Showing papers on "Condition monitoring published in 1994"
02 Oct 1994
TL;DR: In this article, a new method for online induction motor fault detection is presented, which utilizes artificial neural networks to learn the spectral characteristics of a good motor operating online, which may contain many harmonics due to the load which correspond to normal operating conditions.
Abstract: A new method for online induction motor fault detection is presented in this paper. This system utilizes artificial neural networks to learn the spectral characteristics of a good motor operating online. This learned spectrum may contain many harmonics due to the load which correspond to normal operating conditions. In order to reduce the number of harmonics which are continuously monitored to a manageable number, a selective frequency filter is employed. This frequency filter only passes those harmonics which are known to be of importance in fault detection, or which are continuously above a set level, to a neural net clustering algorithm. After a sufficient training period, the neural network signals a potential failure condition when a new cluster is formed and persists for some time. Since a fault condition is found by comparison to a prior condition of the machine, online failure prediction is possible with this system without requiring information on the motor or load characteristics. The detection algorithm was implemented and its performance verified on various fault types.
316 citations
TL;DR: The area of condition monitoring and fault diagnosis is being seen as of increasing importance and all available techniques have their drawbacks, all are not absolute and there is a plea for more information as to how real faults develop and exhibit themselves in measured parameters.
Abstract: The main theme of the paper is a description of the activity around the world in the way of research, development and application of techniques for condition monitoring and fault diagnosis of machine tools. The paper initially discusses the necessity for planned maintenance, the extension of this into condition based maintenance and the necessity for condition monitoring. It then discusses some definitions relating to this field of activity and in particular differentiates between hard and soft faults and the reason the latter can be used for prediction, whereas the former is easier to diagnose. The paper endeavours intentionally to restrict itself to the condition monitoring and fault diagnostics of the machine tool itself. Thus there is no real discussion of research into tool or process monitoring—although many areas of this type of work cut across that of the machine itself. For this reason there are notes on area of tool condition monitoring where these have techniques and expertise which lend themselves to the machine itself. There is a discussion on on-line and off-line monitoring followed by a brief survey of the automatic monitoring presently available on machine tools. Much of this is related to hard fault diagnostics but there are signs that soft faults are now being monitored. The paper then describes methods for the choice of parameters for condition monitoring and their data acquisition. Two large sections then follow on fault monitoring and diagnosis of both hard and soft faults where much reference is made to recently completed research under contracts ESPRIT 504 and SERC GR/E/12818. These sections also include references to ongoing research and EC contracts, including application of neural networks, expert systems and fuzzy logic. The paper concludes that the area of condition monitoring and fault diagnosis is being seen as of increasing importance. All available techniques have their drawbacks, all are not absolute and there is a plea for more information as to how real faults develop and exhibit themselves in measured parameters.
199 citations
Book•
01 Jan 1994
TL;DR: In this paper, the authors present a survey of condition-based maintenance information systems in the manufacturing domain, focusing on the aspects of maintenance, including basic diagnostic techniques, vibration monitoring, and particle monitoring.
Abstract: 1.Introduction - aspects of maintenance. 2. Maintenance information systems. 3. Basic diagnostic techniques. 4. Vibration monitoring. 5. Fluid condition and particle monitoring. 6. Systems approach to condition monitoring. 7. Application of system condition monitoring. 8. Processing of diagnostics data. 9. Future developments of condition-based maintenance in manufacture. Index
143 citations
01 Nov 1994
TL;DR: The results of applying KFM to condition monitoring of electrical drives reveal the practical advantages of unsupervised systems, which include the ability to learn and produce classifications without supervision.
