Gas Turbine Combustion Instability Diagnosis System and Method Using the Same

Real-time Gas Turbine Combustion Instability Diagnosis Based on Sensitivity Analysis of Combustion Dynamic Pressure Signals

This technology is a gas turbine combustion instability diagnosis system and method that measures combustion dynamic pressure using internal sensors, calculates instability metrics based on signal time periods determined by sensitivity analysis, and controls operation accordingly.

Gas turbine combustors operate at high temperatures around 1500°C, where unstable combustion causes pressure fluctuations and flame displacement, leading to localized stress and thermal cycling that can cause cracks. Existing monitoring systems, which estimate anomalies based on exhaust gas or surface temperature distribution, have limitations in accurately detecting combustion instability in real time.

This technology calculates an instability metric using the entropy of dynamic pressure signal energy and the probability of signal changes within a frame. If the metric is 0.1 or lower, it is identified as unstable, triggering an immediate response from the combustion control unit. Applicable to power and industrial gas turbine monitoring and predictive maintenance, it prevents combustor cracking and turbine blade damage, reducing the risk of unplanned shutdowns.

‍

‍

Key Features:

• A combustion unit including dynamic pressure sensors mounted inside to measure combustion dynamic pressure
• A diagnostic unit that processes dynamic pressure signals, determines signal time periods via sensitivity analysis to calculate instability metrics, and a combustion control unit that manages operation based on these findings
• A diagnostic unit that calculates combustion instability metrics and threshold values using the entropy of dynamic pressure signal energy and the probability of signal magnitude changes within a unit frame
• A step of determining combustion as unstable if the averaged combustion instability threshold value, derived from a selected number of analyzable data points in a specific frame, is 0.1 or less

‍

This invention was developed with support from the Ministry of Science and ICT for the development of AI-based big data analysis technology for predicting and diagnosing gas turbine combustion instability using high-speed temperature, flame imaging, and dynamic pressure signals.

‍