Search

Data reconciliation (DR) is playing an important role in reducing random errors usually occurred in measured data. Principal component analysis (PCA), on the other end, deals with the reduction of dimensions when there is large number of variables involved in a complex process. In this paper, we bring these two techniques together to deal with random errors in measured data of a steam metering circuit. The results prove that PCA-based DR is effective in dealing with random errors than DR alone. The study is also extended to work on a partially measured system where only partial information of the system is known.