Probabilistic Model for Aero-Engines Fleet Condition Monitoring

Abstract

Since aeronautic transportation is responsible for a rising share of polluting emissions, it is of primary importance to minimize the fuel consumption any time during operations. From this perspective, continuous monitoring of engine performance is essential to implement proper corrective actions and avoid excessive fuel consumption due to engine deterioration. This requires, however, automated systems for diagnostics and decision support, which should be able to handle large amounts of data and ensure reliability in all the multiple conditions the engines of a fleet can be found in. In particular, the proposed solution should be robust to engine-to-engine deviations and different sensors availability scenarios. In this paper, a probabilistic Bayesian network for fault detection and identification is applied to a fleet of engines, simulated by an adaptive performance model. The combination of the performance model and the Bayesian network is also studied and compared to the probabilistic model only. The benefit in the suggested hybrid approach is identified as up to 50% higher accuracy. Sensors unavailability due to manufacturing constraints or sensor faults reduce the accuracy of the physics-based method, whereas the Bayesian model is less affected.

Assessment of Dynamic Bayesian Models for Gas Turbine Diagnostics, Part 1: Prior Probability Analysis

“…We assume a Gaussian distribution, without loss of generalization. In a faulty state, the mean of X is shifted by a certain amount ∆, depending on the sensitivity of the measured variable to the fault [11]. We can express ∆ as a function of σ, e.g., 2σ, 3σ, 4σ, etc., depending on how sensitive the measurement is to the fault.…”

“…A performance model was used to generate synthetic data to train and test the BNs. The performance model of the gas turbine system has been extensively described in previous work [11,[30][31][32]. Compared to [11,30], and similarly to [32], the model was modified to simulate a single-shaft gas turbine connected to a generator.…”

“…The performance model of the gas turbine system has been extensively described in previous work [11,[30][31][32]. Compared to [11,30], and similarly to [32], the model was modified to simulate a single-shaft gas turbine connected to a generator. The gas turbine under consideration is a single-shaft 50 MW class turbine such as the Siemens Energy SGT800.…”

“…For this reason, probabilistic models such as Bayesian networks (BNs) [6] and probabilistic neural networks [7], or methods based on fuzzy inference [8], offer an effective tool for diagnostics. Their performance has often surpassed that of deterministic models [9][10][11]. One advantage of BNs compared to other data-driven methods is their ability to handle missing data and to integrate expert knowledge when historical data are inconsistent or incomplete [12].…”

“…Machines 2021, 9, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/machines the whole fleet, engine-to-engine variations have to be taken into consideration. Initial variations in efficiency and flow capacity due to production deviations were considered as ±0.5% and ±1%, respectively [11]. Further, degradation phenomena such as fouling and erosion may manifest with different ratios between flow capacity and efficiency in different machines.…”