Abstract: The feasibility of using an artificial network for identifying faults in induction motors has been demonstrated previously by the authors. In this work, the network was used as a learning and pattern recognition device, and was able to successfully associate input signal patterns with appropriate machine states. The neural network used was the multilayered perceptron (MLP), trained by a backpropagation algorithm. However, MLP lacks flexibility since it requires fully labelled input-output pairs (i.e. training of the network is supervised). This limitation can be removed by the use of an alternative approach, using unsupervised methods, such as the Kohonen feature maps (KFM) technique. The results of applying KFM to condition monitoring of electrical drives are reported in this paper, and they reveal the practical advantages of unsupervised systems, which include the ability to learn and produce classifications without supervision. Because of the natural parallel architecture of neural networks, they are also ideally suited to the use of multiple transducer inputs, which can greatly enhance the reliability of decisions made regarding the state of machine performance or condition.
66 citations
TL;DR: The TIGER ESPRIT Project is investigating a variety of tools and techniques for real-time situation assessment of dynamic systems, including a high-speed rule-based compiler, a situation assessment tool able to monitor dynamic responses over time, qualitative and numerical prediction of turbine response, and model-based diagnosis to help identify unknown faults.
Abstract: Industry is dominated by dynamic systems such as process control of chemical plants or steel plants and power generation, ranging to areas such as ecological systems and traffic systems. A challenging problem is to perform real-time monitoring and diagnosis of such dynamic systems. The paper describes the work of the TIGER ESPRIT Project in performing real-time situation assessment of dynamic systems. The goal of TIGER is to monitor a complex dynamic system in real time and make an assessment of whether it is working properly. If it is not working properly, it is desirable to identify the cause of the problem either from a set of known faults or to characterise an unknown fault. This work is applied in the application domain of condition monitoring of gas turbines. The project is developing and demonstrating its techniques on two industrial gas turbines; a 28 mW gas turbine used as the prime power for a chemical plant and a small auxiliary power gas turbine used in aviation. The Project is investigating a variety of tools and techniques for real-time situation assessment. These include a high-speed rule-based compiler, a situation assessment tool able to monitor dynamic responses over time, qualitative and numerical prediction of turbine response, and model-based diagnosis to help identify unknown faults. A prototype system is currently in use at a chemical plant performing monitoring and diagnosis in real-time.
55 citations
05 Dec 1994
TL;DR: Time domain reflectometry has been used for diagnostic monitoring of a large size photovoltaic (PV) giant in operating conditions and was able to detect, identify and localise the most common fault conditions, such as breaks of the circuit, insulation defects, wiring anomalies.
Abstract: Time domain reflectometry has been used for diagnostic monitoring of a large size photovoltaic (PV) giant in operating conditions. By analysing the waveforms obtained when a step-voltage excitation is propagated down the electrical line connecting the PV generators to the inverter, we were able to detect, identify and localise the most common fault conditions, such as breaks of the circuit, insulation defects, wiring anomalies. Best compromise between analysis time and precision was achieved by testing groups of three paralleled strings (panels): in this way we were able to detect in a short time whether a fault was present, while its position was separately determined with a good precision by repeating the test on the single strings. The whole 1 MW PV plant was tested in a couple of days. Finally, we discuss some economical aspects of this fault finding technique.
50 citations
TL;DR: In this paper, a method of simulation of the performance of jet engines, with the possibility of adapting to engine particularities, is presented, which employs an adaptation procedure coupled to a performance model solving the component matching problem.
Abstract: A method of simulation of the performance of jet engines, with the possibility of adapting to engine particularities, is presented. It employs an adaptation procedure coupled to a performance model solving the component matching problem. The proposed method can provide accurate simulation for engines of the same type, with differences that are due to manufacturing or assembly tolerances. It does not require accurate component maps, because they are derived during the adaptation procedure. It can also be used for health monitoring purposes, for component fault identification, and condition assessment. The effectiveness of the proposed method is demonstrated by application to two commercial jet engines.
38 citations
TL;DR: In this article, the authors used neural networks to identify useful patterns and trends in the vibration signals of the rotating components of the mining and petrochemical industries, which can be used to predict the mechanical state of a machine and its various components.
