Model-based Gas Path Diagnosis For Three-shaft Gas Turbine

As gas turbines are more and more important in aviation, ships and energy industry,their safety and high-efficiency operation has more influence for power system securityand economy. But the high temperature, high pressure and high speed of harsh workingconditions and complex structure makes gas turbine be fault sensitive part. The researchfor Gas turbine gas path fault diagnosis has important theoretical significance andengineering application value.GT25000is three-shaft gas turbine, also a type of industrial and marine gas turbine,which has been realized the localization. The retrofit products of GT25000have had awide range of applications in the industrial power and west-east gas projects. Athree-shaft gas turbine based GT52000is used as object, model-based gas pathdiagnosis is studied in this paper.Firstly, based on modularization design, the simulation model of three-shaft turbineis set up in the MATLAB/SIMULINK environment. To ensure accuracy of the modelsimulation, gas composition, variable thermal properties, cooling, volume inertia, and soon are taken into account. Steady cycle thermodynamic calculation in the designcondition provided the necessary simulation parameters. The results of dynamicsimulation near the design condition were roughly tallying with the actual runningsituation of three-shaft gas turbine, ensured the reliability of gas turbine gas path faultdiagnosis.Secondly, based on the fitting method of linearization, a method to optimize thelittle perturbation was used to realize the linearization of the nonlinear model of thethree-shaft gas turbine, which not only could improve the accuracy of linear model, butalso did not change the original physical meaning of system matrix element. Theoptimized results could be directly applied to the buildup of linear gas path fault modelwhose state had been extended. The health parameters which could represent the healthstate of parts were extended to the system state of the accurate linear model. The lineargas path fault model could describe the deterioration of three-shaft gas turbine partsperformance. The discretization made the linear fault model can be directly applied tothe adaptive filter.Thirdly, the extended kalman filter (EKF) and strong tracking filter (STF) wererespectively used to track the gradual and abrupt gas path fault of the three-shaft gasturbine. This realized the accurate tracking of some health parameters' deterioration.Further simulation results demonstrated the accuracy and quickness of the trackingperformance of strong tracking filter. Using the existing three-shaft gas turbine modeland diagnostic system, two types of filter both only realized accurate tracking of somehealth parameters' deterioration, but were unable to complete all parts of health parameters' estimation. Here involved the observability of the diagnostic system.Finally, respectively through the method based on singular value decomposition(SVD) and linear fault model simulation method, different diagnostic systems'observability were detected and analyzed. These two methods got a consistent result.Based on the different diagnostic systems, STF was again used to diagnosed abrupt gaspath fault of the three-shaft gas turbine. Diagnostic results were in line with the previousobservability analysis. The simulation and observability analysis showed that using theappropriate diagnostic system, and the measurements were sensitive enough to thechanges of parts' performance, the filter could accurately track the changes of all theperformance parameters of gas path components of the three-shaft gas turbine.