摘要
在企业实际中,存在大量非正态过程。此时,如果直接使用正态技术下的控制图监控过程,将会导致严重的错误。以抽样百分位点选择约翰逊分布函数来拟合非正态数据为基础,分析了许多最佳拟合Johnson曲线并转换为正态分布的方法。然后,提出了最佳拟合对应的最适宜z值的选择程序,以及在这种转换下的控制图技术。案例研究验证了该方法的有效性和实用性。
In practice, there are moments of processes with non-normal distribution. If the normal-based techniques are used on these processes, serious errors can result. In this paper, the methods of fitting non-normal data by Johnson curve family are discussed. Then, in conjunction with the sample percentile, the procedure for selecting the most suitable z in best-fit distribution and the control chart based on this transformation are researched. Finally, the case study verifies that the methods researched are valid and practical.
出处
《系统工程理论方法应用》
2004年第4期372-376,共5页
Systems Engineering Theory·Methodology·Applications
基金
国家自然科学基金资助项目(70002007)
航空基金资助项目(00J55005)