Showing papers on "Condition monitoring published in 1987"

Joining Petri nets and knowledge based systems for monitoring purposes

Abstract: The control of FMS is now currently performed by means of distributed and hierarchical systems and Petri nets have been proved useful to specify such systems. In this paper, we propose to cope with monitoring by proposing a hierarchical structuration of monitoring functions and to use jointly Petri nets (for detection and handling purposes) and a knowledge based approach (for diagnosis and decision) in an integrated way.

A Rotor Condition Monitor for Squirrel-Cage Induction Machines

Abstract: Rotor temperature is of concern in both short-term machine protection and in longer term condition monitoring systems for large induction machines. Unduly large temperatures can arise under severe operating conditions such as stall and overload. A cage rotor condition monitoring system using direct temperature measurement is described. Under running conditions, it continuously transmits temperature information about strategic points on the rotor via an optical link. Possible rotor circuit imperfections are broken bars and end rings, single or multiple, and the deterioration of bar/end ring joints. The possibility of using temperature information for the identification of these faults is also investigated.

Vehicle condition monitoring system

Abstract: A condition monitoring system for a vehicle comprises a central control unit and a plurality of remote data acquisition modules, each of the modules being connected to the central control unit over a single wire and having a plurality of inputs connected to local condition-sensing transducers. Each data acquisition module serves to store data representing signals received at its inputs. The central control unit prompts the modules in turn and each module, when prompted, transmits its stored data to the central control unit over the respective single wire. This system provides a simplification of the wiring and connectors required between the various transducers and the central control unit, which drives displays displaying the conditions of the transducers.

TEXMAS - an expert system for gas turbine engine diagnosis and more

Abstract: The Turbine Engine Expert Maintenance Advisor System, TEXMAS, is being implemented on the T53 in order to effect both engine condition monitoring and diagnosis. The monitoring of trends in measured parameters leads to prediction of component failures; diagnosis, or fault-isolation, deduces the defective component or system on the basis of monitoring data. TEXMAS uses the DIGR diagnostic reasoner expert system as the basis of its engine monitoring and diagnosis operations.

Experimental Implementation and Evaluation of the RMI Failure Detection Algorithm

Abstract: This paper describes laboratory experiments with the Reachable Measurement Intervals (RMI) failure detection algorithm for systems with imperfect models. It was implemented on a multiprocessor computer system and used to detect and Isolate failures in an aircraft simulator in the presence of modeling errors. The failure detection computer and the monitored system resembled a commercial system and provided a good vehicle for assessing the suitability of RMI for industrial applications. The algorithm performed very well with real hardware, matching its performance in simulations. It did so because it has been designed specifically to handle imperfect dynamic models. The algorithm is fast enough to monitor large industrial systems when implemented on a single-board computer. Its structure is suitable for parallel processing implementation which makes it fast enough for even very large systems.

An Adaptive Noise Cancellation Approach for Condition Monitoring of Gear Box Bearings

Abstract: This paper describes the practical implementation of the adaptive noise cancellation (ANC) technique, test instrumentation/data acquisition and accelerometer fixing designs and locations for condition monitoring of gearbox bearings. The effects of various accelerometer fixing designs for optimum detection of primary and reference inputs are described and tested with a seeded defect bearing located in a gear box housing corrupted by severe gear meshing noise. For a reasonable size defect, a design based on shear-block fixture shows considerable success in attenuating the bearing signal at the reference input. The ANC output signal using the above input and a primary input derived from a series of offset balls fixture shows considerable success in detecting a seeded defect-bearing signal corrupted by gear background noise.

Marine Gas Turbines - Engine Health Monitoring - New Approaches

Abstract: Recent developments, coupled with field experience of Engine Health Monitoring Techniques within the Royal Navy Gas Turbine Fleet have enabled a fresh initiative. The advantages of adopting a policy of Condition Based Maintenance rather than a rigid hours concept are outlined and the Engine Health Monitoring (EHM) developments and affects on overhaul facilities are explained.Copyright © 1987 by ASME

A Practical Method Of Machine Tool Condition Monitoring By Analysis Of Component Surface Finish Data

