101 results about "LU decomposition" patented technology

The invention relates to a GPU (graphic processing unit) based parallel power flow calculation system and method for a large-scale power system. The system comprises a symbol Jacobian matrix forming and decomposing module, an initialization module, a power flow equation right-hand side calculation module, a jacobian matrix assignment module, an LU decomposing module and a forward and backward substitution module; the symbol Jacobian matrix forming and decomposing module is located on a host side, and the host side transmits calculating data to an equipment side; the power flow equation right-hand side calculation module, the jacobian matrix assignment module, the LU decomposing module and the forward and backward substitution module are sequentially connected on the equipment side. The method includes (1) transmitting data needed by calculation to the host side entirely; (2) generating a symbol Jacobian matrix and performing symbol composition on the symbol Jacobian matrix; (3) transmitting a decomposition result by the host side to the equipment side; (4) executing power flow equation right-hand side calculation; (5) executing Jacobian matrix assignment; (6) executing LU decomposition; (7) executing forward and backward substitution.

The invention discloses a high-efficiency time domain electromagnetic simulation method based on an H matrix algorithm, which can realize electromagnetic simulation on a large three-dimensional target. In the method, a time domain finite element method (TDFEM) is used as a background, a low-rank compression technique is used as a core, and a tree structure is used as a basis for carrying out logical unit (LU) decomposition on a sparse matrix generated by the TDFEM by a four arithmetic algorithm corresponding to the H matrix. The acquired upper and lower triangular factors have low-rank compressible characteristics, and the compressed matrix equation can realize quick solution of high-efficiency time domain electromagnetic simulation by the H matrix algorithm. The high-efficiency time domain electromagnetic simulation method has the advantages of fast computation speed, low memory consumption, controllable computation accuracy, good stability and the like, can reduce the complexity of computation to O(Nlog<2>N) and reduce the memory consumption to O(NlogN), can be widely applied to the solution of a large sparse linear system of equations during high-efficiency time domain electromagnetic simulation, and can provide important reference for analyzing the electromagnetic property of the large three-dimensional target.

Aiming at the characteristics that corner-block sparse matrix is capable of parallel computing, the invention provides a parallel LU decomposition for corner sparse matrix based on FPGA. a simulating-sorting module, a symbol resolution module and a parallel numerical LU decomposition module form a three-layer processing platform structure consisting of, wherein the parallel numerical LU decomposition module is used for executing the parallel LU decomposition to matrix data and is positioned at the tail end of a whole processing structure; the symbol decomposition module is used for marking the position of an element to be modified in the matrix and is positioned at the middle end of the whole processing structure; and the simulating-sorting algorithm module is used for determining the elimination order of the matrix and is positioned at the foremost end of the whole processing structure; and the three modules are connected through a shared memory unit. The decomposition has the advantage of executing parallel LU decomposition to the corner sparse matrix in real time, can greatly save development cost as compared with the prior art using parallel computers and distributed computers, and can be applied to the field of real-time electrical network analysis.

The invention relates to a quick LU factorization method for a circuit sparse matrix in circuit simulation, and belongs to the technical field of electronic design automation (EDA). The method comprises the following steps of: symbolic analysis and the calculation of LU factorization, wherein in the step of the symbolic analysis, a preprocessed matrix is analyzed to forecast each row of a non-zero structure of matrixes L and U formed after the matrix is subjected to the LU factorization; and in the step of the calculation of the LU factorization, based on the non-zero structures, obtained by the symbolic analysis, of the matrixes L and U, each row of the matrix is subjected to numerical solving and numerical distribution to obtain the matrixes L and U after the LU factorization. The method is characterized in that the symbolic analysis is separated from the circulation of the LU factorization, and the symbolic analysis is executed beyond the circulation of Newton-Raphson iteration in the process of the circuit simulation, namely is executed once, so the complexity of the circuit simulation can be reduced effectively, and the speed of the LU factorization can be improved to accelerate the integral circuit simulation.

In accordance with a parallel matrix processing method adopted in a shared-memory scalar computer, a matrix to be subjected to LU factorization is divided into a block D of the diagonal portion and blocks beneath the D diagonal block such as L1, L2 and L3. Then, D+L1, D+L2 and D+L3 are assigned to 3 processors respectively for processing them in parallel. Next, a block U is updated by adopting an LU-factorization method and C1 to C3 are updated with L1 to L3 and U. By carrying out this processing on the inner side gradually decreasing in size as blocks, finally, a portion corresponding to the D diagonal block remains to be processed. By applying the LU factorization to this D portion, the LU factorization for the entire matrix can be completed.

