大系统平衡块对角优分解

Balanced Block Diagonally Dominant Decomposition of Large-Scale Systems

为了实现大系统的并行处理,大系统的分解总是希望分配给各处理器的任务量是均衡的,即分解后的子系统要有平衡的结构(或维数尽量接近)。因此,在系统平衡块三角分解的基础上,利用非平衡补偿方法使系统状态方程的系数矩阵具有块对角优型,使系统模型简化的结果更为理想。系统平衡块对角优分解方法通过实例仿真证明是可行的,并为大系统分散控制的分析与设计提供了新的系统模型简化的手段。

For a parallel processes implementation of large-scale systems, a balance of the processor burdens distributed by the decomposition of large-scale systems is always hoped, that is, the subsystems should have balanced structures after decompositions (or the dimensions of subsystems are similar). Based on the system balanced block triangle decomposition, a coefficient matrix of system state equations can be of block diagonally dominant by using imbalanced compensating scheme; and therefore, the result of system model simplifications is ideal. It is proved that the given balanced block diagonally dominant decomposition of systems is feasible by the simulations, which presents a new system model simplifying method for the analysis and design of large-scale system decentralized controls.