Employment of Intelligent Predictive Maintenance on Thermal Power Plant Component Parts Taking Condenser Vacuum as a Case Study

Abstract

This work proposes deployment of machine learning in the maintenance of individual constituent parts of steam power plant assemblages. With the condenser vacuum of a steam turbine (in a six-turbine plant assemblage) taken as a case study, information on the past operating parameters of the selected plant component was used to forecast its future working condition. Based on Exponential Gaussian Process of Regression, a model was developed, trained using the diachronic operational data, and employed in determining the future. A quantitative evaluation was employed to provide the distribution of the test values of the data about the lines of regression, as well as to measure the prediction accuracy of the model. The results show MAE and RMSE values are 6.1602 and 7.9286 respectively during the training; while for the prediction, the values are 92.6544 and 92.7235 respectively. It is concluded that modern power plants with myriads of instrumentation and data acquisition mechanisms can leverage on the approach of this study to model and plan the maintenance scheme that best suits and fits individual component units of power plants, since understanding of the anticipatory values of operational parameters helps to determine the likelihood of components failures.