Dimitrios Dimogianopoulos

FDI for Aircraft Systems Using Stochastic Pooled-NARMAX Representations: Design and Assessment

In this paper, two statistical schemes aiming at effective fault detection and isolation (FDI) for aircraft systems are introduced. They are based on novel stochastic pooled nonlinear autoregressive moving average with exogenous excitation representations that model the relationships among available aircraft signals, as well as statistical decision making. The first, or ldquodirect,rdquo scheme relates a pilot input to a measurable flight attitude via a two-stage pooled representation. The second, or ldquoindirect,rdquo scheme relates four attitude-dependent flight variables via a proper pooled representation. Both schemes achieve effective FDI operation inside an entire flight regime, under stochastic effects and uncertainty and under various operating or environmental conditions, at the price of increased computational effort during training. Their performance and robustness are assessed via many flights conducted with an aircraft simulator inside the considered flight regime, under different conditions and under faults of various types and magnitudes.

Contact-Free Magnetoelastic Smart Microsensors With Stochastic Noise Filtering for Diagnosing Orthopedic Implant Failures

A novel contact-free magnetoelastic microsensor for diagnosing failing orthopedic implants is introduced. It uses an implant-embedded 30-mum-thick MetGlas-2826 film and a sensing coil placed 30 mm away from the film. Implant loading generates thin-film magnetoelastic response, passively (contact-free) detected as voltage by the coil. A specific integrated stochastic nonlinear filter performs data denoising, and a decision-making module concludes (via data assessment) on the health state of the implants. The sensor is successfully validated in vitro on an external fixation system and a hip prosthesis implant, respectively.

Aircraft fault detection and identification by stochastic functionally pooled modelling of relationships among attitude data

Abstract This paper introduces a statistical fault detection and identification (FDI) scheme for aircraft systems, which uses flight attitude data rather than information from purposely developed physical or virtual sensors. The scheme is based on the modelling of relationships among the considered data via stochastic Time-dependent Functionally Pooled Non-linear AutoRegressive with Exogenous excitation (TFP-NARX) representations. These are globally valid inside a flight regime and under various considered environmental conditions, thanks to the pooling technique used for their identification. Moreover, due to the TFP-NARX coefficients being a function of time-dependent quantities, high modelling accuracy is achieved. The schemes operation involves identifying nominal TFP-NARX models of relationships among the attitude data from an aircraft operating in a healthy state. Owing to fault occurrence, these relationships may change. Then, an in-flight comparison of the nominal and the current aircraft dynamics provides fault-related information, which is statistically evaluated for FDI purposes. The schemes performance and robustness are assessed with numerous flights conducted throughout a flight regime under various manoeuvring settings and turbulence levels.

Fault Detection and Isolation in Aircraft Systems Using Stochastic Nonlinear Modelling of Flight Data Dependencies

This paper introduces a fault detection and isolation (FDI) scheme for aircraft systems based on the modelling of the relationships among flight variables. The modelling is performed by means of pooled nonlinear autoregressive with exogenous (NARX) excitation representations. During the systems operation in healthy mode, these relationships are valid. Hence, a scheme using statistical hypothesis testing is designed to detect changes in these relationships as a result of fault occurrence. The FDI schemes performance and robustness are assessed with flights conducted under various external flight conditions (turbulence)

On-board statistical detection and control of anomalous pilot-aircraft interactions

Abstract A statistical method for the on-board detection and control of oscillatory phenomena in pilot-aircraft systems is presented. Recursive identification is used to obtain a linear model of the system at every time instant. The estimated system parameters are monitored, and the system stability margins are continuously assessed. Oscillations due to stability loss are detected early using a composite statistical hypothesis test. Finally a simple stability augmentation system is designed to assist the pilot-aircraft system during the critical time intervals. The method is successfully tested with data from a detailed nonlinear aircraft model and a flight simulator facility.

Adaptive control for linear time-varying systems using direct least squares estimation

An adaptive controller based on a least squares non-recursive identification algorithm is proposed. The adaptive scheme is capable of dealing with slowly time-varying parameters. The minimized criterion of the estimation algorithm is an L/sub 2/ norm of the identification error with forgetting factor. The proposed estimation algorithm does not require explicit knowledge of the noise bound or the region where the true parameters lie. The control stategy involves a pole placement hybrid control law. We study the case when there is no persistent excitation and show how we can modify the estimates in order to avoid singularities in the control law. Finally we show how the control objectives can be achieved.

