Nonlinear multisplitting method is known as parallel iterative methods for solving a large-scale system of nonlinear equations F(x) = 0. We extend the idea of nonlinear multisplitting and consider a new model ill whic...Nonlinear multisplitting method is known as parallel iterative methods for solving a large-scale system of nonlinear equations F(x) = 0. We extend the idea of nonlinear multisplitting and consider a new model ill which the iteration is executed asynchronously: Each processor calculate the solution of an individual nonlinear system belong to its nonlinear multisplitting and can update the global approximation residing in the shared memory at any time. A local convergence analysis of this model is presented. Finally, we give a uumerical example which shows a 'strange' property that speedup Sp > p and efficiency Ep > 1.展开更多
Based on the hydrodynamic stability theory of distorted laminar flow and the kind of distortion profiles on the mean velocity in parallel shear flow given in paper [1], this paper investigates the nonlinear stability ...Based on the hydrodynamic stability theory of distorted laminar flow and the kind of distortion profiles on the mean velocity in parallel shear flow given in paper [1], this paper investigates the nonlinear stability behaviour of parallel shear flow, carries on stability calculation taking account of the perturbations of background turbulence noise under certain assumption, and obtains some results in accordance qualitatively with those of experiment for plane Poiseuille flow and pipe Poiseuille flow.展开更多
The boundary value problem for nonlinear parabolic system is solved by the finite difference method with intrinsic parallelism. The existence of the discrete vector solution for the general finite difference schemes w...The boundary value problem for nonlinear parabolic system is solved by the finite difference method with intrinsic parallelism. The existence of the discrete vector solution for the general finite difference schemes with intrinsic parallelism is proved by the fixed-point technique in finite-dimensional Euclidean space. The convergence and stability theorems of the discrete vector solutions of the nonlinear difference system with intrinsic parallelism are proved. The limitation vector function is just the unique generalized solution of the original problem for the parabolic system.展开更多
虚拟阻抗控制在并联逆变器控制中得到了广泛应用,但是会带来电压降落的问题,并且虚拟阻抗法需要对输出电流求导,使得系统对输出电流噪声和非线性负载很敏感。为解决此问题,提出一种基于双重二阶广义积分器(dual-second order general in...虚拟阻抗控制在并联逆变器控制中得到了广泛应用,但是会带来电压降落的问题,并且虚拟阻抗法需要对输出电流求导,使得系统对输出电流噪声和非线性负载很敏感。为解决此问题,提出一种基于双重二阶广义积分器(dual-second order general integrator,DSOGI)的虚拟阻抗模型,该模型避免了对输出电流求导,降低了对输出电流的敏感度,从而较好地抑制了输出电流中的噪声,并降低了非线性负载对系统的影响;同时可减少电压降落和抑制系统间的环流,改善了系统的电能质量。对逆变器并联控制系统进行了应用对比分析,结果验证了基于DSOGI虚拟阻抗模型的有效性。展开更多
Fuzzy systems can be used to identify nonlinear dynamic systems in two modes. One is series-parallel modeand the other is parallel mode. The prediction and identification algorithm convergence of nonlinear dynamic sys...Fuzzy systems can be used to identify nonlinear dynamic systems in two modes. One is series-parallel modeand the other is parallel mode. The prediction and identification algorithm convergence of nonlinear dynamic system i-dentification using parallel fuzzy systems is discussed in this paper. It is proved that as long as the parameters of par-allel fuzzy systems meet some prerequisites, the parallel prediction procedure converges and the parallel identificationalgorithm locally converges. This conclusion has instructive significance for parallel fuzzy systems' application.展开更多
文摘Nonlinear multisplitting method is known as parallel iterative methods for solving a large-scale system of nonlinear equations F(x) = 0. We extend the idea of nonlinear multisplitting and consider a new model ill which the iteration is executed asynchronously: Each processor calculate the solution of an individual nonlinear system belong to its nonlinear multisplitting and can update the global approximation residing in the shared memory at any time. A local convergence analysis of this model is presented. Finally, we give a uumerical example which shows a 'strange' property that speedup Sp > p and efficiency Ep > 1.
文摘Based on the hydrodynamic stability theory of distorted laminar flow and the kind of distortion profiles on the mean velocity in parallel shear flow given in paper [1], this paper investigates the nonlinear stability behaviour of parallel shear flow, carries on stability calculation taking account of the perturbations of background turbulence noise under certain assumption, and obtains some results in accordance qualitatively with those of experiment for plane Poiseuille flow and pipe Poiseuille flow.
基金Project supported by the National Natural Science Foundation of China and the Foundation of CAEP.
文摘The boundary value problem for nonlinear parabolic system is solved by the finite difference method with intrinsic parallelism. The existence of the discrete vector solution for the general finite difference schemes with intrinsic parallelism is proved by the fixed-point technique in finite-dimensional Euclidean space. The convergence and stability theorems of the discrete vector solutions of the nonlinear difference system with intrinsic parallelism are proved. The limitation vector function is just the unique generalized solution of the original problem for the parabolic system.
文摘虚拟阻抗控制在并联逆变器控制中得到了广泛应用,但是会带来电压降落的问题,并且虚拟阻抗法需要对输出电流求导,使得系统对输出电流噪声和非线性负载很敏感。为解决此问题,提出一种基于双重二阶广义积分器(dual-second order general integrator,DSOGI)的虚拟阻抗模型,该模型避免了对输出电流求导,降低了对输出电流的敏感度,从而较好地抑制了输出电流中的噪声,并降低了非线性负载对系统的影响;同时可减少电压降落和抑制系统间的环流,改善了系统的电能质量。对逆变器并联控制系统进行了应用对比分析,结果验证了基于DSOGI虚拟阻抗模型的有效性。
文摘Fuzzy systems can be used to identify nonlinear dynamic systems in two modes. One is series-parallel modeand the other is parallel mode. The prediction and identification algorithm convergence of nonlinear dynamic system i-dentification using parallel fuzzy systems is discussed in this paper. It is proved that as long as the parameters of par-allel fuzzy systems meet some prerequisites, the parallel prediction procedure converges and the parallel identificationalgorithm locally converges. This conclusion has instructive significance for parallel fuzzy systems' application.