Abstract: This paper reports on research Aquila Mining Systems Ltd. conducted by J.H. Burrows Electronic Inc. in relation to the vibration condition monitoring of rotating equipment used in the mining and petrochemical industries. Using historical and real time vibration data monitored from compressors, pumps and electric motors, various approaches were designed and evaluated to extract and identify useful patterns and trends in the vibration signals of the rotating components of these machines. Efforts were focused on establishing whether the observed trends could be classified into distinct categories which would be indicative of the mechanical state of the equipment. Subsequent work will examine the feasibility of on-line prediction of component wear that could lead to preventive maintenance in advance of complete and catastrophic failure. Towards this end, data experimentation with neural networks will be undertaken to examine their applicability in accurately and reliably predicting the mechanical state of a machine and its various components.
36 citations
01 Jan 1994
TL;DR: The required communication circuit for realizing data transmission for both the ordinary 60-Hz power line and the special PWM (pulse-width-modulated) inverter-fed power line is constructed.
Abstract: A new method for condition monitoring of an electrical machine is proposed. The method uses the power leads to the machine itself as the communication link between the sending station located within the machine and the receiving station located remotely outside the machine. The required communication circuit to realize data transmission in both the cases of the ordinary 60 Hz power line and the special PWM inveter-fed power line is then constructed. The communication circuit uses an asynchronous serial communication protocol and an FSK modulation for realizing frequency multiplexing in the power line. An on-line winding temperature monitoring system for an inverter-fed induction machine is constructed using this power line communication link
36 citations
Journal Article•
TL;DR: In this paper, the authors describe some of the technical literature regarding engine-oil condition monitoring, including oil analyses techniques and the significance of various analytical results and limit criteria for oil analysis results (that is, values above which there is risk of engine problems).
Abstract: This article describes some of the technical literature regarding engine-oil condition monitoring, including oil analyses techniques and the significance of various analytical results. Limiting criteria for oil analysis results (that is, values above which there is risk of engine problems) are tabulated. Several approaches for monitoring oil condition are described: oil sampling followed by oil analysis, creation of models to predict various aspects of oil degradation, and direct on-board sensing of oil condition. This article is a shortened version of the chapter on {open_quotes}Automotive Engine-Oil Condition Monitoring{close_quotes} from the 1994 Handbook of Lubrication and Tribology, Volume 111, available from the Society of Tribologists and Lubrication Engineers. 1 ref., 1 tab.
31 citations
11 Apr 1994
TL;DR: A prior probability model is developed for spacecraft and receiver anomalies based on non-ideal failure models and the uncertainty present in their failure distribution parameters to suggest that the assumptions of traditional RAIM may be partially invalid.
Abstract: Current methods for GPS receiver autonomous integrity monitoring are limited by the assumptions they make. Using published studies of navigation system reliability, this paper develops a prior probability model for spacecraft and receiver anomalies based on non-ideal failure models and the uncertainty present in their failure distribution parameters. With this model, the thresholds for a residuals test statistic are found to optimize an arbitrary objective function based on relative costs for false alarm and missed detection errors. The outputs of Monte Carlo simulations allow thresholds to be computed for each geometry case using trial-and-error optimization. The simulation outputs suggest that the assumptions of traditional RAIM may be partially invalid. These results are useful for both snapshot RAIM tests as well as multi-step integrity algorithms which use Bayesian updating to generate posterior failure probabilities. Multi-step algorithms may be a valuable addition to the future GIC integrity structure. >
TL;DR: In this paper, a decision-based approach to condition-based maintenance management of rotating machinery is introduced and illustrated by formulating and solving a multiple objective maintenance management problem for a 15 MW industrial gas turbine.
Abstract: A decision-based approach to condition-based maintenance management of rotating machinery is introduced and illustrated by formulating and solving a multiple objective maintenance management problem for a 15 MW industrial gas turbine. The compromise Decision Support Problem approach is used because it provides a convenient way of incorporating both information from condition monitoring and considerations of factors such as machine degradation, operating cost (fuel cost), production loss, maintenance cost, environmental protection, machine availability, etc. The focus in this paper is on explaining the approach rather than on the results per se.
