Identification of measurement variances based on test equations using ANN

The paper addresses methods for the identification of random error variances in measurements. Consideration is given to the algorithms of identifying variances using a numerical method and ANN. Numerical methods, based on the calculation of discrepancy correlation in a pair of test equations with a common measurement, are presented. Application of ANN allows identification of variances early at the stages of implementing the algorithms of state estimation for the larger part of the existing measurements as well as identification of measurements variances involved in one test equation only. Recommendations on the application of the two offered algorithms for electric power system state estimation under real conditions are given based on their study and comparison