Gas turbine performance monitoring based on extended information fusion filter

Abstract

Performance monitoring is a critical issue for gas turbine engine for improving the operation safety and reducing the maintenance cost. With regard to this, variants of Kalman-filters-based state estimation have been employed to detect gas turbine performance, but the classical centralized Kalman filters are subject to heavy computational effort and poor fault tolerance. A novel nonlinear fusion filter algorithm using information description with distributed architecture is proposed and applied to gas turbine performance monitoring. This methodology is developed from federated Kalman filter, and a bank of local extended information filters and one information mixer are combined with extended information fusion filter. The local state estimates and covariance calculated in parallel by the local extended information filters are integrated in the information mixer to yield a global state estimate. The global state estimate of nonlinear system is fed back to the local filters with weighted factor for next iteration. The aim of the proposed methodology is to reduce the computational efforts of state estimation and improve robustness to sensor faults in cases of gas turbine performance monitoring. The simulation results on a turbofan engine confirm the extended information fusion filter's effective capabilities in comparison to the general central ones.

A Review of Information Fusion Methods for Gas Turbine Diagnostics

“…Information fusion was accomplished by means of DS theory with evidence reliability coefficients, and the results showed sufficient accuracy even in the presence of modeling uncertainty. Lu et al successively demonstrated the use of a non-linear Kalman filter with a distributed architecture for information fusion [79].…”

Noise‐robust gas path fault detection and isolation for a power generation gas turbine based on deep residual compensation extreme learning machine

“…This identification process is executed through two distinct approaches: physics‐based models that leverage the governing physical laws of the problem, and data‐driven models. Several physics‐based methodologies have been suggested, encompassing techniques such as Kalman filters, 4,5 weighted least squares, and influence coefficient matrix 6 . In a recent study, Li et al presented a physics‐based method for diagnosing gas turbine faults within a power plant, considering the variable geometry of the compressor and operating under transient conditions 7 .…”