Modified distributed block Jacobi method for block additive unconstrained optimization problems of large systems

In this paper, we propose a modified distributed block Jacobi method for solving block additive unconstrained optimization problems of large practical systems such as electric power systems. Our method has two modifications over conventional block Jacobi method: the first one is the inclusion of a pre-processing step which will greatly improve the computational efficiency; the second modification is the proposed distributed processing step-size determination rule which does not need the knowledge of an unknown asynchronous measure in the distributed environment. We have applied our method to the weighted least square problems of power system state estimation and test the computational efficiency by several cases in a sequential computer and demonstrate the convergence of our method in a real PC-network. In both tests, we obtain very successful results.