役龄累积和状态诱因协同的变压器失效概率模型

Predicting the failure probability of transformers is a key to achieve the quantitative analysis on operation risk of targeted transformers. This paper analyzed the development law of failure probability of transformers from two aspects, i.e., increasing service years and operating state inducements, aiming at merging development trend and differences of failure probability of transformers. We introduced service years characterizing accumulation of fatigue aging into the benchmark function of the proportional failure probability model by means of Weibull function, which represents the development law of failure probability depending on the "bathtub curve". The operating state, describing the degree of device defects, was regarded as a parameter in the exponential connection function of the proportional failure probability model, which reflected the impact of health state on the failure probability. The Levenberg-Marquardt algorithm and the inversion algorithm were combined to estimate the model parameter by using the linear regression theory. Finally, a failure probability model of transformers merging increasing service years and operating state inducement has been proposed. We verified the model using a set of real statistic data. It is expected that the model could provide a new idea for transformer operating risk management.