Deep Fault Diagnosis For Steam Turbines Based On Industrial Big Data

Turbine generator is the core equipment of a thermal power plant.The failure or shutdown of the turbine generator will have a great impact on the production efficiency and safety of the power plant.Turbine generating unit will produce a large number of process data,containing a wealth of equipment status information.How to mine useful information from a large amount of industrial big data for turbine condition monitoring and fault diagnosis has gradually become the turbine fault diagnosis research hot spots.Under this background,this thesis conducted a comprehensive survey on the current turbine generator condition monitoring and fault diagnosis methods and systems,and an in-depth analysis of the existing methods.In order to solve those unsolved problems,a method to analyze turbine faults deeply based on industrial big data is proposed.A practical Web-based visual analysis system is developed to analyze the historical failure cases of the steam turbine of a power plant.The main contributions of this paper are as follows.1.A steam turbine condition monitoring approach based on nonlinear state estimation is proposed.Based on the nonlinear state estimation theory,an improved method of nonlinear state estimation method based on standard Euclidean distance is proposed,which realizes the condition monitoring and early warning of faults for several turbine units.2.A fault diagnosis approach for steam turbines based on delay correlation analysis and visualization is set up.Based on the Pearson correlation coefficient,a delay correlation analysis method is proposed,which can effectively calculate the real correlation coefficients among all parameters and visualize them through the correlation matrix diagram,so as to indicate the faults more efficiently.3.A deep fault analysis system for steam turbines is developed.Combining the human-computer interaction theory and data visualization method,turbine condition monitoring and fault diagnosis are integrated.An interactive turbine fault analysis system is designed and developed to visualize the analysis processes and results through data visualization.