다중혈관 관상동맥 환자에서 y-문합을 이용하여 양쪽 내흉동맥만을 사용한 우회술의 조기 성적

Machinery health prognostics: A systematic review from data acquisition to RUL prediction

https://escholarship.org/content/qt5ns8r3fs/qt5ns8r3fs.pdf?t=pihyfp

Remaining useful life estimation in prognostics using deep convolution neural networks

An overview of time-based and condition-based maintenance in industrial application

Remaining useful life (RUL) prediction of rolling element bearings plays a pivotal role in reducing costly unplanned maintenance and increasing the reliability, availability, and safety of machines. This paper proposes a hybrid prognostics approach for RUL prediction of rolling element bearings. First, degradation data of bearings are sparsely represented using relevance vector machine regressions with different kernel parameters. Then, exponential degradation models coupled with the Frechet distance are employed to estimate the RUL adaptively. The proposed approach is evaluated using the vibration data from accelerated degradation tests of rolling element bearings and the public PRONOSTIA bearing datasets. Experimental results demonstrate the effectiveness of the proposed approach in improving the accuracy and convergence of RUL prediction of rolling element bearings.

A Hybrid Prognostics Approach for Estimating Remaining Useful Life of Rolling Element Bearings

This paper develops ordering heuristics to improve the efficiency of reliability evaluation for multistate two-terminal networks given all minimal path vectors ( $ d $ -MPs for short). In the existing methods, all $d$ -MPs are treated equally. However, we find that the importance of each $d$ -MP is different, and different orderings affect the efficiency of reliability evaluation. Based on the above observations, we introduce the length definitions for $d$ -MPs in a multistate two-terminal network, and develop four ordering heuristics, called O1, O2, O3, and O4, to improve the efficiency of the Recursive Sum of Disjoint Products (RSDP) method for evaluating network reliability. The results show that the proposed ordering heuristics can significantly improve the reliability evaluation efficiency, and O1 performs the best among the four methods. In addition, an ordering heuristic is developed for the reliability evaluation of multistate two-terminal networks given all minimal cut vectors ( $d$ -MCs).

Ordering Heuristics for Reliability Evaluation of Multistate Networks

In condition based maintenance (CBM) optimization, the main optimization objectives include maximizing reliability and minimizing maintenance costs, which are often times conflicting to each other. In this work, we develop a physical programming based approach to deal with the multi-objective condition based maintenance optimization problem. Physical programming presents two major advantages: (1) it is an efficient approach to capture the decision makers' preferences on the objectives by eliminating the iterative process of adjusting the weights of the objectives, and (2) it is easy to use in that decision makers just need to specify physically meaningful boundaries for the objectives. The maintenance cost and reliability objectives are calculated based on proportional hazards model and a control limit CBM replacement policy. With the proposed approach, the decision maker can systematically and efficiently make good tradeoff between the cost objective and reliability objective. An example is used to illustrate the proposed approach.

Condition based maintenance optimization considering multiple objectives

Redundancy allocation for multi-state systems using physical programming and genetic algorithms

The development of accurate health condition prediction approaches has been a key research topic in condition based maintenance (CBM) in recent years. However, current health condition prediction approaches are not accurate enough, which has become the bottleneck for achieving the full power of CBM. Neural network based methods have been considered to be a very promising category of methods for equipment health condition prediction. In this paper, we propose a neural network prediction model called extended recurrent neural network (ERNN). An ERNN based approach is developed for health condition prediction of gearboxes based on the vibration data collected from a gearbox experimental system. The results demonstrate the capability of the ERNN based approach for producing satisfactory health condition prediction results. A comparative study based on the gearbox experiment data further establishes ERNN as an effective recurrent neural network model for equipment health condition prediction.

Health Condition Prediction of Gears Using a Recurrent Neural Network Approach

Model-based gear dynamic analysis and simulation has been a promising way for developing effective gearbox vibration monitoring approaches. In this paper, based on the dynamic model of a one-stage gearbox with spur gears and one tooth crack, statistical indicators and the discrete wavelet transform (DWT) technique are investigated to identify effective and sensitive health indicators for reflecting the crack propagation level. The results suggest that the root mean square (RMS) indicator is a better statistical indicator than the Kurtosis indicator to reflect the crack propagation in the early stage; the RMS indicator based on the residual signal segments that are strongly affected by the crack is more sensitive; the proposed DWT approach can improve the sensitivity of the RMS indicator, and the RMS indicator becomes more sensitive with the increase of the DWT level up to a best DWT level, beyond which either the monotonicity is lost or the sensitivity decreases; the proposed approach is effective with the presence of noise; with the increase of the noise level, the DWT level at which the best performance is achieved, and thus the sensitivity, decreases. Gearbox systems with different sizes and different input shaft frequencies are also investigated, and it is found that the observations presented above hold for different gearbox system settings.

Zhigang Tian

Papers

Ordering Heuristics for Reliability Evaluation of Multistate Networks

Condition based maintenance optimization considering multiple objectives

Redundancy allocation for multi-state systems using physical programming and genetic algorithms

Health Condition Prediction of Gears Using a Recurrent Neural Network Approach

Crack propagation assessment for spur gears using model-based analysis and simulation