DATA-DRIVEN IDENTIFICATION OF OPERATING PATTERNS IN A THERMAL POWER PLANT BY CLUSTERING METHODS

Resumo

Thermal power plant operation depends on the knowledge of a wide range of complex and cross de-
pendent parameters. Information is usually captured through Distributed Control Systems (DCS) which allow to

access up to date data but also long periods of recorded operation. Large and available data sets are decisive for
plant operation, but they must be properly used and interpreted to achieve effectiveness. The purpose of the present
paper is to present an identification of operational patterns from historical data from an actual thermal power plant
based on unsupervised machine learning methods. The proposed methodology is applied to a long term data series

from the 360 MW Brazilian coal-fired Pecem power plant, for 29 selected parameters, concerning its steam gen-
erator and associated mills. Dataset size and redundancy is treated by the Principal Component Analysis (PCA)

approach, which defines a lower dimensional space, proper for clustering while preserving most of its variance.

The K-means clustering method identifies operating point groups according to their degree of similarity. The ap-
propriate cluster number is defined by means of the average silhouette coefficient, which measures the clusters

consistency. Cluster parameter values and distribution are evaluated to verify result consistency. The assessment

with the 29 parameters from the steam generator and mills system is presented, and the results show that the op-
eration may be described globally by a 2 clusters analysis or, for refined observations, by a 10 clusters analysis.

The different patterns encountered facilitate an understanding of the parameters arrangement and resulting perfor-
mance, enabling the identification of low efficiency operation conditions and supporting practices to improve the

plants operation.