The invention discloses a method for calculating stray current and neutral point direct current of transformer caused by stray current in a metro. The method comprises the following steps: 1. The structure of rail-drain network-earth in metro system is equivalent to a three-layer resistance network model, and the resistance network is divided into several sections equally; 2, that flow path of theDC bias magnetic field of the transformer caused by stray current is analyzed, and a current branch of the DC bias magnetic field of the transformer caused by stray current is added into the resistance network model; 3, KVL equations along respective loop are arranged to obtain loop current equations for that built resistance network model; 4, a loop current equation is resolved by using the LU decomposition method, and the stray current of the rail leakage and the neutral point current of the transformer is calculated. This method can be used to calculate the distribution curves of neutral current and stray current of transformer under subway operation, which provides a theoretical basis for evaluating the influence of stray current on DC magnetic bias of transformer.

The invention provides a sparse matrix LU decomposition method based on a GPU. The sparse matrix LU decomposition method includes the following steps: A, building a to-be-simulated circuit matrix of a circuit network on a CPU, and performing preprocessing and rarefaction on the circuit matrix; B, selecting a computing platform for processing a sparse matrix LU according to preprocessing results; C, performing decomposition on the sparse matrix LU through the GPU if the computing platform which is used is the GPU; D, performing back substitution solution on values of non-zero elements in the sparse matrix LU through the CPU so as to complete decomposition of the sparse matrix LU. According to the sparse matrix LU decomposition method, the computing platform can be selected automatically according to the number of floating-point operation times in decomposition, the task on the CPU and the task on the GPU can be distributed reasonably, coordinated operation of a plurality of threads on the GPU is guaranteed, and therefore the computing speed is improved and circuit simulation time is shortened.

A pseudo-spatial-average-covariance-matrix generating unit selects, from matrixes R_f1, R_f2, R_b1 and R_b2, one appropriate matrix or two or more appropriate matrixes for combination to generate a pseudo-spatial-average-covariance matrix R. A pseudo-spatial-average-covariance-matrix Hermitian-conjugate product generating unit generates a target-count-estimation matrix RR^H. A target-count-estimation-matrix decomposing unit performs LU decomposition on RR^H into a lower triangular matrix L and an upper triangular matrix U. An index generating unit generates an index using elements of the upper triangular matrix U. An index-parameter scanning unit estimates a target count by using the index generated by the index generating unit.

A method for determining harmonic voltage comprises the steps of obtaining the fundamental wave voltage and the injection electric quantity of each of the nodes, and determining positive-sequence, negative-sequence and zero-sequence current of each of the nodes; according to the fundamental wave voltage and positive-sequence, negative-sequence and zero-sequence current of each of the nodes, determining positive-sequence, negative-sequence and zero-sequence admittance matrixes of each of the nodes; numbering each of the nodes with a node numbering method, and conducting factorization on positive-sequence, negative-sequence and zero-sequence admittance matrixes with an LU decomposition according to the numbering sequence; and using harmonic power flow models to determine harmonic impedance of each of the branches, determining positive-sequence harmonic voltage, negative-sequence harmonic voltage and zero-sequence harmonic voltage of each of the nodes according to the harmonic impedance of branches and positive-sequence, negative-sequence and zero-sequence admittance matrixes after factorization, and determining the harmonic voltage of each of the node according to the positive-sequence harmonic voltage, the negative-sequence harmonic voltage and the zero-sequence harmonic voltage. The invention also provides a corresponding system. By the aid of the method and the system, the calculation precision can be improved, and the calculation can be simplified.

The solution program causes a computer to execute step 1 for performing, in parallel, the LU decomposition on the column block comprising a plurality of columns and for storing the result in the working area; step 2 for canceling the row permutation on the left of each column to the result in step 1 and for copying back the result in the array in compress mode; step 3 for saving the portion which may be damaged by updating the band matrix corresponding to the result in step 1; step 4 for updating the band matrix in parallel by using the result in step 1; and step 5 for returning the saved portion in step 3 to the result in step 4.

The encoding method for LDPC code based on LU decomposition for optimized search matrix is realized by software pretreatment and hardware encoding. Wherein, eliminating correlative rows of check matrix, realigning the matrix; selecting nonsingular matrix with non-correlative column composite algorism for decomposition with simple LU algorism; optimizing search by diagonal element asymptotic one of nonsingular matrix; finally, selecting nonsingular matrix with lowest density, and reducing delay of whole coding system by time interval control parallel process hardware structure. This invention holds advantages in prior art, reduces calculation amount near Greedy algorism, and fit to any LDPC code special to abnormal code and normal node with correlative rows.

