Pierino Gianni Bonanni

Fusing competing prediction algorithms for prognostics

Two fundamentally different approaches can be employed to estimate remaining life in faulted components. One is to model from first principles the physics of fault initiation and propagation. Such a model must include detailed knowledge of material properties, thermodynamic and mechanical response to loading, and the mechanisms for damage creation and growth. Alternatively, an empirical model of condition-based fault propagation rate can be developed using data from experiments in which the conditions are controlled or otherwise known and the component damage level is carefully measured. These two approaches have competing advantages and disadvantages. However, fusing the results of the two approaches produces a result that is more robust than either approach alone. In this paper, we introduce an approach to fuse competing prediction algorithms for prognostics. Results presented are derived from rig test data wherein multiple bearings were first seeded with small defects, then exposed to a variety of speed and load conditions similar to those encountered in aircraft engines, and run until the ensuing material liberation accumulated to a predetermined damage threshold or cage failure, whichever occurred first

Prognostic Fusion for Uncertainty Reduction

* Work was performed while the author was with GE Global Research Abstract — One of the most critical aspects of prognostics is uncertainty management. That is, the calculation and processing of uncertainty sources that will ultimately allow the assignment of a probability density function (PDF) around the estimate of remaining useful life. This is of prime importance because it will give the decision maker a handle on when to take action based on the current risk tolerance. Ideally, the PDF is narrow such that a risk cut-off is close to the estimate. Accounting for the different uncertainty sources and stacking them up is a key in the proper quantification of the spread of the estimate. This paper investigates how the fusion of two different estimators can aid in the uncertainty management process. The fusion method aggregates two prediction algorithm output PDFs. The fusion algorithm is based on a kernel-based regression through time that discounts events distant in time from the time currently being evaluated. First, raw prediction PDFs are scaled by the individual quality estimates (subjective quality estimates are used here). Next, the PDFs are combined using kernel regression and are then normalized. The resulting spread of the fused PDF is smaller than the original ones at the same level of risk. We present results that use data obtained from bearing rig test data. The results support the premise that prediction output fusion can lead to better performance.

Towards an integrated reasoner for bearings prognostics

This paper describes an architecture and elements for an integrated prognostic bearings reasoner. The goal of the reasoner is to arrive at a reliable measure of damage accumulation, quantify its confidence, and aid in the assessment of remaining life. The reasoning integrates inflight and post-flight functions. During flight, the tasks are primarily diagnostic and assess damage in real time using input from a plurality of sources, both sensor-based and model-based. The damage assessment is further refined after conclusion of the flight with full-order models and additional information from historical failure data, operational data, and inspection data. Load profiles from future missions are used to calculate the damage propagation which will allow the reasoner to assess remaining life. This paper lays out the overall process and then focus in more detail on the in-flight reasoner. The operation of the architecture is demonstrated for bearing prognosis via an illustrative example

Towards In-Flight Detection and Accommodation of Faults in Aircraft Engines

To effectively accommodate safety-critical faults in-flight it is necessary to rapidly detect them and to have a means to accommodate the fault. We present results on model-based fault detection using sensor residuals from an extended Kalman filter with an embedded real-time engine model to characterize un-faulted behavior over the flight envelope. Thereafter, we present an approach for online fault accommodation via optimal changes in a set of suitable adjustments in the existing FADEC control logic. These optimal adjustments are obtained through off-line optimization for recovery of stall margins and thrust, then interpolated online for the existing flight conditions. We present results of the fault detection & accommodation applied to a high-bypass commercial aircraft engine over the flight envelope.