Abstract: Random process analysis of component surface finish data is used to establish the 'fingerprint' of the machine tool condition when applied to a particular machining operation. Vibrations occurring during the machining process can be determined and the nature of the vibration isolated. It will be shown that for a turning operation it is possible to distinguish among types of error, such as, vibrations and errors causing radial movement of the cutting tool, variation in feed rate, vertical vibration of the cutter or component, or worn bearings. Existing methods used for condition monitoring involve the use of expensive vibration analysers with skilled personnel to assess the results and make a judgement of the machine tool capability. This means that a machine must be taken off line to be checked and hence cannot be continually assessed. Random process analysis of the surface texture produced on the component permits condition monitoring and assessment of machine capability to be made during production runs. The control parameters are not based on an arbitary judgement but on maintaining an acceptable quality of component according to its specification. This method effectively closes the control loop closely around the component. It modifies the control parameters to meet the required precision for the component and assesses if the machine capability is acceptable.

Studies and analyses of the space shuttle main engine

Abstract: The primary objectives were to: evaluate ways to maximize the information yield from the current Space Shuttle Main Engine (SSME) condition monitoring sensors, identify additional sensors or monitoring capabilities which would significantly improve SSME data, and provide continuing support of the Main Engine Cost/Operations (MECO) model In the area of SSME condition monitoring, the principal tasks were a review of selected SSME failure data, a general survey of condition monitoring, and an evaluation of the current engine monitoring system A computerized data base was developed to assist in modeling engine failure information propagations Each of the above items is discussed in detail Also included is a brief discussion of the activities conducted in support of the MECO model

An Apparatus for Automatic Assessment of Machine Condition

Abstract: A new apparatus for automatic assessment of machine condition has been developed. The 'life oil' sample extracted from the operating machinery is analysed for wear particles suspended in the oil. The operation of the apparatus, based on the ferrography principle, is completely automated and controlled by a microprocessor. The operations such as calibration, assessment of the measurement conditions, offset, oil sample and sample volume measurements are controlled by a microcomputer. The results of the measurements are computed and displayed in the form of commonly used wear parameters. The results can also be conveniently stored in the memory of any compatible PC computer and utilized further for a condition monitoring program. The apparatus is also self-calibrating and self-testing, thus any possible human and apparatus faults are detected and displayed, and can be quickly rectified. The apparatus has been designed to process two oil samples simultaneously. The apparatus is a powerful machine condition-monitoring tool for many engineering applications. It is efficient, reliable and reducing human error to a minimum. In the paper the design details, measurement and operating principles of the apparatus are described and possible sources of measurement errors discussed.

Noncontacting measurement technologies for space propulsion condition monitoring

Abstract: This paper describes four noncontacting measurement technologies that can be used in a turbopump condition monitoring system. The isotope wear analyzer, fiberoptic deflectometer, brushless torque-meter, and fiberoptic pyrometer can be used to monitor component wear, bearing degradation, instantaneous shaft torque, and turbine blade cracking, respectively. A complete turbopump condition monitoring system including these four technologies could predict remaining component life, thus reducing engine operating costs and increasing reliability.

Condition Monitoring and the Maintenance Plan

Abstract: Condition monitoring is becoming more cost effective but should be selected only if cost effective relative to other maintenance activities and it should be considered as part of the total maintenance plan for all assets that are significantly interrelated. The way in which a component fails, the failure type, will have an effect on the maintenance activities that are appropriate for that component.

Wear-particle analysis: maturing tool for predictive-maintenance

Abstract: Wear-particle analysis is a technique for machinery condition monitoring through an empirical measurement of the relative amounts of ferrous materials in the machinery's lubricant. It is explained how this technique has become an economic tool for any size powerplant owing to the recent availability of service contracts for the periodic analysis of lubricating oil.

Model-Based Failure Diagnosis in Power Plants

Abstract: This paper describes methods, applications and experiences dealt with during the research project Failure Diagnosis and Condition Monitoring in Power Plants. The diagnosis algorithms developed are model-based, i.e. they are based on monitoring the process using process measurements and models describing the quantitative operation of the process. A model is here understood as being any description of the operation of the process and the relations between the measurements. The stucture of the model is chosen to suit the process monitored and its failure mechanisms in the best possible way: a dynamic or static mathematical model, a characteristic curve or a presentative value or function. The methods are also designed to be as fault selective as possible in order to make the location more effective.