21 Jun 1994
TL;DR: An algorithm is described that, given a safe timed Petri net model of the monitored software, can determine the uncorrupted timestamp values, i.e. those that would have been observed had the delays not been present.
Abstract: Execution monitoring plays a central role in most software development tools for parallel and distributed computer systems. However, such monitoring may induce delays that corrupt event timing. If this corruption can be quantified, it may be possible to determine the intrusion-free behavior. In this paper, we describe an algorithm that, given a safe timed Petri net model of the monitored software, can determine the uncorrupted timestamp values, i.e. those that would have been observed had the delays not been present. Monitoring conditions sufficient to ensure correct operation of the algorithm, and examples illustrating the algorithm's applicability to message-passing systems are also presented. This work is part of a larger effort aimed at identifying cost-effective software alternatives to custom hardware monitoring. >
25 Oct 1994
TL;DR: The authors propose the use of a frame-based system in which the data in each frame is modeling as a multiscale stochastic process, where each frame of features is modeled as a sample of a multivariate, multimodal distribution.
Abstract: Presents results on using a statistical model motivated by the wavelet transform to represent non-stationary signals typically encountered in machinery monitoring applications. The authors propose the use of a frame-based system in which the data in each frame is modeled as a multiscale stochastic process. The parameters of a multiscale model are used as features for each frame, where each frame of features is modeled as a sample of a multivariate, multimodal distribution. Classification of machine states based on monitoring signals is performed by comparing likelihood scores for each machine state. The authors present an example of applying the system to data consisting of a superposition of damped sinusoids, as a way of illustrating system performance for the case of transient monitoring signals. They compare their system to one which is trained using a DFT-based (non-time-frequency-based) representation (in particular, LPC coefficients) and show that their system exhibits both superior performance as well as greater robustness to noise in the signals. They also compare results using multiscale parameters versus LPC coefficients for the case of synthesized autoregressive signals and for the case of actual, measured signals from a weld depth monitoring system. >
TL;DR: In this article, the authors provide an overview of the use of both performance and mechanical transient analysis as a means to detect gas turbine problems, and a discussion of how transient analysis can be integrated within an existing on-line monitoring system.
Abstract: Most engine health monitoring systems used for land based systems are based on steady state operation. Diagnostic analysis has traditionally been conducted under steady state conditions with on-line systems tending to concentrate on map based performance diagnostics using pattern analysis, fault matrices or expert systems. Transient analysis is a relatively new technique and is being applied to some aeroengines. There is significant diagnostic content in turbine startup and shutdown data and in data obtained during power or speed changes. This data can be captured if an automatic on-line system is employed. This paper provides an overview of the use of both performance and mechanical transient analysis as a means to detect gas turbine problems. The paper covers the need for transient analysis and covers transient analysis techniques. Examples and a discussion of how transient analysis can be integrated within an existing on-line monitoring system is made.
19 Apr 1994
TL;DR: This work identifies different features which seem to contain tool wear information, document what they found to be superior signal processing tools to identify, extract and process these non-stationary features, and stresses the need for a fully annotated public-domain manufacturing signal database.
Abstract: We address the general problem of reliable, real-time detection of faults in metal-removal processes in manufacturing. As has long been recognized by skilled machine operators, mechanical and acoustic vibrations can be reliable sources of cues for such monitoring. However, conventional dull-tool monitoring systems, which are generally based on stationary signal processing methods, are inadequate for real-time control of drilling procedure. Making use of a database from nine different drill bits, we (a) identify different features which seem to contain tool wear information, (b) document what we found to be superior signal processing tools to identify, extract and process these non-stationary features, and (c) stress the need for a fully annotated public-domain manufacturing signal database. >
Book•
20 Apr 1994
TL;DR: Part 1 Machine tool monitoring systems: feed force sensors current/power sensors pressure sensors accelerometers torque-controlled machining tool setting system and discussion: sensors condition monitoring summary.
