Reliability data calibration based on load point interruption indices using nonlinear and quadratic optimization

The predictive reliability assessment estimates the future performance of distribution networks, regarding to interruptions of power supply, based on components reliability data and system topology. The indices estimated by predictive models must match with their respective historical values otherwise the credibility of the model is compromised. An alternative to increase the accuracy of the predictive reliability models is to set reliability components data (failure rates and repair times) so that the estimated indices are near to their historical values. This process to adjust reliability data is called calibration. In this paper, nonlinear and quadratic optimization models are proposed to calibrate failure rates and repair times, respectively, of power distribution equipments. Both models are based on the load point reliability indices. The test results in a large scale distribution networks demonstrate that the proposed models can significantly reduce the mismatches between historical and predicted reliability indices.