1,721,469 research outputs found
A non-gaussian continuous state space model for asset degradation
No full textThe degradation model plays an essential role in asset life prediction and condition based maintenance. Various degradation models have been proposed. Within these models, the state space model has the ability to combine degradation data and failure event data. The state space model is also an effective approach to deal with the multiple observations and missing data issues. Using the state space degradation model, the deterioration process of assets is presented by a system state process which can be revealed by a sequence of observations. Current research largely assumes that the underlying system development process is discrete in time or states. Although some models have been developed to consider continuous time and space, these state space models are based on the Wiener process with the Gaussian assumption. This paper proposes a Gamma-based state space degradation model in order to remove the Gaussian assumption. Both condition monitoring observations and failure events are considered in the model so as to improve the accuracy of asset life prediction. A simulation study is carried out to illustrate the application procedure of the proposed model
Feature group optimization for machinery fault diagnosis based on fuzzy measures
No full textWith the development of modern multi-sensor based data acquisition technology often used with advanced signal processing techniques, more and more features are being extracted for the purposes of fault diagnostics and prognostics of machinery integrity. Applying multiple features can enhance the condition monitoring capability and improve the fault diagnosis accuracy. However, an excessive number of features also increases the complexity of the data analysis task and often increases the time associated with the analysis process. A method of bringing some efficiency into this process is to choose the most sensitive feature subset instead. Fuzzy measures are helpful in this regard and have the ability to represent the importance and interactions among different criteria. Based on fuzzy measure theory, a feature selection approach for machinery fault diagnosis is presented in this paper. A heuristic least mean square algorithm is adopted to identify the fuzzy measures using training data set. Shapley values with respect to the fuzzy measures are applied as importance indexes to help choose the most sensitive features from a set of features. Interaction indexes with respect to the fuzzy measures are then employed to remove the redundant features. Vibration signals from a rolling element bearing test rig are used to validate the method. The results show that the proposed feature selection approach based on fuzzy measures is effective for fault diagnosis
Latent degradation indicator estimation using condition monitoring information
No full textAsset health prediction is imperative to optimal asset management. Online and offline inspections can provide useful information for predicting asset health. The information from an asset health inspection can be divided into two types. (1) Direct indicators which directly determine failures (e.g. the thickness of a brake pad, or the wear in a component) and (2) indirect indicators which are not related to failures directly (e.g. vibration signals or oil analysis results). The direct indicators can provide more precise reference for the maintenance strategy determination. However, these direct degradation indicators are often technically or economically impossible to inspect frequently and accurately. The indirect indicators, on the other hand, can be acquired more easily using various condition monitoring techniques. This paper proposes two continuous state space models to estimate and predict direct degradation indicators using indirect degradation indicators. The two continuous state space models adopt the Wiener process and the Gamma process respectively. The Expectation Maximization (EM) algorithms based on the modified Kalman smoother and the modified particle smoother are used to estimate the parameters of the proposed models. The application process of the EM algorithms and the characteristics of the state space models are illuminated through a simulation study. Finally, a case study using the data from an accelerated test of a gear box is conducted to justify the feasibility of the proposed models
Signal patterns of piston slap of a four-cylinder diesel engine
No full textThis paper presents an experimental study on the vibration signal patterns associated with a simulated piston slap test of a four-cylinder diesel engine. It is found that a simulated worn-off piston results in an increase in vibration RMS peak amplitudes associated with the major mechanical events of the corresponding cylinder (i.e., inlet and exhaust valve closing and combustion of Cylinder 1). This then led to an increase of overall vibration amplitude of the time domain statistical features such as RMS, Crest Factor, Skewness and Kurtosis in all loading conditions. The simulated worn-off piston not only increased the impact amplitude of piston slap during the engine combustion, it also produced a distinct impulse response during the air induction stroke of the cylinder attributing to an increase of lateral impact force as a result of piston reciprocating motion and the increased clearance between the worn-off piston and the cylinder. The unique signal patterns of piston slap disclosed in this paper can be utilized to assist in the development of condition monitoring tools for automated diagnosis of similar diesel engine faults in practical applications
Utilising Reliability and Condition Monitoring Data for Asset Health Prognosis
No full textThe ability to forecast machinery health is vital to reducing maintenance costs, operation downtime and safety hazards. Recent advances in condition monitoring technologies have given rise to a number of prognostic models which attempt to forecast machinery health based on condition data such as vibration measurements. This paper demonstrates how the population characteristics and condition monitoring data (both complete and suspended) of historical items can be integrated for training an intelligent agent to predict asset health multiple steps ahead. The model consists of a feed-forward neural network whose training targets are asset survival probabilities estimated using a variation of the Kaplan–Meier estimator and a degradation-based failure probability density function estimator. The trained network is capable of estimating the future survival probabilities when a series of asset condition readings are inputted. The output survival probabilities collectively form an estimated survival curve. Pump data from a pulp and paper mill were used for model validation and comparison. The results indicate that the proposed model can predict more accurately as well as further ahead than similar models which neglect population characteristics and suspended data. This work presents a compelling concept for longer-range fault prognosis utilising available information more fully and accurately
Optimising preventive maintenance strategy for production Lines
No full textPreventive Maintenance (PM) is often applied to improve the reliability of production lines. A Split System Approach (SSA) based methodology is presented to assist in making optimal PM decisions for serial production lines. The methodology treats a production line as a complex series system with multiple (imperfect) PM actions over multiple intervals. The conditional and overall reliability of the entire production line over these multiple PM intervals are hierarchically calculated using SSA, and provide a foundation for cost analysis. Both risk-related cost and maintenance-related cost are factored into the methodology as either deterministic or random variables. This SSA based methodology enables Asset Management (AM) decisions to be optimised considering a variety of factors including failure probability, failure cost, maintenance cost, PM performance, and the type of PM strategy. The application of this new methodology and an evaluation of the effects of these factors on PM decisions are demonstrated using an example. The results of this work show that the performance of a PM strategy can be measured by its Total Expected Cost Index (TECI). The optimal PM interval is dependent on TECI, PM performance and types of PM strategies. These factors are interrelated. Generally, it was found that a trade-off between reliability and the number of PM actions needs to be made so that one can minimise Total Expected Cost (TEC) for asset maintenance.\u
Improving asset management process modelling and integration\ud
No full textAsset management (AM) processes play an important role in assisting enterprises to manage their assets more efficiently. To visualise and improve AM processes, the processes need to be modelled using certain process modelling methodologies. Understanding the requirements for AM process modelling is essential for selecting or developing effective AM process modelling methodologies. However, little research has been done on analysing the requirements. This paper attempts to fill this gap by investigating the features of AM processes. It is concluded that AM process modelling requires intuitive representation of its processes, ‘fast’ implementation of the process modelling, effective evaluation of the processes and sound system integration.\u
Statistical condition monitoring based on vibration signals
No full textDesigning control limits for condition monitoring is an important aspect of setting maintenance schedules and has been virtually ignored by researchers to date. This paper proposes a novel statistical process control tool, the Weighted Loss function CUSUM (WLC) chart, for the detection of condition variation. The control limit was designed using baseline condition data, where the process was fitted by an autoregressive model and the residuals were used as the chart statistic. The condition variation is reflected by the changes of mean and variance of the statistic’s distribution against baseline condition, which can be detected by a single WLC chart. The approach was evaluated using a case study which showed that the chart can detect faulty conditions as well as their severity. The proposed approach has the advantage of requiring healthy baseline data only for the design of condition classifiers. It is applicable in numerous practical situations where data from faulty conditions are unavailable
- …
