Estimation of Gas Turbines Gradual Deterioration Through a Dempster-Schafer based Fusion Method

Authors:Romessis C., Mathioudakis K.

Abstract

This paper presents a fusion procedure of independently acting diagnostic methods, allowing gas turbines health condition assessment given a series of measurements. The proposed procedure incorporates a fusion technique, which is based on the Dempster-Schafer theory. The novel element of the method is its ability to cope with the problem of overall engine gradual performance deterioration, instead of identification of individual component fault events. The effectiveness of the technique is evaluated through its application on scenarios representing drifting gas turbine faults encountered in practice, using independently acting diagnostic methods, already established. Through this application, the efficiency of the proposed fusion procedure is demonstrated, along with the improvement it provides over its constituent methods to both the accuracy and the reliability of diagnosis.

Gas Turbine Fault Identification by Fusing Vibration Trending and Gas Path Analysis

Authors:Kyriazis A., Tsalavoutas A., Mathioudakis K., Bauer M., Johanssen O.

Abstract

A fusion method that utilizes performance data and vibration measurements for gas turbine component fault identification is presented. The proposed method operates during the diagnostic processing of available data (process level) and adopts the principles of certainty factors theory. Both performance and vibration measurements are analyzed separately, in a first step, and their results are transformed into a common form of probabilities. These forms are interweaved, in order to derive a set of possible faulty components prior to deriving a final diagnostic decision. Then, in the second step, a new diagnostic problem is formulated and a final set of faulty health parameters are defined with higher confidence. In the proposed method the non-linear gas path analysis is the core diagnostic method, while information provided by vibration measurements trends is used to narrow the domain of unknown health parameters and lead to a well defined solution. It is shown that the presented technique combines effectively different sources of information, by interpreting them into a common form and may lead to improved and safer diagnosis.

Gas Turbines Diagnostics Using Weighted Parallel Decision Fusion Framework

Authors:Kyriazis A., Mathioudakis K.

Abstract

A technique that allows the fusion of decisions provided by independent diagnostic methods for gas turbines faults is presented. It utilizes a Bayesian Belief Network to handle the various diagnostic decisions and provides a final assessment about the engine health condition. This task is performed in a parallel fusion framework with the optional addition of a weighting module, in two variants, for further diagnostic enhancement. Effectiveness of the proposed technique is illustrated by application to the detection of mechanical and aerothermodynamic component faults. In the latter case diagnostic conclusions of GPA (Gas Path Analysis) methods are merged delivering an improved diagnosis. In the case of mechanical faults, a radial compressor is examined by utilization of fast response data and diagnostic decisions are also combined for final diagnosis. The proposed method exhibits broad generality by utilizing different sources of information and fault scenarios alongside with improved diagnostic effectiveness.

Gas Turbine Fault Diagnosis Using Fuzzy-based Decision Fusion

Authors:Kyriazis A., Mathioudakis K.

Abstract

A two-step information fusion technique allowing the combination of fast response and performance data for further improvement of gas turbines diagnostic procedure is proposed in this paper. The proposed technique derives from the notion of Decision level Fusion where different diagnostic methods provide assessments for the condition of an engine and the final decision results from a combination of these assessments. The diagnostic method of Probabilistic Neural Networks (PNNs) acts independently and in parallel on data of a different nature. Their conclusions are entered in the first step of the fusion technique and are aggregated deriving the probability consensus. In second step this consensus is classified concerning a set of examined faults utilizing Fuzzy Set theory and Fuzzy Logic, providing thus the final diagnostic decision.

Fusion of Gas Turbines Diagnostic Inference: The Dempster-Schafer Approach

Authors:Romessis C., Kyriazis A., Mathioudakis K.

