Abstract:

According to the failure site data of the well pumping unit in active service,its life statistical model is determined and the service reliability is evaluated quantitatively.The one-shat verification of the mufti-spare se(ected hypothesis distribution for the well pumping unit life statistical made could be realized by using the fuzzy pattern recognition principle and the maximum total mean subordinate superiority standard.The method demonstrate that the verification result is only depend on the level of the total mean subordination degree but not the leveC of the significance.The defect existed in the classics statistic distribution analysis that the mufti-hypothesis distribution in same significance level could be accepted is surmounted.The efficiency of the hypothesis verification for the well pumping unit statistic madel is improved and the calculation of the using reliability index for the well pumping unit in active service is more accurate and credible by use the method in this paper.

Key words: well pumping unit, life, fault, reliability, fuzzy pattern recognition, statistical analysis

摘要：

由在役抽油机故障的现场数据来判定其寿命统计模型,实现对抽油机使用可靠性的定量评定.根据模糊模式识别原理,采用最大总体平均隶属优势准则,可实现对抽油机寿命统计模型的多个备择假设分布进行一次性检验.这种方法表明,检验结果不依赖于显着性水平的高低,只取决于总体平均隶属度的大小.从而克服了在经典统计分析中存在的在同一显着性水平下,可同时接受多个假设分布而难以区分的缺点,提高了对抽油机统计模型假设检验的功效,使在役抽油机使用可靠性指标的计算更加准确、可信.

关键词: 抽油机, 寿命, 故障, 可靠性, 模糊识别, 统计分析