Serious problem for failure probability evaluation of equipment due to voltage sag is the uncertainties and mathematical expression of influencing factors. The uncertainties contained in voltage sag, equipment voltage tolerance and possible operation state are researched. The intension and extension uncertainties of influencing factors are used to distinguish their property. Stochastic and fuzzy variables are introduced to express intension and extension uncertainties.The mathematical models of influencing factors are established using stochastic and fuzzy models and a multi-uncertainty evaluation model is proposed also. In this method, maximum entropy principle is used to extract the probability distribution of voltage sag intensity and the determination principle of membership function of fuzzy safety event is used to determine the multi-uncertainty evaluation model. As a case study, personal computer is simulated. The simulation results show that this method is correct and viable and it can be easily used in other fields. Keywordsvoltage sag; sensitive equipment; failure probability; multiuncertainty; intension and extension uncertainty; mathematical model; maximum entropy principle; fuzzy safety eventI.

There is complex uncertainty involved in evaluation of equipment sensitivity to voltage sags, and that makes it difficult for the point-valued assessment method to evaluate the impact of voltage sags on the sensitive equipment, especially in the case of small sample size. Considering the multi-valued nature of equipment status during the voltage sags, the voltage tolerance level of sensitive equipment is characterized by interval data, so as to derive interval probabilities of equipment operation status. Meanwhile, hybrid entropy is then used to measure the uncertainty of randomness and fuzziness in equipment voltage tolerance level. It develops one novel model by transforming the interval probability into point-valued probability based on maximum hybrid entropy model. By the help of personal computers, the proposed model is finally verified in a test distribution system.

No abstract