Abstract

This paper proposes a fusion technique allowing the merge of conclusions provided by diagnostic methods that act independently for the detection of gas turbine faults. The proposed technique adopts the principles of Dempster-Schafer theory for the fusion of two diagnostic methods output; these are the method of Bayesian Belief Networks (BBN) and the method of Probabilistic Neural Networks (PNN). The proposed technique has been applied for the detection of thermodynamic as well as mechanical faults on gas turbines. First, the case of a turbofan engine of civil aviation is examined. The proposed technique allows the fusion of diagnostic inference on the presence of several faults of thermodynamic nature. Then the case of a radial and an axial compressor are examined, were several mechanical faults are deliberately implemented. In all cases, the effectiveness of the proposed fusion technique demonstrates that the merge of diagnostic information from different sources leads to better and safer diagnosis.

Gas Turbine Fault Diagnosis From Fast Response Data Using Probabilistic Methods and Information Fusion

Authors:Kyriazis A., Aretakis N., Mathioudakis K.

Abstract

The paper covers first the use of probabilistic neural networks for the classification of spectral fault signatures obtained from fast response data (sound, vibration, unsteady pressure). The method is compared to other alternatives, such as geometrical and statistical pattern recognition. Effectiveness of the method is demonstrated by presenting the results from application to data from a radial compressor and an industrial gas turbine. Further, probabilistic methods are used to perform information fusion. The outcomes of different diagnostic methods are used as a first level of diagnostic inference, and are fed to two different fusion processes which are also based on i)Probabilistic Neural Networks and ii)Bayesian Belief Networks. It is demonstrated that these fusion processes provides powerful tools for effective fault classification.

Combining Classification Techniques With Kalman Filters For Aircraft Engine Diagnostics

Authors:Dewallef P., Romessis C., Leonard O., Mathioudakis K.

Abstract

A diagnostic method consisting of a combination of Kalman filters and Bayesian Belief Networks (BBN) is presented. A soft constrained Kalman filter uses a priori information derived by a BBN at each time step, to derive estimations of the unknown health parameters. The resulting algorithm has improved identification capability in comparison to the stand alone Kalman filter. The paper focuses on the way of combining the information produced by the BBN with the Kalman filter. An extensive set of fault cases is used to test the method, on a typical civil turbofan layout. The effectiveness of the method is thus demonstrated and its advantages over individual constituent methods are shown.

Bayesian Network Approach for Gas Path Fault Diagnosis

Authors:Romessis C., Mathioudakis K.

Abstract

A method for solving the gas path analysis problem of jet engine diagnostics based on a probabilistic approach is presented. The method is materialized through the use of a Bayesian Belief Network (BBN). Building a BBN for gas turbine performance fault diagnosis requires information of a stochastic nature expressing the probability of whether a series of events occurred or not. This information can be extracted by a deterministic model and does not depend on hard to find flight data of different faulty operations of the engine. The diagnostic problem and the overall diagnostic procedure are first described. A detailed description of the way the diagnostic procedure is set-up, with focus on building the BBN from an engine performance model, follows. The case of a turbofan engine is used to evaluate the effectiveness of the method. Several simulated and benchmark fault case scenarios have been considered for this reason. The examined cases demonstrate that the proposed BBN-based diagnostic method composes a powerful tool. This work also shows that building a diagnostic tool, based on information provided by an engine performance model, is feasible and can be efficient as well.

Implementation of Stochastic Methods For Industrial Gas Turbine Fault Diagnosis

Authors:Romessis C., Mathioudakis K.

Abstract

Implementation of stochastic diagnostic methods for diagnosis of sensor or component faults is presented. Two industrial gas turbines are considered as test cases, one twin and one single shaft arrangement. Methods based on Probabilistic Neural Networks (PNN) and Bayesian Belief Networks (BBN), are implemented. The ability for successful diagnosis is demonstrated on specific cases of sensor malfunctions, as well as on two types of compressor deterioration, fouling and variable vane mistuning. The examined diagnostic problem and the methods of PNN for sensor fault diagnosis and BBN for the diagnosis of component faults are first described. For each gas turbine case, the implementation of the diagnostic methods is shown and application to fault cases that occurred is presented. The effectiveness of the stochastic diagnostic methods demonstrates that they offer a powerful alternative diagnostic tool.

Identification of Sensor Faults on Turbofan Engines Using Pattern Recognition Techniques

Authors:Aretakis N., Mathioudakis K., Stamatis A.

