Data Mining for the Analytical Redundancy of Power Plant Critical Parameters

As a new modeling thought, the accurate analytical redundancy model of power plant critical parameters was established by data mining method, which obtained effective information from the large number of real-time operation data. The basic modeling mode, including data preprocessing, mining model, verification model and the strategy from data to analytical redundancy model, was proposed in the paper. Under this mode intermediate point temperature modeling was given as an example. Considering the system was non-linear, time-varied, and multivariate and coupled, PSO-LS-SVM algorithm was used and compared to LS-SVM, BPNN, and PLS. Results showed that the obtained model was enough accurate and inexpensive in terms of memory and time required. The model maximum error was 2.59°C. So the proposed modeling thought and the mode were effective for the analytical redundancy and could enhance the modeling accuracy.