In data reconciliation

it is common practice to detect gross errors and delete those measurements with gross error(MGEs) before using the algorithm of reconciliation. Each measurements deletion affects the structure of measurement network as well as reconciliation precision. Since some MGEs deletion will perhaps result in the unreasonable solution of the least squares algorithm of data reconciliation

MGEs can not be deleted totally at one time. They must be treated differently in order to delete as many MGEs as possible to ensure reconciliation precision. To solve this problem

a new method of redundancy analysis is proposed in this paper. Supposing one MGE is deleted and treated as unmeasured variable

the corresponding formulation of the data reconciliation problem is adjusted and its reconciliation precision can be calculated. Thereupon

the first order redundancy degree of the deleted MGE is defined. After all of the MGEs are checked

the MGE with the largest first order redundancy degree will be deleted and treated as unmeasured. In the same way

the higher order redundancy degree of the remained MGEs can be calculated

and they can be removed one by one until the deletion of each one left will lead to no redundancy of the network. In this way

the information in the data can be used sufficiently and the reconciliation precision is ensured. Simulation result demonstrates the efficiency of the proposed method.