Intelligent prognostics of machinery health utilising suspended condition monitoring data

The ability to forecast machinery failure 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 for forecasting machinery health based on condition data. Although these models have aided the advancement of the discipline, they have made only a limited contribution to developing an effective machinery health prognostic system. The literature review indicates that there is not yet a prognostic model that directly models and fully utilises suspended condition histories (which are very common in practice since organisations rarely allow their assets to run to failure); that effectively integrates population characteristics into prognostics for longer-range prediction in a probabilistic sense; which deduces the non-linear relationship between measured condition data and actual asset health; and which involves minimal assumptions and requirements. This work presents a novel approach to addressing the above-mentioned challenges. The proposed model consists of a feed-forward neural network, the training targets of which are asset survival probabilities estimated using a variation of the Kaplan-Meier estimator and a degradation-based failure probability density estimator. The adapted Kaplan-Meier estimator is able to model the actual survival status of individual failed units and estimate the survival probability of individual suspended units. The degradation-based failure probability density estimator, on the other hand, extracts population characteristics and computes conditional reliability from available condition histories instead of from reliability data. The estimated survival probability and the relevant condition histories are respectively presented as “training target” and “training input” to the neural network. The trained network is capable of estimating the future survival curve of a unit when a series of condition indices are inputted. Although the concept proposed may be applied to the prognosis of various machine components, rolling element bearings were chosen as the research object because rolling element bearing failure is one of the foremost causes of machinery breakdowns. Computer simulated and industry case study data were used to compare the prognostic performance of the proposed model and four control models, namely: two feed-forward neural networks with the same training function and structure as the proposed model, but neglected suspended histories; a time series prediction recurrent neural network; and a traditional Weibull distribution model. The results support the assertion that the proposed model performs better than the other four models and that it produces adaptive prediction outputs with useful representation of survival probabilities. This work presents a compelling concept for non-parametric data-driven prognosis, and for utilising available asset condition information more fully and accurately. It demonstrates that machinery health can indeed be forecasted. The proposed prognostic technique, together with ongoing advances in sensors and data-fusion techniques, and increasingly comprehensive databases of asset condition data, holds the promise for increased asset availability, maintenance cost effectiveness, operational safety and – ultimately – organisation competitiveness.

Reliability prediction using the non-parametric explicit hazard model : a case study

Survival probability prediction using covariate-based hazard approach is a known statistical methodology in engineering asset health management. We have previously reported the semi-parametric Explicit Hazard Model (EHM) which incorporates three types of information: population characteristics; condition indicators; and operating environment indicators for hazard prediction. This model assumes the baseline hazard has the form of the Weibull distribution. To avoid this assumption, this paper presents the non-parametric EHM which is a distribution-free covariate-based hazard model. In this paper, an application of the non-parametric EHM is demonstrated via a case study. In this case study, survival probabilities of a set of resistance elements using the non-parametric EHM are compared with the Weibull proportional hazard model and traditional Weibull model. The results show that the non-parametric EHM can effectively predict asset life using the condition indicator, operating environment indicator, and failure history.

Component reliability estimations without field data

Maintenance activities in a large-scale engineering system are usually scheduled according to the lifetimes of various components in order to ensure the overall reliability of the system. Lifetimes of components can be deduced by the corresponding probability distributions with parameters estimated from past failure data. While failure data of the components is not always readily available, the engineers have to be content with the primitive information from the manufacturers only, such as the mean and standard deviation of lifetime, to plan for the maintenance activities. In this paper, the moment-based piecewise polynomial model (MPPM) are proposed to estimate the parameters of the reliability probability distribution of the products when only the mean and standard deviation of the product lifetime are known. This method employs a group of polynomial functions to estimate the two parameters of the Weibull Distribution according to the mathematical relationship between the shape parameter of two-parameters Weibull Distribution and the ratio of mean and standard deviation. Tests are carried out to evaluate the validity and accuracy of the proposed methods with discussions on its suitability of applications. The proposed method is particularly useful for reliability-critical systems, such as railway and power systems, in which the maintenance activities are scheduled according to the expected lifetimes of the system components.

