An adaptive sensor selection framework for multisensor prognostics
TL;DR: A novel sensor selection framework is proposed to address the challenge of monitor the degradation of a system using multiple sensors simultaneously by adaptively deciding which sensors to use at the moment to enhance remaining useful life prediction.
Abstract: Recent advances in sensor technology have made it possible to monitor the degradation of a system using multiple sensors simultaneously. Accordingly, many neural network-based prognostic models hav...
Citations
More filters
Journal Article•
TL;DR: In this paper, the Adaptive Distributed Resource Allocation (ADRA) scheme is proposed, which specifies relatively simple local actions to be performed by individual sensor nodes in a sensor network for mode management.
Abstract: A major research challenge in the field of sensor networks is the distributed resource allocation problem, which concerns how the limited resources in a sensor network should be allocated or scheduled to minimize costs and maximize the network capability. In this paper, we propose the Adaptive Distributed Resource Allocation (ADRA) scheme, which specifies relatively simple local actions to be performed by individual sensor nodes in a sensor network for mode management. Each node adapts its operation over time in response to the status and feedback of its neighboring nodes. Desirable global behavior results from the local interactions between nodes. We study the effectiveness of the ADRA scheme for a realistic application scenario; namely, the sensor mode management for an acoustic wireless sensor network to track vehicle movement. We evaluated the scheme via simulations, and also prototyped the acoustic wireless sensor network scenario using the Crossbow MICA2 motes. Our simulation and hardware implementation results indicate that the ADRA scheme provides a good tradeoff between performance objectives such as coverage area, power consumption, and network lifetime.
17 citations
TL;DR: In many applied and industrial settings, the use of Artificial Intelligence (AI) for quality technology is gaining growing attention as mentioned in this paper, which refers to the broad set of techniques which replicate human skills.
Abstract: In many applied and industrial settings, the use of Artificial Intelligence (AI) for quality technology is gaining growing attention. AI refers to the broad set of techniques which replicate human ...
9 citations
TL;DR: In this article , a degradation index building method for multivariate sensory data with censoring is proposed based on an additive nonlinear model with variable selection, which can handle censored data, and can automatically select the informative sensor signals to be used in the degradation index.
3 citations
DOI•
13 Oct 2022
TL;DR: Wang et al. as mentioned in this paper proposed a deep learning-based probabilistic RUL prediction framework with multi self-attention mechanisms, which can adaptively extract useful information from both time dimension and feature dimension by weighting measurements from multiple in-suit sensors.
Abstract: Massive condition monitoring (CM) data from industrial systems has increased the usability of data-driven methods in prognostics. Remaining useful life (RUL) prediction plays a vital role in helping to improve system reliability and to reduce system risks. However, most of existing data-driven methods for RUL prediction only support point estimation and cannot adaptively extract information from different system features and time periods, but it is important to provide probabilistic RUL prediction results in practice. In this context, we propose a deep learning-based probabilistic RUL prediction framework with multi self-attention mechanisms. It is able to weight CM data in two dimensions and predict the probability density of the target RUL. Specifically, based on the multi self-attention mechanisms, the proposed framework can adaptively extract useful information from both time dimension and feature dimension by weighting measurements from multiple in-suit sensors. Then, a temporal convolution network with the shared weights is applied to feature extraction of the CM data. A non-parametric method is used to obtain a confidence interval (CI) of the target RUL with aleatoric uncertainty. The performance of the proposed framework is evaluated via a public turbofan CM dataset. The results show that the proposed framework can output high-accuracy CI for RUL prediction.
1 citations
13 Oct 2022
TL;DR: Wang et al. as mentioned in this paper proposed a deep learning-based probabilistic RUL prediction framework with multi self-attention mechanisms, which can adaptively extract useful information from both time dimension and feature dimension by weighting measurements from multiple in-suit sensors.
Abstract: Massive condition monitoring (CM) data from industrial systems has increased the usability of data-driven methods in prognostics. Remaining useful life (RUL) prediction plays a vital role in helping to improve system reliability and to reduce system risks. However, most of existing data-driven methods for RUL prediction only support point estimation and cannot adaptively extract information from different system features and time periods, but it is important to provide probabilistic RUL prediction results in practice. In this context, we propose a deep learning-based probabilistic RUL prediction framework with multi self-attention mechanisms. It is able to weight CM data in two dimensions and predict the probability density of the target RUL. Specifically, based on the multi self-attention mechanisms, the proposed framework can adaptively extract useful information from both time dimension and feature dimension by weighting measurements from multiple in-suit sensors. Then, a temporal convolution network with the shared weights is applied to feature extraction of the CM data. A non-parametric method is used to obtain a confidence interval (CI) of the target RUL with aleatoric uncertainty. The performance of the proposed framework is evaluated via a public turbofan CM dataset. The results show that the proposed framework can output high-accuracy CI for RUL prediction.