Endpoint-finding algorithm for structured light metrology of large objects in cluttered environments

Imaging of objects under linear structured-light illumination is commonly performed to ascertain location, dimensions, and morphology. The method is particularly attractive for prismatic objects distinguished by a single flat surface, because the processing algorithm can focus on purely linear features, and need only search for abrupt discontinuities in those features. In this paper, we consider a steel mill gantry crane application, wherein the targets to be localized are large steel slabs illuminated by laser line projectors, imaged from very close range to preserve spatial resolution. Conventional Radon transform- based methods for locating target edges yield highly ambiguous results, due to an abundance of image clutter coupled with the limited view of the overall object. This paper presents a modified Radon transform-based method for locating the target edges defined by the laser lines, and makes use of coarse a priori knowledge of the line pattern to resolve ambiguous edge points caused by neighboring clutter. A one-dimensional derivative of the Radon transform is used to improve the localization of short line segments. The algorithm is robust to the case in which the target object appears off axis and the clutter dominates the field of view.

Contour and Angle-Function Based Scoliosis Monitoring: Relaxing the Requirement on Image Quality in the Measurement of Spinal Curvature

Purpose A method for measuring spinal curvature that provides a useful analog to the Cobb angle and is tolerant of degraded image quality is proposed. Conventional methods require a higher standard of discernibility for vertebra features and suffer high variability. Methods Assumption is made that the natural representation of the spine for the purpose of scoliosis monitoring is that of a continuous curved contour rather than a series of discrete vertebral bodies with individual orientations. The angle that a tangent line to this contour makes with the vertical, expressed as a continuous function of height, is proposed as a metric for characterization of the curve. The Cobb angle can be approximated as the difference between the extrema of this function, and details of the function shape can provide additional markers for tracking curve variation and evolution. A method for deriving the angle function from coronal images of the spine is proposed, and both manual and automatic variants of the procedure are described. Results The method is applied to conventional coronal radiographs and to magnetic resonance (MR) coronal views derived from volumetric acquisitions of the spine. Included in the latter category is an image exhibiting poor discrimination of vertebra features due to motion artifacts. The method permits extraction of the curve and Cobb angles in all cases. Conclusions Because the spine contour is discernible even in low quality images where vertebral endplates may be obscured or poorly contrasted from surrounding tissue, the approach offers improved reliability, applicability across imaging modalities, and, in the case of x-rays, the possibility of a reduced radiation dose. Moreover, since it relies on larger image features and exploits the continuity of the spine, the contour-based approach is expected to reduce the variability associated with Cobb angle measurement.

Computer Assisted Cobb Angle Measurements: A novel algorithm

Background The standard for evaluating scoliosis is PA radiographs using Cobb angle to measure curve magnitude. Newer PACS systems allow easier Cobb angle calculations, but have not improved inter/intra observer precision of measurement. Cobb angle and its progression are important to determine treatment; therefore, angle variability is not optimal. This study seeks to demonstrate that a performance equivalent to that achieved in the manual method is possible using a novel computer algorithm with limited user input. The authors compared Cobb angles from predetermined spinal levels in the average attending score versus the computer assisted approach. Methods Retrospective analysis of PA radiographs from 58 patients previously evaluated for scoliosis was collected. Predesignated spinal levels (e.g., T2-T10) were assigned for different curves and calculated by Cobb method. Four spine surgeons evaluated these Cobb angles. Their average scores were measured and compared to formulated values using the novel computer-based algorithm. Literature reports inter-observer reliability is 6.3-7.2degrees. Limits of accuracy were set at 5 degrees of average orthopedic surgeons’ score. Results The computer-based algorithm calculated Cobb angles within 5 degrees of orthopedic surgeons’ average with a standard deviation of 3.2 degrees. This result was based on a 95% confidence interval with p values <0.001. The computer algorithm was plotted against average angle determined by the surgeons, with individual determinations and linear regression (r2 =0.90). The average difference between surgeons’ measures and computer algorithm was 0.4 degrees(SD= 3.2degrees, n=79). There was a tendency for the computer algorithm program to overestimate the angle at larger angles, but difference was small with r2 = 0.09. Conclusions Our study showed the novel computer based algorithm was an efficient and reliable method to assess scoliotic curvature in the coronal plane with the possibility of expediting clinic visits, ensuring reliability of calculation and decreasing patient exposure to radiation. Level of Evidence: III.