A limitation faced by the utilities when monitoring the voltages and power flow, i.e. its state, in their distribution system is that metering is expensive to install and maintain. Thus, mathematical methods to calculate the state of the network from minimal measurements are highly desirable. This paper presents a modification of the traditional technique to estimate the static state of a power system, improving its robustness to measurement deficiency. A significant improvement to the traditional technique it the elimination of the requirement that the monitored power system's state must be fully observable. The technique is applied in two stages. Firstly, a Newton iteration, which uses singular value decomposition to generate the Moore-Penrose inverse of the input's Jacobian matrix. Secondly, an analytic sensitivity analysis determining the observability of variables calculated from the partially observable system state.
展开▼