Abstract: Part 1 Machine tool monitoring systems: feed force sensors current/power sensors pressure sensors accelerometers torque-controlled machining tool setting system. Part 2 Identification systems: probing systems identification systems using a read-write chip identification systems using proximity sensors. Part 3 Non-contact vision systems: technology and information processing video camera video grey scales colour vision systems vision sensor software lasers other noncontact sensors. Part 4 Miscellaneous sensing devices for industrial use: pressure sensors inclinometers velocity and displacement sensors fluid level sensors sensor control units. Part 5 Condition monitoring and predictive maintenance: predictive maintenance and sensors maintenance methods the decision to monitor common monitoring techniques online or off-line monitoring? data collection condition monitoring/predictive maintenance as used in industry. Part 6 Discussion: sensors condition monitoring summary.
Journal Article•
TL;DR: In this paper, the use of fine (10 micron) filters in the F404 engines of the CF-18 aircraft has severely limited the usefulness of oil analysis techniques because the filters remove almost all debris of significant from the oil.
Abstract: In oil-wetted aircraft machinery fitted with coarse (>50 micron) filtration systems, sufficient fine particulate wear debris passes through the filter to enable meaningful spectrometric oil analysis (SOA) of the oil analysis (SOA) of the oil sample debris for off-line condition monitoring. The use of fine (10 micron) filters in the F404 engines of the CF-18 aircraft has severely limited the usefulness of oil analysis techniques because the filters remove almost all debris of significant from the oil. The Defense Research Establishment Pacific (DREP) has developed Filter Debris Analysis (FDA) as an alternative off-line monitoring procedure for the CF-18 F404 engines. In addition, FDA can supplement conventional oil analysis techniques for condition monitoring of machinery with coarse filtration by evaluating the accumulated wear debris generated between filter changes. The method has been very successful at predicting wear anomalies and is being adopted into the routine maintenance procedures for selected machinery. The main drawbacks of FDA at present are that it is time-consuming, it is manpower intensive, and the interpretation of debris levels is subjective. The authors are presently investigating the use of computer-aided analysis to interpret the debris found on filters with an ultimate goal of incorporating the FDA results intomore » an expert system for tracking the condition of oil-wetted equipment. Results of image capture and interpretation of FDA data are presented, and implications of these developments on condition monitoring of oil-wetted machinery are discussed.« less
05 Jun 1994
TL;DR: In this paper, the authors report on the development phase of an on-line continuous tan(/spl delta/) monitoring system, which combines data such as type of equipment, age of equipment and load profiles, with the objective of establishing a system to diagnose the condition of oil-paper insulation.
Abstract: The ageing and deterioration of oil-paper insulation in HV equipment is a matter of continuous concern. Explosions resulting in damage to property, loss of capital and human life, and the need for replacement has prompted the drive towards an improved transformer design and recognition of the need for insulation condition monitoring. Insulation testing is reasonably well defined for routine and type testing of HV equipment in laboratories. However diagnostic criteria and procedures for testing the insulation condition of equipment in service are not well established nor have they been standardised. Continuous online monitoring of insulation condition presents several advantages such as an increased probability of predicting insulation failure and increased safety. This paper reports on the development phase of an on-line continuous tan(/spl delta/) monitoring system. The criteria for such a system in terms of practical and conceptual ideas for in-service operation and future development are discussed. Results from laboratory tests and those from a system which has been installed in a substation demonstrate the concepts of absolute and relative tan(/spl delta/) measurement and the characteristic trends and patterns in the tan(/spl delta/) of oil-paper insulation. The approaches towards combining data such as type of equipment, age of equipment and load profiles, and methods of data evaluation are discussed with the objective of establishing a system to diagnose the condition of oil-paper insulation. >
24 Jan 1994
TL;DR: This paper presents analytic techniques to compute the reliability of a mission based on a component redundancy scenario and predicts accurate reliability values in a variety of scenarios.