Abstract

The possibility to identify faults in the readings of the sensors used to monitor the performance of a high by-pass ratio turbofan engine is examined. A novel method is proposed, based on the following principle: if a measurement set is fed to an adaptive performance analysis algorithm, a set of component performance modification factors (fault parameters) is produced. Faults, which may be present in the measurement set, may be recognized from the patterns they produce on the modification factors. The constitution of a method of this type based on pattern recognition techniques is discussed here. Test cases corresponding to different sensor faults, simulating operation in real conditions are examined. Three kinds of pattern recognition techniques with increasing complexity are used, in order to correctly identify the examined sensor faults. It is demonstrated that by choosing an appropriate formulation it is possible to have a 100% success in the identification of the examined sensor faults.

Probabilistic Neural Networks For Validation of On-Board Jet Engine Data

Authors:Mathioudakis K., Romessis C.

Abstract

A method is presented for identification of faults in the readings of sensors used to monitor the performance and the condition of jet engines. Probabilistic neural networks are used to detect the presence and identify the location and magnitude of faults (biases) in sensor readings. The faults can be detected on sets comprising a limited number of instruments, typical of those available for on-board monitoring of jet engines. An engine performance model is used to support the constitution of a network. Training information is built using the model to produce data for a comprehensive set of healthy and faulty situations. The network performance in detecting and quantifying sensor faults is validated on a large number of fault cases, also generated by a model, which are used for testing the network and cover a wide range of conditions that can be encountered in practice. An engine, representative of current large civil engine designs (large bypass, partially mixed turbofan), serves as the test vehicle for demonstration of the way the method is materialized.

Setting Up Of a Probabilistic Neural Network for Sensor Fault Detection Including Operation with Component Faults

Authors:Romessis C., Mathioudakis K.

Abstract

The diagnostic ability of Probabilistic Neural Networks (PNN) for detecting sensor faults on gas turbines is examined. The structure and the features of a PNN, for sensor fault detection, are presented. It is shown that with the proposed formulation, a powerful tool for sensor fault identification is produced. A particular feature of the PNN produced is the ability to detect sensor faults even in the presence of engine component malfunction, as well as on deteriorated engines. In such situations, the size of bias that can be identified increases. The way to establish the limits of sensor bias that can be detected is presented along with results from application to test cases with realistic noise magnitudes. The diagnostic procedure proposed here is also supported by an engine performance model. The data used for setting up and testing the PNN are generated by such a model.

Jet Engine Sensor Validation with Probabilistic Neural Networks

Authors:Romessis C., Mathioudakis K.

Abstract

A method of identifying faults on the sensors used to monitor the performance and the condition of jet engines is presented. The method uses probabilistic neural networks and can identify the presence of a fault in the reading of a sensor and quantify the magnitude of this fault, namely the bias from the true reading. The possibility to detect this magnitude is a novel aspect of the paper, as compared to previous work of the authors as well as the methods that exist in the open literature. The method uses an engine performance model to support the construction of the network. A sensor fault data base is formed in order to train the network. The capability of the network to correctly detect and quantify sensor faults is tested on a large number of fault cases, which are used for testing the network and covers a large area of conditions that can be encountered in practice. The effectiveness of the method is shown through its implementation in an engine which is representative of current large civil engine designs (large bypass, partially mixed turbofan).

Setting up a Belief Network for Turbofan Diagnosis With the Aid of an Engine Performance Model

Authors:Romessis C., Stamatis A., Mathioudakis K.

Abstract

This paper presents a method for building Bayesian Belief Networks (BBN) for gas turbine performance fault diagnosis. Building a BBN requires information of stochastic nature. It is a common practice to extract this kind of information from statistical analysis of large data sets. In the field of gas turbine diagnostics, though, such data are usually hard to find. With the present method, the required information is extracted from an engine performance model. In this way, stochastic information, expressing the probability of whether a series of events occurred or not, can be extracted by a deterministic model and does not depend on, hard to find, flight data of different faulty operations of the engine. The diagnostic problem is first described. Some basic concepts of BBN, though briefly described, are also presented in relation to turbofan engines. A detailed description of the proposed way to set-up a diagnostic BBN from an engine performance model follows. Several simulated, but realistic, fault cases are then used for inference with the constructed network. Inference with BBN showed that such a network is very reliable, since in the 96% of the cases where a fault was detected, it was detected correctly. Only a 4% of the cases was attributed to a wrong fault. In some cases, the network was not ësensitiveí in the presence of a fault, since it did not detect any fault at all. Further, preliminary work, though, shows that the ësensitivityí of the network can be increased. It is shown that building a BBN, based on information provided by an engine performance model, is feasible and can be efficient as well.