Asset health prediction using condition indicators and operating environment indicators : a case study

Prognostics and asset life prediction is one of research potentials in engineering asset health management. We previously developed the Explicit Hazard Model (EHM) to effectively and explicitly predict asset life using three types of information: population characteristics; condition indicators; and operating environment indicators. We have formerly studied the application of both the semi-parametric EHM and non-parametric EHM to the survival probability estimation in the reliability field. The survival time in these models is dependent not only upon the age of the asset monitored, but also upon the condition and operating environment information obtained. This paper is a further study of the semi-parametric and non-parametric EHMs to the hazard and residual life prediction of a set of resistance elements. The resistance elements were used as corrosion sensors for measuring the atmospheric corrosion rate in a laboratory experiment. In this paper, the estimated hazard of the resistance element using the semi-parametric EHM and the non-parametric EHM is compared to the traditional Weibull model and the Aalen Linear Regression Model (ALRM), respectively. Due to assuming a Weibull distribution in the baseline hazard of the semi-parametric EHM, the estimated hazard using this model is compared to the traditional Weibull model. The estimated hazard using the non-parametric EHM is compared to ALRM which is a well-known non-parametric covariate-based hazard model. At last, the predicted residual life of the resistance element using both EHMs is compared to the actual life data.

Quantitative studies of lower motor neuron degeneration in amyotrophic lateral sclerosis : Evidence for exponential decay of motor unit numbers and greatest rate of loss at the site of onset

Objective: To use our Bayesian method of motor unit number estimation (MUNE) to evaluate lower motor neuron degeneration in ALS. Methods: In subjects with ALS we performed serial MUNE studies. We examined the repeatability of the test and then determined whether the loss of MUs was fitted by an exponential or Weibull distribution. Results: The decline in motor unit (MU) numbers was well-fitted by an exponential decay curve. We calculated the half life of MUs in the abductor digiti minimi (ADM), abductor pollicis brevis (APB) and/or extensor digitorum brevis (EDB) muscles. The mean half life of the MUs of ADM muscle was greater than those of the APB or EDB muscles. The half-life of MUs was less in the ADM muscle of subjects with upper limb than in those with lower limb onset. Conclusions: The rate of loss of lower motor neurons in ALS is exponential, the motor units of the APB decay more quickly than those of the ADM muscle and the rate of loss of motor units is greater at the site of onset of disease. Significance: This shows that the Bayesian MUNE method is useful in following the course and exploring the clinical features of ALS. 2012 International Federation of Clinical Neurophysiology.

Security constrained economic dispatch considering wind energy conversion systems

The available wind power is stochastic and requires appropriate tools in the OPF model for economic and reliable power system operation. This paper exhibit the OPF formulation with factors involved in the intermittency of wind power. Weibull distribution is adopted to find the stochastic wind speed and power distribution. The reserve requirement is evaluated based on the wind distribution and risk of under/over estimation of the wind power. In addition, the Wind Energy Conversion System (WECS) is represented by Doubly Fed Induction Generator (DFIG) based wind farms. The reactive power capability for DFIG based wind farm is also analyzed. The study is performed on IEEE-30 bus system with wind farm located at different buses and with different wind profiles. Also the reactive power capacity to be installed in the wind farm to maintain a satisfactory voltage profile under the various wind flow scenario is demonstrated.