Damage assessment of carbon fiber reinforced composites under accelerated aging and validation via stochastic model-based analysis

Composite materials used in technically advanced structures are subjected to constant aging from exposure to changing environmental conditions. Studying the effects on such materials due to exposure to varying temperature, humidity, ultraviolet radiation, etc. reveals the impact on their mechanical behavior. This study assesses alterations in static, dynamic, and viscoelastic response of polymer matrix woven carbon fiber lamina composites upon exposure to varying environmental conditions recreated in a climatic chamber. Therein, specimens suffered temperature changes from –35 to +40℃ and humidity variations from <10% to 95% RH (noncondensing) over a period of up to 30 days. Alternating cycles simulating conditions of actual 3–4 h flights were specified. Additionally, specimens of the same material were subjected to thermal shock under similar (to the aging scenario) temperature extremes. All specimens were comparatively assessed via experimental procedures involving three-point bending tests performed in both static and dynamic mechanical analysis for a range of temperatures and frequencies, frequency and thermal scans, and finally impact tests. Results indicate that aged materials exhibit increased dynamic stiffness (expressed by the storage moduli) and decreased material damping ability (expressed by the tan δ parameter). Macroscopic assessment of impact test data was performed via stochastic model-based damage detection methodologies. Results indicate that differences in the impact behavior between pristine and aged specimens are statistically detectable and quantifiable, without input from mechanical testing analysis. More importantly, this assessment of aging-induced effects on the specimens corroborates the findings on storage moduli and tan δ from mechanical testing analysis, thus validating the latter.

ECG diagnosis via a sequential recursive time series — Wavelet classification scheme

A novel scheme for diagnosing non-stationary electrocardiogram (ECG) records using a combination of stochastic time-series detection and wavelet-based classification methods is presented. The ECG diagnosis algorithm stems from a two-stage procedure, which initially detects and subsequently classifies ECG segments bearing cardiac abnormalities. In the first stage, recursive stochastic time-series representations (as applied to fault diagnosis of mechanical systems) are used for detecting any potentially abnormal heartbeat incidents in the ECG signal. During the second stage, the detected incidents are fed into a wavelet classifier, which assigns the corresponding heartbeats to different classes, i.e. Supraventricular, Ventricular, Fusion and Normal. The resulting classification task is thus focused on abnormal ECG segments, as the segments related to healthy heart status are discarded by the detection step. The schemes performance is evaluated on several ECG recordings from the MIT-BIH Arrhythmia database. Despite minimal preprocessing of the ECG recordings and the simplicity of the ECG features extraction scheme with respect to other well-established schemes, the algorithmic performance is comparable.

Statistical damage diagnosis in smart systems via contact-free MetGlas ® sensors and stochastic non-linear modelling of system output data

A contact-free, non-destructive concept for damage diagnosis in smart systems is introduced. It utilises in-house developed magnetoelastic contact-free sensors, providing output measurements of the system under load without bearing any physical contact with it. The system’s health state is diagnosed via a specifically developed data processing scheme: firstly, the measured data are modelled via stochastic non-linear autoregressive (NAR) representations for capturing the health state-related system dynamics, and secondly, advanced statistical decision-making tests are used for evaluating this information and concluding on the system’s health state. The experiments involve smart systems (formed by magnetoelastic MetGlas ® alloy stripes attached to polymer epoxy resin slabs) undergoing vibration testing of growing amplitude in a dynamic mechanical analyser. Output data from such ‘healthy’ and ‘damaged’ systems are then assessed using the scheme, and finally, detection and severity evaluation, that is diagnosis, of damage are reliably concluded.

Integral minimum variance-like control for pooled non-linear representations with application to an aircraft system

This paper presents an integral minimum variance-like controller design based upon a constant coefficient pooled non-linear autoregressive moving average with exogenous excitation (CCP-NARMAX) representation. The use of pooling techniques significantly enhances the NARMAX representations ability to accurately describe systems performing under various operating conditions such as aircraft systems, chemical processes, industrial systems and so on. The controller design introduces suitable modifications to account for the characteristics of the CCP-NARMAX representation. The control strategy is subsequently applied to a non-linear aircraft system in order to obtain regulation of the pitch rate around a predetermined value. Comparisons with a conventional PID control design are also made under various operating conditions, including disturbances due to external input and turbulence.