The invention discloses a rapid decomposition method trend calculating method based on a Matlab. The method comprises the following steps that matrix operation and complex number operation are used; an array bt1 is set and a node number of a non-equilibrium node is recorded, and an array bt2 is set and a node number of a PQ node is recorded; when equation set coefficient matrixes B' and B'' are formed, a node type is not considered, and then according to the arrays bt1 and bt2, a matrix element is extracted, and redundant rows and columns are removed; and before iteration, according to the arrays bt1 and bt2, node-injected active and reactive independent elements are removed. In the invention, program codes are reduced, programming is simplified, and a program is clear so that scientific research personnel can conveniently modify the program, debug and improve the program, and add a new function. In the invention, during LU decomposition, an independent element of a coefficient matrix is removed, and before iteration, node-injected active and reactive independent elements are removed; and during an iteration process, work of mentioning a correlation matrix or a vector element repeatedly can be avoided, calculating workloads are reduced and a calculating speed is increased.

The invention discloses a Spark-based distributed matrix inversion parallel operation method. The method comprises the following steps: carrying out parallel LU decomposition operation on an input matrix in an iteration process; taking a distributed upper triangular matrix and a distributed lower triangular matrix obtained through the LU composition as basis so as to the inverse matrixes of the distributed upper triangular matrix and the distributed lower triangular matrix by using a recursive algorithm; and finally taking a permutation matrix and the inverse matrixes of the triangular matrixes obtained in the above two steps as basis so as to implement distributed matrix multiplication to obtain the inverse matrix of any original input matrix. According to the method, the dense matrixes with large dimensionalities can be processed, and relatively high operation efficiency as well as relatively good fault tolerance and expandability can be obtained.

The invention discloses a method of column pivoting LU decomposition based on FPGA, the technical problem to be solved is to reduce time complexity of LU decomposition and accelerate solving of dense matrix linear equation system. The technical solution comprises that: at first a FPGA-based master-slave parallel computation system composed of a master processing unit and P slave processing units is constructed, the master-slave parallel computation system carries out column pivoting LU decomposition on the matrix to be decomposed, the master processing unit sends parameters and the matrix to be decomposed to a first slave processing unit, the slave processing unit processes the matrix in a pipeline way, and the processing result is sent to the master processing unit from the P slave processing unit. Computation accuracy is dramatically improved in contrast to non-pivoting LU decomposition with the invention being used, computing speed is high, and the master-slave parallel computation system thereof has expandability.

The present invention proposes a magnetotelluric meshless numerical simulation method for a random conductive medium model. The method is for heterogeneity of underground space, and according to the method, a shape function is constructed based on discrete nodes, and dependence on a mesh no longer exists. The method provided by the present invention comprises: using the spectral decomposition theory and the hybrid autocorrelation function theory of the stochastic modeling process to construct a random conductive medium model of underground space; establishing a function of the magnetotelluric corresponding boundary value problem; establishing a shape function of a meshless method by using a least squares method; using the Lagrange multipliers method to load essential boundary conditions; resolving linear equations by using the biconjugate gradient stabilized (BICGSTAB) algorithm of incomplete LU decomposition preprocessing, to obtain a field value of each node in the region; and using the field values to calculate the apparent resistivity and phase of each point, thereby completing the whole numerical simulation calculation process. The section calculated according to values of the random conductive medium model better complies with the actual situation, and is more advantageous for the interpretation and processing work of the magnetotelluric data.

The invention discloses an LDPC (Low Density Parity Check) encoding method based on FPGA (Field Programmable Gate Array) and used in CMMB (China Mobile Multimedia Broadcasting) exciter. The method comprises the following steps: firstly, processing a verification matrix H of an LDPC system code on an MATLAB (Matrix Laboratory) platform, so as to generate verification matrixes Hp and Hs which correspond to a verification bit column vector P and an information bit column vector S; carrying out LU (Logical Unit) decomposition to the verification matrixes so as to obtain a lower triangular matrix L and an upper triangular matrix U; and realizing the LDPC encoding on the FPGA platform, which mainly involves the storage of a large matrix, matrix multiplication, forward iteration and backward iteration. According to the encoding method, an encoding mode based on a LU decomposition verification matrix is adopted; the logic calculation that large matrixes are multiplied is avoided; and the problem of large requirement on FPGA internal storage resource caused by large data quantity storage is solved, thereby simplifying the logic calculation operation, saving the storage space, and being beneficial for the realization of the LDPC encoding of the CMMB system.

