Comparison of Methods of Gross Error Location Technology in Hydraulic Safety Monitoring
Jian Liu, Xinghe Liu, Xiuwen Li, Shunfu Zhang · 2021 3rd International Academic Exchange Conference on Science and Technology Innovation (IAECST) · 2021
Gross error location is an important work in data preprocessing of hydraulic safety monitoring. Several different gross error location methods are compared and introduced, and their applicable scope and limitations are discussed in this paper. Based on the normal distribution function, the traditional statistical test method can be applied to the gross error detection of repeated measurements of small samples, but it cannot adapt to the dynamic external load conditions faced by large samples and long sequences of monitoring data; the logic test method improved by "3 σ criterion" can take the correlation of monitoring values in time series into account; ARIMA modeling method is suitable for detecting gross errors of monitoring sequences which are relatively continuous in time; for the monitoring effect quantity with clear related influencing factors, the statistical model method can effectively locate the gross error points; Grey System GM (1, 1) and GM (2, 1) models are suitable for gross error identification of data with insufficient information. The research results have certain reference value.