论文部分内容阅读
在金属疲劳试验中,有时会出现一个或多个明显小于或大于其它数据的异常数据。查找导致这些异常数据的原因是很困难的,难以直接剔除。通常,相关文献只简要地提供基于统计学的识别异常数据的判据。为了帮助数据分析人员更好地理解这些统计判别方法,在处理金属疲劳试验数据时更好地运用统计判别方法,对分布于各文献中的统计判别法进行了归纳整理,并给出必要的推导过程和公式,最后对所列统计判别法进行总结,介绍各方法的优势和不足,并给出建议。
In metal fatigue tests, one or more anomalies that are significantly smaller or larger than other data sometimes appear. Find the cause of these abnormal data is very difficult, it is difficult to directly remove. In general, the relevant literature provides only a brief summary of the criteria for identifying abnormal data based on statistics. To help data analysts better understand these statistical discriminant methods, the statistical discriminant method is better used in dealing with metal fatigue test data, and the statistical discriminant methods distributed in various literatures are summarized and the necessary deductions are given Process and formula, and finally summarize the statistical discriminant method listed, introduce the advantages and disadvantages of each method, and give suggestions.