The invention provides a Ka frequency range deep space communication method and system of a q-LDPC-LT cascade fountain code. The method comprises that an LT encoder encodes original data to generate first data; a q-LDPC encoder encodes the first data to generate second data using a low density sparse parity check matrix construction method with an LU decomposition encoding algorithm; a q-LDPC decoder decodes the second data to generate third data; the second data is decoded to generate the third data, wherein error code detection is performed after bit decision of each coding iteration and error bits are searched and corrected; an LT decoder decodes the second data to generate decoding data; and the second data is decoded to generate the decoding data, wherein when there is no encoding data package with the degree of 1, a data package with the degree of 1 is constructed to perform decoding. The invention can reduce the encoding complexity, improve the encoding efficiency, and further improving the decoding reliability, thereby reducing dependence of decoding on the existence of an encoding data package with the degree of 1.

Disclosed is a design method of a real-time simulation solver of an active power distribution network on basis of an FPGA. The method comprises the steps of offline preprocessing and online solving. During offline preprocessing, a column approximate minimum degree algorithm is adopted for processing a node admittance matrix; the address vectors of current and the address vectors of current voltageare prestored; an LU decomposition method is adopted for decomposing the node admittance matrix; the starting time of each sub-task in the previous substitution process and the back substitution process is prestored. During online solving, the simulation starting time is set; the simulation time is advanced by one step length; information initialization of each sub-task and addressing sorting ofthe current vectors are completed; when solving begins, a starting timer of each processing unit is initialized according to the prestored starting time of the corresponding sub-task; solving in the previous substitution process is completed, and an intermediate solution vector is generated; solving in the back substitution process is completed, and a final solution vector is generated after addressing sorting; whether or not the simulation time reaches the final simulation moment is determined. By means of the method, while the solving precision and speed are ensured, sparse linear equationsare solved accurately and efficiently.

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 sparse node positioning algorithm. According to the algorithm, through a grid sensing area, a node positioning problem is transformed into a sparse signal reconstruction problem. Then a pretreatment method of LU decomposition is employed, the preprocessing of an observation matrix is carried out, and a restricted isometry condition is effectively satisfied. Finally, aiming at a problem that a determined sparse signal is an approximate sparse signal in a sparse positioning model, a centroid algorithm is employed to improve the positioning performance of the algorithm. According to the algorithm, the compressed sensing theory is introduced, through the grid sensing area, the node positioning problem is effectively converted into an N-dimensional vector reconstruction problem with sparsity of 1, and the characteristics of a node is effectively excavated to complete the node self-positioning.

The invention belongs to the field of matrix computing technology, and particularly relates to a scheduling method of a reconfigurable computing architecture based on LU decomposition of an arbitrary dimension matrix. The reconfigurable computing architecture based on LU decomposition of the arbitrary dimension matrix consists of processing units, an interface controller, a scheduling module and a storage module. The method comprises the following steps: interconnecting the processing units, and forming a reconfigurable processing array with the interface controller; enabling the scheduling module to design a scheduling mechanism, generating configuration information, and issuing the configuration information to the reconfigurable processing array; constructing a computing architecture that meets the LU decomposition of a current dimension matrix; based on the constructed computing architecture, distributing computing data to the processing units according to the scheduling mechanism, and performing LU decomposition computation; and returning an LU decomposition computation result to the storage module through the interface controller. According to the scheduling method disclosed by the invention, the LU decomposition computation of the arbitrary dimension matrix can be achieved through the mode that the fixed processing units can be reconfigured, and the flexibility of LU decomposition can be improved.

The invention discloses a multi-phase structure of a two-channel orthogonal mirror image filter bank and a coefficient design method of thereof. At present, most known two-channel QFMB design methodsare optimization design of prototype filter coefficients, and do not pay much attention to reduction of hardware implementation complexity of a QMFB structure. The multi-phase structure of the two-channel orthogonal mirror image filter bank comprises an H0 module, an M (z) module, q modules, a module connecting part, a main input port Input, a main output port Output0 and a main output port Output1. According to the multi-phase structure of the two-channel QMFB analysis filter bank part provided by the invention, a traditional QMFB prototype filter coefficient is converted into a form of an extrapolation pulse response filter, and LU decomposition is adopted, so that more 0 and 1 exist in the coefficient, and multiplier and adder resources are saved.