Asset health prediction using the explicit hazard model

The ability to estimate the asset reliability and the probability of failure is critical to reducing maintenance costs, operation downtime, and safety hazards. Predicting the survival time and the probability of failure in future time is an indispensable requirement in prognostics and asset health management. In traditional reliability models, the lifetime of an asset is estimated using failure event data, alone; however, statistically sufficient failure event data are often difficult to attain in real-life situations due to poor data management, effective preventive maintenance, and the small population of identical assets in use. Condition indicators and operating environment indicators are two types of covariate data that are normally obtained in addition to failure event and suspended data. These data contain significant information about the state and health of an asset. Condition indicators reflect the level of degradation of assets while operating environment indicators accelerate or decelerate the lifetime of assets. When these data are available, an alternative approach to the traditional reliability analysis is the modelling of condition indicators and operating environment indicators and their failure-generating mechanisms using a covariate-based hazard model. The literature review indicates that a number of covariate-based hazard models have been developed. All of these existing covariate-based hazard models were developed based on the principle theory of the Proportional Hazard Model (PHM). However, most of these models have not attracted much attention in the field of machinery prognostics. Moreover, due to the prominence of PHM, attempts at developing alternative models, to some extent, have been stifled, although a number of alternative models to PHM have been suggested. The existing covariate-based hazard models neglect to fully utilise three types of asset health information (including failure event data (i.e. observed and/or suspended), condition data, and operating environment data) into a model to have more effective hazard and reliability predictions. In addition, current research shows that condition indicators and operating environment indicators have different characteristics and they are non-homogeneous covariate data. Condition indicators act as response variables (or dependent variables) whereas operating environment indicators act as explanatory variables (or independent variables). However, these non-homogenous covariate data were modelled in the same way for hazard prediction in the existing covariate-based hazard models. The related and yet more imperative question is how both of these indicators should be effectively modelled and integrated into the covariate-based hazard model. This work presents a new approach for addressing the aforementioned challenges. The new covariate-based hazard model, which termed as Explicit Hazard Model (EHM), explicitly and effectively incorporates all three available asset health information into the modelling of hazard and reliability predictions and also drives the relationship between actual asset health and condition measurements as well as operating environment measurements. The theoretical development of the model and its parameter estimation method are demonstrated in this work. EHM assumes that the baseline hazard is a function of the both time and condition indicators. Condition indicators provide information about the health condition of an asset; therefore they update and reform the baseline hazard of EHM according to the health state of asset at given time t. Some examples of condition indicators are the vibration of rotating machinery, the level of metal particles in engine oil analysis, and wear in a component, to name but a few. Operating environment indicators in this model are failure accelerators and/or decelerators that are included in the covariate function of EHM and may increase or decrease the value of the hazard from the baseline hazard. These indicators caused by the environment in which an asset operates, and that have not been explicitly identified by the condition indicators (e.g. Loads, environmental stresses, and other dynamically changing environment factors). While the effects of operating environment indicators could be nought in EHM; condition indicators could emerge because these indicators are observed and measured as long as an asset is operational and survived. EHM has several advantages over the existing covariate-based hazard models. One is this model utilises three different sources of asset health data (i.e. population characteristics, condition indicators, and operating environment indicators) to effectively predict hazard and reliability. Another is that EHM explicitly investigates the relationship between condition and operating environment indicators associated with the hazard of an asset. Furthermore, the proportionality assumption, which most of the covariate-based hazard models suffer from it, does not exist in EHM. According to the sample size of failure/suspension times, EHM is extended into two forms: semi-parametric and non-parametric. The semi-parametric EHM assumes a specified lifetime distribution (i.e. Weibull distribution) in the form of the baseline hazard. However, for more industry applications, due to sparse failure event data of assets, the analysis of such data often involves complex distributional shapes about which little is known. Therefore, to avoid the restrictive assumption of the semi-parametric EHM about assuming a specified lifetime distribution for failure event histories, the non-parametric EHM, which is a distribution free model, has been developed. The development of EHM into two forms is another merit of the model. A case study was conducted using laboratory experiment data to validate the practicality of the both semi-parametric and non-parametric EHMs. The performance of the newly-developed models is appraised using the comparison amongst the estimated results of these models and the other existing covariate-based hazard models. The comparison results demonstrated that both the semi-parametric and non-parametric EHMs outperform the existing covariate-based hazard models. Future research directions regarding to the new parameter estimation method in the case of time-dependent effects of covariates and missing data, application of EHM in both repairable and non-repairable systems using field data, and a decision support model in which linked to the estimated reliability results, are also identified.