Abstract: Component redundancy is used to improve reliability and to postpone maintenance in many system operations. In such system, a failure of a component may occur and the system may still be used in operation. Thus the system dispatch state may not be a fully-operational state. Some subsystems may employ monitoring to determine if a unit is operational or not at the beginning of a mission. Redundant components may be checked and repaired at the end of periodic intervals, called maintenance check times, to reduce the cost of maintenance. In such scheduled maintenance (SM) systems, one may be interested in determining the mission reliability for a particular mission. We, in this paper, present analytic techniques to compute the reliability of a mission based on this maintenance scenario. Our model predicts accurate reliability values in a variety of scenarios. We also compare our solution with most used "back-of-envelope" computation methods and show shortcomings of those approaches. Several examples are shown and we validate our results using tools HARP, HARP-PMS, HARP-SMS, and EHARP. >
Dissertation•
01 Aug 1994
27 Jun 1994
TL;DR: The work presented here forms part of a study into the application of self-learning networks to the complex field of machine condition monitoring, which involves a simple conversion of microphone TES acoustic data into a matrix of frequency of code occurrence which can be directly applied to an artificial neural network (ANN).
Abstract: The work presented here forms part of a study into the application of self-learning networks to the complex field of machine condition monitoring. There are already several methods by which machines can be automatically monitored, but the development of a simplified nonintrusive "intelligent" system would be advantageous. Some work has been undertaken on the application of time encoded speech (TES) to automatic speech recognition using neural networks. It seemed feasible to try a similar technique to classify the acoustic emissions of a mechanical object. Initial experimentation was carried out using the speech system on a diesel engine. However the implementation described here involves a simplified form of data application to that employed previously. It consists of a simple conversion of microphone TES acoustic data into a matrix of frequency of code occurrence which can be directly applied to an artificial neural network (ANN). >
TL;DR: In this paper, the problem of process fault detection is considered using a feature detection network topology to reduce the dimensionality of the problem and extract from the process data important attributes that indicate the presence of process malfunctions.
TL;DR: In this paper, the authors describe the equipment and processes utilized in the Microelectronics Manufacturing Science and Technology (MMST) program, and describe the processing methodology that was developed and followed in order to operate in this CIM environment and successfully execute an approximately 150 step 0.35 /spl mu/m CMOS process in less than 72 hours.
Abstract: This paper describes the equipment and processes utilized in the Microelectronics Manufacturing Science and Technology (MMST) program. The processes were carried out in a combination of testbeds (AVP, the TI designed and built Advanced Vacuum Processor) and commercial equipment, all in the single-wafer mode. All AVP processing was performed with the wafers in an inverted, face-down, configuration. All the processing equipment was connected to a Computer-Integrated Manufacturing (CIM) system, which both collected the designated data and communicated the process parameters from the CIM database to the particular processing unit. Where available, in situ sensors were utilized for monitoring the process parameters, with measurements made on a metrology die in the center of the wafer. Many of these processes were controlled by the model-based process control algorithms in the CIM system. Otherwise, the processes were controlled by standard statistical process control (SPC) methods. This paper emphasizes the processing methodology that was developed and followed in order to operate in this CIM environment and successfully execute an approximately 150 step 0.35 /spl mu/m CMOS process in less than 72 hours. >
TL;DR: A condition monitoring and diagnostic system for the solid fuel gasification process and implement the system in a standard digital automation system based on static non-linear models and statistical properties of measured signals is defined.
TL;DR: This paper will present an integrated technology for truck condition monitoring, troubleshooting and maintenance, and will discuss the problem definition, and functions and configurations of the IMSS along with its implementation.
Abstract: An intelligent maintenance support system (IMSS) for truck condition monitoring has been developed by Syncrude Canada Ltd. and the University of Alberta. The IMSS implements normal and abnormal condition monitoring and fault diagnosis as well as providing maintenance assistance. The IMSS uses the integrated distributed intelligent system architecture, which integrates two symbolic inference engines with an intelligent hypermedia system. The system can process data from the Vital Signs Monitor (VSMTM) developed by Marathon LeTourneau as well as information from the operators. This paper will present an integrated technology for truck condition monitoring, troubleshooting and maintenance, and will discuss the problem definition, and functions and configurations of the IMSS along with its implementation.
