Helicopter gearbox condition monitoring systems and methods

Helicopter Gearbox Condition Monitoring

The work of Vecer, Kreidl, and Smid (2005) entitled: "Condition Indicators for Gearbox Conditioning Monitoring Systems" states that condition monitoring systems are very important for researchers in gearbox development. They enable detection of gear cracks during testing and stop the test before the gear crack progresses. Then the researchers are able to recognize where the crack began and to decide about the reason for the gearbox fault." (Vecer, Kreidl, and Smid, 2005, p. 35) This enables the designers to "take appropriate steps in gearbox design to improve gearbox performance." (Vecer, Kreidl, and Smid, 2005, p.35) Condition monitoring systems are stated to deal with "various types of input data, for instance vibration, acoustic emission, temperature, oil debris analysis etc." (Vecer, Kreidl, and Smid, 2005, p.35)

Systems that are based on vibration analysis, acoustic emission and oil debris are stated to be the most common and are very well established in industry." (Vecer, Kreidl, and Smid, 2005, p.35) Systems based on acoustic emission are stated to have "a more obvious application for bearing monitoring than for gearing monitoring." (Vecer, Kreidl, and Smid, 2005, p.35) Some applications have been introduced for gearbox monitoring. Acoustic emission (AE) is generally defined as "transient elastic waves generated from a rapid release of strain energy caused a deformation or by damage within or on the surface of the material." (Vecer, Kreidl, and Smid, 2005, p. 35)

The work of Rogers is stated to provide a description of the application "for monitoring of a slowly rotating anti-friction slew bearing mounted in cranes for gas production." (Vecer, Kreidl, and Smid, 2005, p.36) It is reported that a very reliable method for detection of gearing damage in the earlier stages is oil debris analysis which also enables estimation of the level of wear. (Vecer, Kreidl, and Smid, 2005, paraphrased) It is reported that the contacting surfaces of the gearwheels and bearings during gearbox operations are "gradually abraded. Small pieces of material break down from the contact surfaces. These small pieces of material are carried away by oil lubricating the gearwheels and bearing. By detecting the number and size of particles in the oil" it is possible to identify "gear-pitting damage in an early stage, which is unidentifiable by vibration analysis. Oil debris sensors are generally based on magnetic or optical principles. The change in the magnetic field is measured by magnetic sensors and this change is stated to be caused by "metal particles in monitored samples of oil. The oil debris monitoring system can be off-line or online and it is held that there is more reliability in oil debris monitoring systems than in vibration-based systems in the early detection of pitting failure. (Dempsey, and Afjeh, 2002)

The work of Tapioa (1992) entitled: "Gear-Systems Condition Monitoring and Diagnostics Through Phase Domain Measurements and Analysis" reports the extensive analysis that has been conducted in relation to gearbox vibration and specifically notes the work of Nakada and Utagawa in which the "source of the gear meshing frequency is said to be produced by the deflections of the gear teeth as they mesh under load." (Tapioa, 1992, p. 12) it is reported that Honda and Conway researched theoretical modeling of geared systems which "concentrated on each individual tooth and how its strength was affected by the formation of a crack at its base..." And who state the conclusion that the "gear tooth stiffness was greatly lowered by the formation of a crack." (Tapioa, 1992, p. 12)

Dale (1984) is stated to have conducted one of the earliest studies that analyze gear meshing vibrations in what was an approach "taken to analyze the acquired data...through frequency domain analysis via FFT.." now the norm in the analysis of vibration. ( p.13) The work of Hauser (1985) provides an experimental procedure that "moves away from FFT analysis and suggest that small increments of the rotation should be isolated and analyzed individually." The work of McFadden is said to be mentionable as McFadden "suggests that the implementation of phase synchronous averaging of the signals to eliminate noise." (Tumer and Huff, 2003, p.14) It is reported that Jacob (1992) and Sharaf-Eldeen and Jacob "investigation the feasibility of using optical encoders to monitor gear train systems and compared the results with those obtained using accelerometer measurements of vibrations..." through use of frequency domain analysis." (Tumer and Huff, 2003, p.15)

Tumer and Huff (2003) in the work entitled" Analysis of Triaxial Vibration Data for Health Monitoring of Helicopter Gearboxes" states that research on the nature of vibration data that has been collected during flight experimentations form helicopter transmission "has led to several crucial observations believed to be responsible for the high rates of false alarms and missed detections in aircraft vibration monitoring systems." (Tumer and Huff, 2003, p. 120)