Bayesian survival analysis using gene expression

This thesis developed and applied Bayesian models for the analysis of survival data. The gene expression was considered as explanatory variables within the Bayesian survival model which can be considered the new contribution in the analysis of such data. The censoring factor that is inherent of survival data has also been addressed in terms of its impact on the fitting of a finite mixture of Weibull distribution with and without covariates. To investigate this, simulation study were carried out under several censoring percentages. Censoring percentage as high as 80% is acceptable here as the work involved high dimensional data. Lastly the Bayesian model averaging approach was developed to incorporate model uncertainty in the prediction of survival.

Ultimate flexural limit states analysis of prestressed concrete sleeper

Railway is one of the most important, reliable and widely used means of transportation, carrying freight, passengers, minerals, grains, etc. Thus, research on railway tracks is extremely important for the development of railway engineering and technologies. The safe operation of a railway track is based on the railway track structure that includes rails, fasteners, pads, sleepers, ballast, subballast and formation. Sleepers are very important components of the entire structure and may be made of timber, concrete, steel or synthetic materials. Concrete sleepers were first installed around the middle of last century and currently are installed in great numbers around the world. Consequently, the design of concrete sleepers has a direct impact on the safe operation of railways. The "permissible stress" method is currently most commonly used to design sleepers. However, the permissible stress principle does not consider the ultimate strength of materials, probabilities of actual loads, and the risks associated with failure, all of which could lead to the conclusion of cost-ineffectiveness and over design of current prestressed concrete sleepers. Recently the limit states design method, which appeared in the last century and has been already applied in the design of buildings, bridges, etc, is proposed as a better method for the design of prestressed concrete sleepers. The limit states design has significant advantages compared to the permissible stress design, such as the utilisation of the full strength of the member, and a rational analysis of the probabilities related to sleeper strength and applied loads. This research aims to apply the ultimate limit states design to the prestressed concrete sleeper, namely to obtain the load factors of both static and dynamic loads for the ultimate limit states design equations. However, the sleepers in rail tracks require different safety levels for different types of tracks, which mean the different types of tracks have different load factors of limit states design equations. Therefore, the core tasks of this research are to find the load factors of the static component and dynamic component of loads on track and the strength reduction factor of the sleeper bending strength for the ultimate limit states design equations for four main types of tracks, i.e., heavy haul, freight, medium speed passenger and high speed passenger tracks. To find those factors, the multiple samples of static loads, dynamic loads and their distributions are needed. In the four types of tracks, the heavy haul track has the measured data from Braeside Line (A heavy haul line in Central Queensland), and the distributions of both static and dynamic loads can be found from these data. The other three types of tracks have no measured data from sites and the experimental data are hardly available. In order to generate the data samples and obtain their distributions, the computer based simulations were employed and assumed the wheel-track impacts as induced by different sizes of wheel flats. A valid simulation package named DTrack was firstly employed to generate the dynamic loads for the freight and medium speed passenger tracks. However, DTrack is only valid for the tracks which carry low or medium speed vehicles. Therefore, a 3-D finite element (FE) model was then established for the wheel-track impact analysis of the high speed track. This FE model has been validated by comparing its simulation results with the DTrack simulation results, and with the results from traditional theoretical calculations based on the case of heavy haul track. Furthermore, the dynamic load data of the high speed track were obtained from the FE model and the distributions of both static and dynamic loads were extracted accordingly. All derived distributions of loads were fitted by appropriate functions. Through extrapolating those distributions, the important parameters of distributions for the static load induced sleeper bending moment and the extreme wheel-rail impact force induced sleeper dynamic bending moments and finally, the load factors, were obtained. Eventually, the load factors were obtained by the limit states design calibration based on reliability analyses with the derived distributions. After that, a sensitivity analysis was performed and the reliability of the achieved limit states design equations was confirmed. It has been found that the limit states design can be effectively applied to railway concrete sleepers. This research significantly contributes to railway engineering and the track safety area. It helps to decrease the failure and risks of track structure and accidents; better determines the load range for existing sleepers in track; better rates the strength of concrete sleepers to support bigger impact and loads on railway track; increases the reliability of the concrete sleepers and hugely saves investments on railway industries. Based on this research, many other bodies of research can be promoted in the future. Firstly, it has been found that the 3-D FE model is suitable for the study of track loadings and track structure vibrations. Secondly, the equations for serviceability and damageability limit states can be developed based on the concepts of limit states design equations of concrete sleepers obtained in this research, which are for the ultimate limit states.

