摘要
目前大坝监测的数据处理一般采用误差服从正态分布的假设。而现实中严格服从正态分布的观测序列是很少的，尤其当数据中含有粗差，与正态分布相差较大时，建立的模型将会偏离实际情况。为此，通过应用抗差最小二乘法，笔者提出了大坝监测数据处理的抗差多元回归模型，阐述了其建模的机理、步骤，同时还可利用该法的权重进行粗差检测。
At present the classic least square method. which is only adaptable to the data series of normal distribution. is usually adopted in statistic computing. However, the actual series are usually contaminative distribution instead of normal distribution. When the survey is not of normal distribution due to the gross error in the data series. the models based on the method deviate seriously from the actual situalion. Therefore, a new method is presented to mend the existing data processing methods with the application of the theory of robust estimation. Through introduction of pollution distribution and survey weights. gross error can be detected and processed step by step by iterative computation. and the best parameters can be estimated after the effect of gross error is eliminated.
出处
《大坝观测与土工测试》
2000年第3期18-21,共4页
Dam Observation and Geotechnical Tests