Tumer and Huff report the need to give consideration to other information sources of system vibrations and explore helicopter transmission vibration data and do so through using triaxial accelerometer collection in three different directions which are analyzed first for content and then in combination of Principal Components Analysis (PCA) for analysis of changes in directionality. Compared is the frequency of the three different direction and analysis is conducted through sue of time-synchronously averaged vibration data. Tumer and Huff states that the triaxial data are "decorrelated using a mathematical transformation and compared to the original axes to determine their differences." (Tumer and Huff, 2003, p.12)

The study results are stated to indicate that "triaxial accelerometers can provide additional information about the frequency content of helicopter gearbox vibrations and provide researchers and industry with a novel method of capturing and monitoring triaxial changes in the baseline vibration signatures." (Tumer and Huff, 2003, p.120)

The work of Lia, Liu, Shi and Zhang (2004) entitled: "Gearbox Condition Monitoring Using Self-Organizing Feature Maps" proposes a "novel technique for the condition monitoring of gearboxes based on self-organizing feature maps (SOFM) networking. In order to visualize the SOFM results used is a method that is stated to be improved and that is based "on the unified distance matrix (U-matrix) in which the overall topological information condensed into the map units is considered so as to project the high dimensional input vectors into a two-dimensional space and give a better picture of their intrinsic structure that the original U-matrix method." (Lia, Liu, Shi and Zhang, 2004, p.1)

The feature data is stated to be extracted "from the industrial gearbox vibration signals which are measured under different operating conditions with this proposed technique. Results are stated to show that "trained with the SOFM network and visualized with the improved method, the feature data are mapped into a two-dimensional space and formed clustering regions, each indicative of a specific gearbox condition. Therefore, the gearbox operating condition with a fatigue crack or broken tooth compared with the normal condition is identified clearly. Furthermore, with the trajectory of the image points for the feature data in two-dimensional space, the variation of gearbox conditions is observed visually, and the development of gearbox early-stage failures is monitored in time." (Lia, Liu, Shi and Zhang, 2004, p.1)

The work of Worden and Barton entitled: "Condition Monitoring" presented in a lecture entitled: "Intelligent Fault Detection in Systems and Structures" states that in the process of devising "an intelligent fault detection system the primary consideration is defining an unambiguous definition of damage as a prerequisite to providing a unified approach to damage evaluation across all the engineering disciplines." (Worden and Barton, 2003, p.3) There are stated to be four key multidisciplinary areas for which monitoring and assessing damage are principal concerns: (1) structural health monitoring (SHM); (2) condition monitoring (CM); (3) Non-destructive evaluation (NDE); and (4) Statistical Process Control (SPC). (2003, p.2) SHM is stated to be relevant to structures including aircraft and buildings and is stated to implies a sensor network that monitors the behavior of the structure online." (Worden and Barton, 2003, p.3) It is related that optical fibers are typical sensors used such as electrical resistance stain gauges or acoustic devices." (Worden and Barton, 2003, p.3) CM is stated to be relevant to "rotating and reciprocating machinery, such as used in manufacturing" and it is additionally stated that CM also uses vibration-based online techniques using accelerometers as sensors. NDE is stated to be typically carried offline following the location of damage with online sensors. Therefore, NDE is used mainly for "characterization and as a severity check when there is a priori knowledge of the location of the damage." (Worden and Barton, 2003, p.3) Typical techniques are stated to include those of: (1) ultrasound; (2) thermography; and (3) shearography. (Worden and Barton, 2003, p.3) It is related that SPC is process rather than structure-based using various sensors in monitoring process changes. (Worden and Barton, 2003, paraphrased)

Intelligent fault detection is stated to entail detection of the damage that when not corrected results in a fault. Detection is a method that provides a qualitative indication that there may be damage present in the structure while localization is a method that provides information about the "probable position of the damage." (Worden and Barton, 2003) Assessment is a method that provides an estimate of the extent of the damage and prediction is the method that offers information concerning the structural safety and that provides estimation of the residual life.

It is reported that many modern approaches to damage identification are "based on the idea of pattern recognition (PR)." (Worden and Barton, 2003) A PR algorithm is stated to be one that assigns to a sample of measured data a class label, usually from a finite set. In the case of damage identification, the measured data could be vibration modeshapes, full-field thermoelastic data, scattered wave profiles etc. The appropriate class labels would encode damage type, location etc. In order to carry out the higher levels of identification using PR, it will almost certainly be necessary to construct examples of data corresponding to each class." (Worden and Barton, 2003, p.3) It is additionally reported that the "holistic approach to fault identification "requires that the diagnostic system should be carefully designed with the objectives in mind." (Worden and Barton, 2003, p.3)