A parametric duration model of the reaction times of drivers distracted by mobile phone conversations

The use of mobile phones while driving is more prevalent among young drivers—a less experienced cohort with elevated crash risk. The objective of this study was to examine and better understand the reaction times of young drivers to a traffic event originating in their peripheral vision whilst engaged in a mobile phone conversation. The CARRS-Q Advanced Driving Simulator was used to test a sample of young drivers on various simulated driving tasks, including an event that originated within the driver’s peripheral vision, whereby a pedestrian enters a zebra crossing from a sidewalk. Thirty-two licensed drivers drove the simulator in three phone conditions: baseline (no phone conversation), hands-free and handheld. In addition to driving the simulator each participant completed questionnaires related to driver demographics, driving history, usage of mobile phones while driving, and general mobile phone usage history. The participants were 21 to 26 years old and split evenly by gender. Drivers’ reaction times to a pedestrian in the zebra crossing were modelled using a parametric accelerated failure time (AFT) duration model with a Weibull distribution. Also tested where two different model specifications to account for the structured heterogeneity arising from the repeated measures experimental design. The Weibull AFT model with gamma heterogeneity was found to be the best fitting model and identified four significant variables influencing the reaction times, including phone condition, driver’s age, license type (Provisional license holder or not), and self-reported frequency of usage of handheld phones while driving. The reaction times of drivers were more than 40% longer in the distracted condition compared to baseline (not distracted). Moreover, the impairment of reaction times due to mobile phone conversations was almost double for provisional compared to open license holders. A reduction in the ability to detect traffic events in the periphery whilst distracted presents a significant and measurable safety concern that will undoubtedly persist unless mitigated.

Modelling survival data to account for model uncertainty: a single model or model averaging?

This study considered the problem of predicting survival, based on three alternative models: a single Weibull, a mixture of Weibulls and a cure model. Instead of the common procedure of choosing a single “best” model, where “best” is defined in terms of goodness of fit to the data, a Bayesian model averaging (BMA) approach was adopted to account for model uncertainty. This was illustrated using a case study in which the aim was the description of lymphoma cancer survival with covariates given by phenotypes and gene expression. The results of this study indicate that if the sample size is sufficiently large, one of the three models emerge as having highest probability given the data, as indicated by the goodness of fit measure; the Bayesian information criterion (BIC). However, when the sample size was reduced, no single model was revealed as “best”, suggesting that a BMA approach would be appropriate. Although a BMA approach can compromise on goodness of fit to the data (when compared to the true model), it can provide robust predictions and facilitate more detailed investigation of the relationships between gene expression and patient survival. Keywords: Bayesian modelling; Bayesian model averaging; Cure model; Markov Chain Monte Carlo; Mixture model; Survival analysis; Weibull distribution

Statistical analysis of reduction in tensile strength of cotton strips as a measure of soil microbial activity

