石化安全设备寿命参数极大似然估计

Maximum likelihood estimation of petrochemical safety equipment lifetime parameters

为弥补中值方法在评估区间截尾数据的寿命分布参数时不够准确的缺点,同时方便工程实际应用,提出一种分位数填充算法。首先使用分位数填充算法将区间截尾数据转化为虚拟完全数据,然后将虚拟完全数据用于极大似然评估(MLE)中,最后以蜡油加氢装置紧急泄压阀为案例,通过比较小样本数据量评估结果来验证该算法的有效性。结果表明:在小样本下该算法具有很好的稳定性,且经过5次以内的迭代即可收敛;利用该算法得到的寿命分布参数的准确度比中值方法提升7. 5%,有助于制定有效的维护策略以保证安全设备的机械完整性,保证石化装置的安全运行。

In order to compensate for the shortcomings of the mid-point method in evaluating the lifetime distribution parameters of interval censored data,and make it easy to use in engineering practice,a quantile-filling algorithm was worked out,which can convert the interval censored data into virtual complete data,and the obtained virtual complete data will be used in MLE to obtain parameter estimation of the probability distribution of equipment lifetime.In the case of the emergency relief valve of the wax oil hydrogenation unit,the use and performance of the algorithm were verified by comparing the estimation results of the 50-100 data volume.The results show that the algorithm has good stability under small sample,that it can converge within 5 times of iteration,the accuracy of life distribution parameters obtained by this algorithm is 7.5%higher than that by the mid-point method,which helps to develop effective maintenance strategy to ensure the mechanical integrity of the safety equipment,and then ensure the safe operation of the petrochemical plant.