Fault Detection for Gas Turbine Sensors Using I/O Dynamic Linear Models: Methodology of Fault Code Generation

Abstract

This paper presents a methodology of sensor diagnosis which appears to be particularly suitable also for application in the field of small/medium power size industrial gas turbines. The methodology is based on the Analytical Redundancy technique and uses ARX (Auto Regressive with eXternal input) MISO (Multi-Input/Single-Output) linear dynamic models obtained from time series data of the gas turbine operating condition. The linear models allow the on-line calculation of some measurable parameter starting from the values of other measured parameters. The comparison between computed and measured values of the same parameters allows setting-up a vector of residuals which, if compared with the columns of the fault matrix, permits the identification of a possible sensor fault. The initial applications of the methodology to a single-shaft industrial gas turbine show an unambiguous and certain detection and isolation of fault in sensors used both in the measurement only and in feedback by the machine control system.

ARX Linear Model Set-Up for Fault Diagnosis of Gas Turbine Sensors

“…In this paper, the case of sensor fault diagnosis of a single-shaft industrial gas turbine with variable IGV angle, working in parallel with electrical mains in a cogeneration plant, was specifically considered (Bettocchi et al, 1996b).…”

“…This paper illustrates a methodology for setting-up a set of linear models for diagnosing sensor faults of a single-shaft industrial gas turbine (Bettocchi et al, 1996b), pointing out some of the difficulties in applying system modeling and system identification methods.…”

Diagnosis of Gas Turbine Operating Conditions by Means of the Inverse Cycle Calculation

“…If the number of measurements is increased it is possible to eliminate, among the problem inputs, the more critical measurements (i.e., the measurements of fuel mass flow rate and lower heating value). b) To use techniques for the detection and isolation of measurement sensor faults to ascertain that measurement errors are lower than the minimal detectable faults (Bettocchi et al, 1996(Bettocchi et al, , 1998Simani and Spina, 1998;Simani et al, 1998;Bettocchi and Spina, 1999). c) To submit the results of the calulations to statistical analyses, using for example weighted least-squares techniques (Doel, 1993(Doel, , 1994, which can be coupled to expert systems in the maintenance decision process (Doel, 1990;Palmer, 1998).…”

A Method for the Diagnosis of Gas Turbine Sensor Faults in Presence of Measurement Noise

“…The faults were simulated during transients different from the ones relative to the time series of data used in the model identification phase, since this is the worse case to isolate a fault. In particular the residuals in fault-free conditions due to ARX model approximation during transients are higher than the ones in steady state conditions (Bettocchi et al, 1996b). Table 2 shows, for each model, the residual standard deviation referred to mean value and the residual mean value in a fault-free condition, which are due both to ARX model approximation and noise.…”