The cotton strip assay (CSA) is an established technique for measuring soil microbial activity. The technique involves burying cotton strips and measuring their tensile strength after a certain time. This gives a measure of the rotting rate, R, of the cotton strips. R is then a measure of soil microbial activity. This paper examines properties of the technique and indicates how the assay can be optimised. Humidity conditioning of the cotton strips before measuring their tensile strength reduced the within and between day variance and enabled the distribution of the tensile strength measurements to approximate normality. The test data came from a three-way factorial experiment (two soils, two temperatures, three moisture levels). The cotton strips were buried in the soil for intervals of time ranging up to 6 weeks. This enabled the rate of loss of cotton tensile strength with time to be studied under a range of conditions. An inverse cubic model accounted for greater than 90% of the total variation within each treatment combination. This offers support for summarising the decomposition process by a single parameter R. The approximate variance of the decomposition rate was estimated from a function incorporating the variance of tensile strength and the differential of the function for the rate of decomposition, R, with respect to tensile strength. This variance function has a minimum when the measured strength is approximately 2/3 that of the original strength. The estimates of R are almost unbiased and relatively robust against the cotton strips being left in the soil for more or less than the optimal time. We conclude that the rotting rate X should be measured using the inverse cubic equation, and that the cotton strips should be left in the soil until their strength has been reduced to about 2/3.

The impact of mobile phone distraction on the braking behaviour of young drivers : a hazard-based duration model

Braking is a crucial driving task with a direct relationship with crash risk, as both excess and inadequate braking can lead to collisions. The objective of this study was to compare the braking profile of young drivers distracted by mobile phone conversations to non-distracted braking. In particular, the braking behaviour of drivers in response to a pedestrian entering a zebra crossing was examined using the CARRS-Q Advanced Driving Simulator. Thirty-two licensed drivers drove the simulator in three phone conditions: baseline (no phone conversation), hands-free, and handheld. In addition to driving the simulator, each participant completed questionnaires related to driver demographics, driving history, usage of mobile phones while driving, and general mobile phone usage history. The drivers were 18–26 years old and split evenly by gender. A linear mixed model analysis of braking profiles along the roadway before the pedestrian crossing revealed comparatively increased decelerations among distracted drivers, particularly during the initial 20 kph of deceleration. Drivers’ initial 20 kph deceleration time was modelled using a parametric accelerated failure time (AFT) hazard-based duration model with a Weibull distribution with clustered heterogeneity to account for the repeated measures experiment design. Factors found to significantly influence the braking task included vehicle dynamics variables like initial speed and maximum deceleration, phone condition, and driver-specific variables such as licence type, crash involvement history, and self-reported experience of using a mobile phone whilst driving. Distracted drivers on average appear to reduce the speed of their vehicle faster and more abruptly than non-distracted drivers, exhibiting excess braking comparatively and revealing perhaps risk compensation. The braking appears to be more aggressive for distracted drivers with provisional licenses compared to drivers with open licenses. Abrupt or excessive braking by distracted drivers might pose significant safety concerns to following vehicles in a traffic stream.

Data driven modeling for power transformer lifespan evaluation

Large sized power transformers are important parts of the power supply chain. These very critical networks of engineering assets are an essential base of a nation’s energy resource infrastructure. This research identifies the key factors influencing transformer normal operating conditions and predicts the asset management lifespan. Engineering asset research has developed few lifespan forecasting methods combining real-time monitoring solutions for transformer maintenance and replacement. Utilizing the rich data source from a remote terminal unit (RTU) system for sensor-data driven analysis, this research develops an innovative real-time lifespan forecasting approach applying logistic regression based on the Weibull distribution. The methodology and the implementation prototype are verified using a data series from 161 kV transformers to evaluate the efficiency and accuracy for energy sector applications. The asset stakeholders and suppliers significantly benefit from the real-time power transformer lifespan evaluation for maintenance and replacement decision support.