1 citations
References
More filters
TL;DR: A novel, efficient, gradient based method called long short-term memory (LSTM) is introduced, which can learn to bridge minimal time lags in excess of 1000 discrete-time steps by enforcing constant error flow through constant error carousels within special units.
Abstract: Learning to store information over extended time intervals by recurrent backpropagation takes a very long time, mostly because of insufficient, decaying error backflow. We briefly review Hochreiter's (1991) analysis of this problem, then address it by introducing a novel, efficient, gradient based method called long short-term memory (LSTM). Truncating the gradient where this does not do harm, LSTM can learn to bridge minimal time lags in excess of 1000 discrete-time steps by enforcing constant error flow through constant error carousels within special units. Multiplicative gate units learn to open and close access to the constant error flow. LSTM is local in space and time; its computational complexity per time step and weight is O. 1. Our experiments with artificial data involve local, distributed, real-valued, and noisy pattern representations. In comparisons with real-time recurrent learning, back propagation through time, recurrent cascade correlation, Elman nets, and neural sequence chunking, LSTM leads to many more successful runs, and learns much faster. LSTM also solves complex, artificial long-time-lag tasks that have never been solved by previous recurrent network algorithms.
72,897 citations
"An adaptive sensor selection framew..." refers background in this paper
...LSTM and GRU are designed to handle these long-term dependencies more efficiently (Cho et al. 2014; Hochreiter and Schmidhuber 1997)....
[...]
01 Jan 2014
TL;DR: In this paper, the encoder and decoder of the RNN Encoder-Decoder model are jointly trained to maximize the conditional probability of a target sequence given a source sequence.
Abstract: In this paper, we propose a novel neural network model called RNN Encoder‐ Decoder that consists of two recurrent neural networks (RNN). One RNN encodes a sequence of symbols into a fixedlength vector representation, and the other decodes the representation into another sequence of symbols. The encoder and decoder of the proposed model are jointly trained to maximize the conditional probability of a target sequence given a source sequence. The performance of a statistical machine translation system is empirically found to improve by using the conditional probabilities of phrase pairs computed by the RNN Encoder‐Decoder as an additional feature in the existing log-linear model. Qualitatively, we show that the proposed model learns a semantically and syntactically meaningful representation of linguistic phrases.
19,998 citations
TL;DR: This work shows why gradient based learning algorithms face an increasingly difficult problem as the duration of the dependencies to be captured increases, and exposes a trade-off between efficient learning by gradient descent and latching on information for long periods.
Abstract: Recurrent neural networks can be used to map input sequences to output sequences, such as for recognition, production or prediction problems. However, practical difficulties have been reported in training recurrent neural networks to perform tasks in which the temporal contingencies present in the input/output sequences span long intervals. We show why gradient based learning algorithms face an increasingly difficult problem as the duration of the dependencies to be captured increases. These results expose a trade-off between efficient learning by gradient descent and latching on information for long periods. Based on an understanding of this problem, alternatives to standard gradient descent are considered. >
7,309 citations
"An adaptive sensor selection framew..." refers methods in this paper
...…(LSTM) can be used as g: One drawback of a vanilla RNN is the vanishing gradient problem; the error gradients vanish (become exceedingly close to 0) as they backpropagate through multiple time steps, which makes learning long-term information nearly impossible (Bengio, Simard, and Frasconi 1994)....
[...]
Posted Content•
TL;DR: The authors randomly omits half of the feature detectors on each training case to prevent complex co-adaptations in which a feature detector is only helpful in the context of several other specific feature detectors.
Abstract: When a large feedforward neural network is trained on a small training set,
it typically performs poorly on held-out test data. This "overfitting" is
greatly reduced by randomly omitting half of the feature detectors on each
training case. This prevents complex co-adaptations in which a feature detector
is only helpful in the context of several other specific feature detectors.
Instead, each neuron learns to detect a feature that is generally helpful for
producing the correct answer given the combinatorially large variety of
internal contexts in which it must operate. Random "dropout" gives big
improvements on many benchmark tasks and sets new records for speech and object
recognition.
6,899 citations
"An adaptive sensor selection framew..." refers background in this paper
...Dropout is a standard approach to tackle this issue (Hinton et al. 2012)....
[...]
...Overlooking informative sensors can be viewed as similar to the co-adaptation concept in the NN literature, where different hidden neurons are highly dependent on others, and thus it deteriorates the generalization performance (Hinton et al. 2012)....
[...]
Proceedings Article•
14 Jun 2011
TL;DR: This paper shows that rectifying neurons are an even better model of biological neurons and yield equal or better performance than hyperbolic tangent networks in spite of the hard non-linearity and non-dierentiabil ity.
Abstract: While logistic sigmoid neurons are more biologically plausible than hyperbolic tangent neurons, the latter work better for training multi-layer neural networks. This paper shows that rectifying neurons are an even better model of biological neurons and yield equal or better performance than hyperbolic tangent networks in spite of the hard non-linearity and non-dierentiabil ity
6,790 citations