Prognostics Performance Research Articles

Online machine health prognostics based on modified duration-dependent hidden semi-Markov model and high-order particle filtering

Accurate online health prognostics is considered as a significant part of the condition-based maintenance (CBM), and it contributes to reduce downtime and achieve the most reliable running condition for machinery equipment. In this paper, an online machine health prognostics approach is proposed based on the modified duration-dependent hidden semi-Markov model (MDD-HSMM) and the high-order particle filter (HOPF) method. In the MDD-HSMM, the health state transition probabilities and the observation probabilities are both defined not only as state dependent like traditional HSMM does, but also as duration dependent, which is more realistic to describe the state space model to model the mechanical failure propagation process. And a new forward-backward algorithm is developed to facilitate the training process and to reduce computational complexity of the proposed MDD-HSMM. Then, the HOPF method with an online update scheme is applied to recognize the health states and predict the residual useful lifetime (RUL) value of machine in real time, which is based on the health state space model established by the MDD-HSMM and the online sensing monitoring data. And, a sliding window with variable length, which represents the relationship between current state and several previous states, is applied to adjust the order of HOPF. Finally, a real case study is used to illustrate the prognostics performance of the proposed approach and the experiment results indicate that the proposed approach has higher effectiveness than conventional HSMM methods.

Prognostics of Proton Exchange Membrane Fuel Cells Using A Model-based Method

Proton exchange membrane fuel cells (PEMFCs) are becoming one of the most popular future power technologies with the advantages including low operating temperature, high power conversion efficiency, and zero pollution. However, there are still many shortcomings restricting the large scale commercialization of PEMFCs, such as the short lifetime and high cost. Accurate prognostics of PEMFCs are very significant for its life and cost management. This paper presents an efficient semi-empirical model-based prognostics method estimating the health state and the remaining useful life of PEMFCs based on the adaptive unscented Kalman filter (AUKF) algorithm. AUKF algorithm can automatically adjust the system process covariance and observation covariance according to the estimation residuals, and it also solves the initial parameters setting problem of conventional UKF algorithm. Finally, simulation results demonstrate that the proposed model-based prognostics method using the AUKF algorithm has better prognostics performance than using the UKF algorithm.

A novel concept and assessment method for trustworthiness of prognostics

The existence of the unavoidable uncertainties has a great effect on the prognostics accuracy and performance. However, their influences are rarely taken into account in the current prognostics performance evaluation. Consequently, the validity and/or trustiness degree of prognostics and health management system has not been comprehensively evaluated or measured. Considering the definition of prognostics performance and the characteristics of uncertainties, four prognostics performance metrics and two kinds of quantitative evaluation methods are proposed. On the basis of prognostics performance definition, a novel trustworthiness of prognostics concept is presented. Then, three kinds of assessment methods for trustworthiness of prognostics are provided. Finally, a case study has been carried out on the aviation hydraulic pump for verification purposes, in which the oil-return flow is monitored for predicting the degradation of the pump. The systematic application of trustworthiness of prognostics assessment is implemented based on the back propagation neural network prognostics algorithm. The results show that the proposed trustworthiness of prognostics concept and assessment methods can contribute to the systematic and comprehensive understanding of the performance of prognostics and health management system.

Review and Analysis of Algorithmic Approaches Developed for Prognostics on CMAPSS Dataset


 
 
 Benchmarking of prognostic algorithms has been challenging due to limited availability of common datasets suitable for prognostics. In an attempt to alleviate this problem, several benchmarking datasets have been collected by NASA’s prognostic center of excellence and made available to the Prognostics and Health Management (PHM) community to allow evaluation and comparison of prognostics algorithms. Among those datasets are five C-MAPSS datasets that have been extremely popular due to their unique characteristics making them suitable for prognostics. The C-MAPSS datasets pose several challenges that have been tackled by different methods in the PHM literature. In particular, management of high variability due to sensor noise, effects of operating conditions, and presence of multiple simultaneous fault modes are some factors that have great impact on the generalization capabilities of prognostics algorithms. More than 70 publications have used the C-MAPSS datasets for developing data-driven prognostic algorithms. The C-MAPSS datasets are also shown to be well-suited for development of new machine learning and pattern recognition tools for several key preprocessing steps such as feature extraction and selection, failure mode assessment, operating conditions assessment, health status estimation, uncertainty management, and prognostics performance evaluation. This paper summarizes a comprehensive literature review of publications using C-MAPSS datasets and provides guidelines and references to further usage of these datasets in a manner that allows clear and consistent comparison between different approaches.
 