24 Jan 1994
TL;DR: Predictive maintenance techniques such as vibration analysis, thermography and oil analysis are being applied to Strategic Petroleum Reserve rotating equipment with the expectation of improving equipment reliability and availability while lowering maintenance costs.
Abstract: Careful monitoring of the condition of the engine can detect any strange noises (vibrations) that may indicate abnormal wear of the moving parts. Additionally, nondestructive testing such as oil analysis can be used to determine the extent of contamination in the lubricant. The composition of the contamination in the oil can indicate the source and extent of mechanical problems within the engine. Predictive maintenance techniques such as vibration analysis, thermography and oil analysis are being applied to Strategic Petroleum Reserve (SPR) rotating equipment with the expectation of improving equipment reliability and availability while lowering maintenance costs. The SPR predictive maintenance program is just getting started but expectations are running high. The benefits of such a program are expected to include a sizable reduction in the rework of rebuilt pumps and motors, a reduction of improper pump and motor alignment, and repair of worn bearings before catastrophic bearing failure causes related damage to other components of the equipment. These actions are expected to increase equipment reliability, and availability while saving money. >
TL;DR: In this paper, the first approach in integrating condition monitoring technology and multi-objective optimization for gas turbines is presented, and the results have shown that the construct has significant promise in managing machinery maintenance on an on-condition basis in a close to optimal manner.
Abstract: This paper provides what is, to our knowledge, the first approach in integrating condition monitoring technology and multiobjective optimization for gas turbines. In the past, the two topics have been treated separately and no coherent attempt has been made to unify these two powerful techniques together in a formal manner. This paper provides a construct utilizing the decision support problem technique developed at the University of Houston that can accomplish this. The background and needs for the creation of this construct are explored and a model derived for maintenance management of a 12 MW gas turbine. The results have shown that the construct has significant promise in managing machinery maintenance on an on-condition basis in a close to optimal manner. The ability to use expert systems for the modification of constraints and objective values is also covered.
Journal Article•
TL;DR: In this paper, a knowledge-based expert system is developed which will incorporate FDA results and offer advice to the used by following the logical reasoning of an experienced analyst in determining the type of metallic material and the wear condition of the machinery based on the color, morphological attributers, and surface texture of the wear particles.
Abstract: In oil-wetted aircraft machinery fitted with coarse (>50 micron) filtration systems, sufficient fine particulate wear debris can pass through the filter to enable meaningful spectrometric oil analysis (SOA) of the oil sample debris for off-line condition monitoring. The use of finer (10 {mu}m) filters in the F404 engines of the CF-18 aircraft has severely limited the usefulness of SOA because the filters remove almost all debris of significance from the oil. To acquire useful information from this trapped debris Filter Debris Analysis (FDA) has been developed as an alternative off-line monitoring procedure for the CF-18 F404 engines. In addition, FDA can supplement conventional oil analysis techniques for condition monitoring of coarse filtration machinery by evaluating the accumulated wear debris generated between filter changers. A knowledge based expert system is currently being developed which will incorporate FDA results and offer advice to the used by following the logical reasoning of an experienced analyst in determining the type of metallic material and the wear condition of the machinery based on the color, morphological attributers, and surface texture of the wear particles. Using this expert system, as a foundation, it is hoped that the other condition monitoring (CM) techniques will be assimilated tomore » provide the CM analyst with a tool to correlate data and results thereby increasing the utility of each individual method of analysis. 22 refs., 5 figs., 1 tab.« less
TL;DR: In this article, a balanced combination of neural networks and expert system techniques was used to improve the performance of the engine condition monitoring (ECM) in a supporting role. But the results showed the validity of the approach by successfully identifying certain engine problems.