Factor-space dimension reduction for sensitivity analysis of intelligent electronic devices

Implementation of uncertainty and sensitivity analysis for testing measurement accuracy of a commercial Intelligent Electronic Device (IED) requires unaffordable number of an IED tests. The main reason is a large number of factors (i.e. high dimension of factor space) that potentially affects the accuracy. The factor values are sampled within practical domains by using Monte Carlo simulation to produce test cases. Therefore for high-dimensional factor space, a number of test cases is prohibitively large, making the systematic uncertainty and sensitivity analysis impossible in practice. To deal with this limitation, we propose in this paper an application of Morris factor screening algorithm for the factor space dimension reduction. The screening method identifies the input factors that have a negligible impact on the measurement accuracy. The measurement accuracy is tested for transient condition by injecting signals having step-change. The accuracy is quantified using two performance indices: overshoot and steady state error. Factors selected for this study model deviation from the perfect step-change that are possible during faults on transmission lines. The proposed methodology is demonstrated by testing Cosine filter, which is commonly implemented in IEDs. In addition, we have tested a commercial IED using injection test set. It should be emphasized that although measurement algorithm for a commercial IED is not fully documented, we are able, through a number of systematically designed tests, to describe performance of the measurement algorithm for a complete range of practical cases.