 

Options for Prognostics Methods: A review of data-driven and physics- based prognostics

Condition-based maintenance is a cost effective maintenance strategy, in which maintenance schedules are predicted based on the results provided from diagnostics and prognostics. Although there are several reviews on diagnostics methods and CBM, a relatively small number of reviews on prognostics are available. Moreover, most of them either provide a simple comparison of different prognostics methods or focus on algorithms rather than interpreting the algorithms in the context of prognostics. The goal of this paper is to provide a practical review of prognostics methods so that beginners in prognostics can select appropriate methods for their field of applications in terms of implementation and prognostics performance. To achieve this goal, this paper introduces not only various prognostics algorithms, but also their attributes and pros and cons using simple examples.

Sensor Selection with Grey Correlation Analysis for Remaining Useful Life Evaluation


 
 
 Sensor selection in data modeling is an important research topic for prognostics. The performance of prediction model may vary considerably under different variable subset. Hence it is of great important to devise a systematic sensor selection method that offers guidance on choosing the most representative sensors for prognostics. This paper proposes a sensor selection method based on the improved grey correlation analysis. From empirical observation, all the continuous-value sensors with a consistent monotonic trend are firstly selected for data fusion, and a linear regression model is used to convert the multi-dimensional sensor readings into one-dimensional health factor (HF). The correlation between HF and each of the selected sensors is evaluated by calculating the grey correlation degree defined on two time series. The optimal sensor subset with a relatively large correlation degree is selected to execute the final fusion. The effectiveness of the proposed method was verified experimentally on the turbofan engine simulation data supplied by NASA Ames, using instance-based learning methodology, and the experimental results showed that RUL prediction with fewer sensor inputs can obtain a more accurate prognostics performance than using all sensors initially considered relevant.
 
 

Diagnostics Driven PHM The Balanced Solution

Much effort has been made to develop technologies and define metrics for Prognostics Health Management (PHM). The problem is that most of this effort has focused on theoretical and high risk concepts of Prognostics performance while ignoring the real needs in “System Health Management”. In the wake of this technological attention, the importance of true Integrated Systems Health Management (ISHM) has been masked by the focus on single failure mode physics of failure solutions. The critical PHM metrics, derived from Integrated Systems Diagnostics Design (ISDD) have mostly been ignored. These critical metrics include Reliability, Safety, Testability, and System Maintainability & Sustainment, as well as the impact of prognostics performance on Systems Diagnostics. A key point to be made is that the ISDD process is much larger than just developing metrics. ISDD results in a well-designed system that meets true health management needs, as well as significantly lowering development costs, and the cost of ownership. Another point that needs to be made is that the core of ISDD is a proven and highly effective analysis solution in Model Based Diagnostics. This paper discusses the approach of using Model Based Diagnostics in the ISDD process to determine the best balance of the Health Management design. It will be shown how the impact and effectiveness of prognostics as integrated with the ISDD process provides true value to performance and cost avoidance.

Remaining Useful Life Estimation of Critical Components With Application to Bearings

Prognostics activity deals with the estimation of the Remaining Useful Life (RUL) of physical systems based on their current health state and their future operating conditions. RUL estimation can be done by using two main approaches, namely model-based and data-driven approaches. The first approach is based on the utilization of physics of failure models of the degradation, while the second approach is based on the transformation of the data provided by the sensors into models that represent the behavior of the degradation. This paper deals with a data-driven prognostics method, where the RUL of the physical system is assessed depending on its critical component. Once the critical component is identified, and the appropriate sensors installed, the data provided by these sensors are exploited to model the degradation's behavior. For this purpose, Mixture of Gaussians Hidden Markov Models (MoG-HMMs), represented by Dynamic Bayesian Networks (DBNs), are used as a modeling tool. MoG-HMMs allow us to represent the evolution of the component's health condition by hidden states by using temporal or frequency features extracted from the raw signals provided by the sensors. The prognostics process is then done in two phases: a learning phase to generate the behavior model, and an exploitation phase to estimate the current health state and calculate the RUL. Furthermore, the performance of the proposed method is verified by implementing prognostics performance metrics, such as accuracy, precision, and prediction horizon. Finally, the proposed method is applied to real data corresponding to the accelerated life of bearings, and experimental results are discussed.

Investigating the Effect of Damage Progression Model Choice on Prognostics Performance

The success of model-based approaches to systems health management depends largely on the quality of the underlying models. In model-based prognostics, it is especially the quality of the damage progression models, i.e., the models describing how damage evolves as the system operates, that determines the accuracy and precision of remaining useful life predictions. Several common forms of these models are generally assumed in the literature but are often not supported by physical evidence or physics-based analysis. In this paper, using a centrifugal pump as a case study, we develop different damage progression models. In simulation, we investigate how model changes influence prognostics performance. Results demonstrate that, in some cases, simple damage progression models are sufficient. But, in general, the results show a clear need for damage progression models that are accurate over long time horizons under varied loading conditions.