Abstract: Measurements can be contaminated with gross errors due to various reasons such as measurement irreproducibility, sensor problem and other reasons. The direct data reconciliation will diffuse gross errors, which affect reliability and accuracy of data reconciliation results. As a result, before the data reconciliation, it needs to identify and remove the measurement data contaminated with gross errors. The existing methods of gross errors detection have not a good detection result for limited gross errors (less than or equal to 3). This paper presents a new method of detecting and identifying gross errors based on “3σ rule”, which not only has a good detection result for limited gross errors but also for more gross errors. By using the method of collective compensation for gross errors, this method can effectively prevent the emergence of the singular matrix.

Key words: gross errors, data reconciliation, probability and mathematical statistics, collective compensation