论文部分内容阅读
电能计量设备故障率分析机制是电能准确计量的前提, 而通常计量设备故障率数据样本量少且数据缺失严重, 难以对故障率数据进行准确评估分析.为此, 提出一种基于多层贝叶斯B样条 (MBBS) 的电能计量设备故障率可靠性评估与分析模型:首先采用Z分数法检测原始故障率数据中的异常值;然后采用基于Weibull分布的MBBS模型对电能计量设备故障率数据进行拟合评估, 其中对B样条基函数的样条系数进行改进, 采用一阶自回归作为先验分布, 以平滑拟合曲线;利用马尔科夫蒙特卡洛方法求解模型, 获得后验参数及置信区间估计.通过工程实例进行验证, 结果表明采用本方法能评估电能计量设备故障率随时间的变化趋势, 兼顾数据整体变化的渐变规律, 并准确分析出可靠度.“,”The fault rate analysis mechanism of power metering equipment is the premise of accurate measurement of electric energy. However, it is usually difficult to accurately evaluate and analyze the failure rate data due to the small sample size and serious data missing. Therefore, a reliability evaluation and analysis model of power metering equipment failure rate is proposed based on multilayered Bayesian B-spline (MBBS). Firstly, the Z-score method is used to detect the abnormal data in the raw failure rate data. Then, an multilayered Bayesian cubic B-spline model based on Weibull distribution is applied to evaluate the failure rate data of the electricity meter. To smooth the fitting curve, the first-order autoregressive distribution is adopted to the spline coefficients of B-spline basis function. The posterior parameters and confidence interval estimation are obtained by Markov Monte Carlo method. Finally, the verification on the engineering case shows that the proposed method can evaluate the variation trend over time of the power metering equipment failure rate, reflect the he gradual change law of the data, and accurately analyze the reliability.