摘要
统计过程控制是保证现代生产制造业产品或服务质量水平的重要工具。传统的统计过程控制图对于监控过程出现大波动有很好的效果,但是对于过程的小波动并不灵敏。累积和控制图对过程出现的小偏移应用累积放大原理,能够及时的发现偏移并产生报警,避免产生严重的后果。本文对累积和控制图的原理和实现工具做了说明。给出了计算机实现累积和控制图比较完整的设计方案,此方案可以完成对统计过程控制系统的搭建,实现对过程的监控预警。
Statistical process control is one of the key factors for the products and services in manufacturing enterprise. Traditional statistical process control chart performs well when fluctuations in the monitoring system are great. However, when fluctuations are small, traditional statistical process control charts are in general not sensitive enough. Cumulative sum control chart（CUSUM）, which sums up all small drifts, can detect shifts and report events much earlier than those traditional statistical process control charts and therefore is helpful to avoid huge lost. In this paper, we first introduce CUSUM and its implementation. Besides, we also provide an algorithm that utilizes CUSUM method to detect anomalies in a statistical process control system.
基金
国家科技重大专项课题(2011ZX02507-006)资助
