Application of regression analysis for predication of voltage collapse in power systems

In this paper we present a new application of the Least Error Square (LES) Estimation algorithm for the predication of voltage collapse in a power system using the local measurements of voltage and current of a load bus. Using these measurements a polynomial of order n is assumed for the relation of the load voltage and load current, and hence the kVA. The coefficients of this polynomial are estimated using the LES estimation algorithm. The collapse point is defined as the point where the load draws its maximum volt-ampere from the bus. Having obtained this point the estimated voltage can be obtained using the assumed polynomial. This method of prediction of voltage collapse supercedes the conventional load flow methods by avoiding repeated load flows. The proposed algorithm is tested using the IEEE-30 bus system and compared with the conventional load flow methods.