'of the matrix P'; S5, the matrix P' being substituted into a positioning equation set PX=Q', and an equivalent linear equation set P'X=Q' being obtained; S6, performing singular value decomposition on the matrix P' to obtain two orthogonal matrixes U=(u1,..., un) and V=(v1,..., vn) of the matrix P', and designing a filtering factor fi; and S7, solving P'X =Q' by applying a truncated singular value decomposition method. The method is reliable and effective in PD signal positioning in the GIS, and is high in positioning precision.

The invention discloses a flood disaster prediction and early warning method based on QR-BC-ELM. The invention discloses a flood early warning method, relates to the technical field of flood early warning, considers various flood causes, establishes a flood model based on an extreme learning machine and a geographic information system to simulate a flood-prone area of the Yellow River basin, and verifies the efficiency and precision advantages of the extreme learning machine relative to an artificial neural network. The learning speed of the improved extreme learning model provided by the invention is 32 times that of an artificial neural network and 1.2 times that of a traditional extreme learning model. Moreover, the noise processing capacities of the proposed orthogonal triangular decomposition extreme learning model and the full-rank decomposition extreme learning model are greatly superior to those of an artificial neural network, and the BC-ELM and the QR-ELM have great advantages in the aspects of prediction precision and prediction efficiency, and are relatively suitable for flood prediction models.