Neural networks and pseudo-measurements for real-time monitoring of distribution systems

A state estimation scheme for power distribution system, based on Artificial Neural Networks, is proposed. It allows the quantities which describe the distribution system operations to be identified on-line, and also in the presence of measurement uncertainties in the input data. Details of the design and optimization of such a neural scheme are discussed. The analysis of the performance for the state estimation scheme for a power distribution system is reported; the accuracy obtainable with such neural pseudo-instruments, the capability to measure outwith the operating range, and also in the presence of increased measurement uncertainties, are highlighted