MCHSE: a versatile framework for the design of two-level power system state estimators

A new method for the design of static WLS state-estimators on large-scale power systems is described. The approach is based on a simple partitioning, termed `α-decomposition´ of the measurement model, and the estimator is implemented on two levels within a multiple-processor environment. The lower level computation, executed within subsystem processors operating in parallel, is noniterative and of reduced order; the upper level algorithm, also noniterative and of low order, is executed in a single processor. The solution is obtained from alternate entries to the lower and upper levels, continued to a convergence. Necessary and sufficient conditions for convergence are derived, and lead to new definitions for observability on decomposed power systems. Tests for `α-observability´ may be distributed amongst the processors on two levels. The method is optimal and structurally efficient, and convergence is fast (theoretically quadratic). The structure of the estimator is simple and well-defined, and possibilities for suboptimal variants are suggested