基于Moffat休息型HMM退化过程的220 kV断路器可靠性预测

Reliability Prediction of 220 kV Circuit Breakers Based on Moffat Rest Hidden Markov Degradation Process

针对220 kV断路器状态评估参量获取困难,以及计划检修和突发性维修手段造成的断路器前期维修过剩、后期维修不足的问题,分析了断路器的温度与状态之间的关系,以及其自身性能衰退的过程,并考虑人为修缮因素,构建了一种以温度为状态评估参量、以服役时间和状态检修为模型修正参量、基于Moffat休息型隐马尔科夫(hidden Markov model, HMM)退化过程的220 kV断路器可靠性预测模型。该模型阐述了设备健康状态在其全寿命周期的"Moffat休息"式退化规律,在此基础上,通过分析红外检测图像与可见光检测图像融合得到的多个关键构件温度,评估了当前运行状态的可靠性,并预测了后续趋势。实验结果表明:设备状态周期性修复抑制了时变状态转移矩阵的退化速度,通过分析关键构件的温度,状态评估结果滞后于实际情况的误差得到了改善,可用系数、停运率的评估结果与实际统计数据相比,误差分别缩减至0.044%、0.032%。研究结果表明,采用所提方法得到的断路器可靠性预测结果更加符合设备退化规律。

It is difficult to obtain all the parameters for evaluating 220 kV circuit breakers. And the excessive maintenance of circuit breakers after installation and the insufficient maintenance afterwards caused by planned overhauls and sudden maintenance methods are also observed. To deal with the problems, the relationship between the temperature and state of a circuit breaker, and the circuit breaker’s degradation process with artificial repairs, are discussed in this paper. A reliability prediction model for 220 kV circuit breaker based on the Moffat resting hidden Markov model(HMM) degradation process is built, with temperature as the status evaluation parameter, and service time and status maintenance as the model modification parameters. The model describes the "Moffat rest" degradation of the health state of a circuit breaker during its entire life cycle. On this basis, the temperature of multiple key components obtained by fusing infrared detection images and visible light detection images is analyzed to evaluate and predict operating reliability of the circuit breaker. The experimental results show that the periodic repair of equipment suppresses the degradation of the time-varying state transition matrix. By analyzing the temperature of key components, the error of the state evaluation lagging behind the actual situation is decreased. Comparing the evaluations of available coefficient and outage rate with the actual data, the errors are reduced to 0.044% and 0.032% respectively. The results show that, the obtained reliability prediction of circuit breakers by using the proposed method is more in line with the law of equipment degradation.