Laura Lohiniva and Kimmo Leppäkoski
data analysis, self-organizing map, monitoring, power plants
Preliminary data analysis can be used for studying data properties and variable dependence. Data mining methods such as the Self-Organizing Map (SOM) can be effectively and easily applied even on large data sets. In this case study, process measurement data from normal operation of a circulating fluidized bed (CFB) boiler was tentatively analyzed by using correlation calculation for quantitative results and SOM for visualization. Results were compared and found to support each other. SOM efficiently visualizes the numerical results obtained by correlation analysis. The used methods enable the recognition and evaluation of dependence among variables in a computationally inexpensive way.
Important Links:
Go Back
