粗差对数学期望值计算有偏态影响,并会传递到残差序列中,仅用3σ准则探测粗差,极易产生"去真"或"纳伪"的错误。通过分析观测序列极差、观测值改正数极差与粗差之间的关系,推导了粗差的判断条件,构造双极差判断矩阵进行粗差探测,同时比较了小样本和大样本序列探测的不同之处,通过实例分析,验证了该方法是一种有效的探测手段。
Gross errors make skewness influence on mathematical expectation of series, which will transfer to the residual sequence. The gross errors detection only with 3cr criterion may make a mistake of discarding the truth or accepting the wrong. The relationship between the poor of observations or error correction and gross errors was analyzed and the formulae to judge gross errors were deduced. The judgment condition matrix based on bio-poor for detecting gross errors was constructed and the difference of small sample with large sample in detecting gross errors was compared. This method is verified effective through example analyses.