A Parametric Investigation of the Diagnostic Ability of Probabilistic Neural Networks on Turbofan Engines

Authors:Romessis C., Stamatis A., Mathioudakis K.

Abstract

Fault identification through the use of Artificial Neural Networks has become very popular recently. Probabilistic Neural Networks (PNN) is one of the architectures, which have mostly been investigated for gas turbine diagnostics. In this paper, the influence of parameters related to the structure and training on the diagnostic performance of a probabilistic Neural Network (PNN), is investigated. In particular, the parametric investigation examines the effect of the training set on the diagnostic performance of a PNN. The effect of noise level was also examined and found to be important. Another parameter examined is the severity of a fault, which was found to affect seriously the performance of the diagnostic PNN. Other parameters also examined are the effect of the operating conditions as well as the considered output parameters of the network. Guidelines useful for setting up this type of network, are derived.

Classification of Radial Compressor Faults Using Pattern Recognition Techniques

Authors:Aretakis N., Mathioudakis K.

Abstract

An application of pattern recognition techniques for classification of faults in a radial compressor is presented. A number of mechanical alterations, simulating faults, are introduced in a test compressor. They include the insertion of an inlet obstruction, an obstruction in a diffuser passage, variation of impeller tip clearance and impeller fouling. Two kinds of measurements namely sound emission and casing vibration are examined. Three kinds of pattern recognition techniques with increasing complexity are used in order to correctly classify the examined faults according to engine condition. The possibility of using each one of these techniques for diagnosing faults in a radial compressor is also examined. It is demonstrated that minor faults, which do not affect performance, can be identified using the proposed techniques

Incorporating Neural Networks into Gas Turbine Performance Diagnostics

Authors:Kanelopoulos K., Stamatis A., Mathioudakis K.

Abstract

Possibilities of incorporating neural nets in different tasks of a gas turbine performance diagnostic procedure are investigated. The purpose is to examine how neural nets can be implemented and what advantages they may offer. First, the possibility to constitute a performance model by using neural nets is considered. Different modes of operation are examined and the neural net architectures for achieving better accuracy are discussed. Subsequently, different problems of fault detection and identification are considered. Classification of faults is performed on the basis of diagnostic parameters produced by adaptive modelling. Both sensor faults and actual engine component faults are examined. A decision logic based on several neural nets is proposed. At a first level it is decided whether a fault exists, and if yes, checks are performed in order to identify the fault in as much detail as possible. Summarizing, the paper discusses different aspects of neural net implementation, in an effort to provide guidelines for application of this type of technique in the field of gas turbine diagnostics.

Gas Turbines Compressor Fault Identification by Utilizing Fuzzy Logic-Based Diagnostic Systems

Authors:Kyriazis A., Helmis I., Aretakis N., Roumeliotis I., Mathioudakis K.

Abstract

This paper proposes the development of Fuzzy Logic diagnostic systems for the detection and identification of gas turbines faults by utilizing performance and fast response data. The proposed diagnostic systems are developed in two successive steps. First the fuzzy rules are extracted and filtered and second the complete fuzzy inference system is built with rule base from first step. In the case of performance data, deliberately implanted mechanical faults in an axial and a radial compressor are examined, while in the case of fast response data blade faults in the axial compressor are also studied. Diagnostic results from pattern recognition and Probabilistic Neural Networks methods are provided for comparison in both cases. The effectiveness of the proposed systems is illustrated by this comparison, revealing thus an alternate diagnostic tool. Also, their broad generality is exhibited by application to different fault scenarios and utilization of different nature data.