Learning the health index of complex systems using dynamic conditional variational autoencoders

Wei, Yupeng; Wu, Dazhong; Terpenny, Janis P.

doi:10.1016/j.ress.2021.108004

Cited by 35 publications

(11 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since there are some sensors with irregular measurements that cannot offer useful deterioration information, we selected 14 sensor data to train the model based on signal correlation, monotonicity, and signal-to-noise ratio. These sensors are T24, T30, T50, P30, Nf, Nc, Ps30, phi, NRf, NRc, BPR, htBleed, W31 and W32, respectively [18,23].…”

Section: Experiments Studymentioning

confidence: 99%

“…In real cases, aircraft engines are usually operated under various operating conditions, such as different altitudes and throttle rotation angles, etc. These complex operating conditions often result in data distribution shifts, which significantly impact the predictive performance of LSTM [18]. Although previous studies have shown that integrating multiple sensor data can improve the RUL prediction performance of LSTM under complex operating conditions [10,11], it should be noted that these methods often overlook the distribution characteristics and spatial relationships of sensors under different operating conditions.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An interpretable RUL prediction method of aircraft engines under complex operating conditions using spatio-temporal features

Gao,

Wang,

Sun

2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

Long short-term memory (LSTM) based prediction methods have achieved remarkable achievements in remaining useful life (RUL) prediction for aircraft engines. However, their prediction performance and interpretability are unsatisfactory under complex operating conditions. For aircraft engines with high hazard levels, it is important to ensure the interpretability of the models while maintaining excellent prediction accuracy. To address these issues, an interpretable RUL prediction method of aircraft engines under complex operating conditions using spatio-temporal features (STFs), referred to as iSTLSTM, is proposed in this paper. First, we develop a feature extraction framework called Bi-ConvLSTM1D. This framework can effectively capture the spatial and temporal dependencies of sensor measurements, significantly enhancing the feature extraction capabilities of LSTM. Then, an interpretation module for STFs based on a hybrid attention mechanism is designed to quantitatively assess the contribution of STFs and output interpretable RUL predictions. The effectiveness of iSTLSTM is evidenced by extensive experiments on the C-MAPSS and N-CMAPSS datasets, confirming the superiority and reliability of our method for aircraft engine RUL prediction.

show abstract

Section: Experiments Studymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

An interpretable RUL prediction method of aircraft engines under complex operating conditions using spatio-temporal features

Gao,

Wang,

Sun

2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…In the decoder network, the context vector c i is calculated by using Equation (15). Then, the context vector and the output of the LSTM network are given to a fully connected layer to predict the SOH of the battery cell i in cycle t, that is, y i,t…”

Section: Attention-based Deep Learning Modelmentioning

confidence: 99%

“…To address this issue, an attention-based deep learning predictive algorithm is developed in this work, where an attention matrix referring to the significance level of a time series at distinctive time is generated so that the deep learning predictive model can utilize the most relevant portion of a time series for SOH predictions. Moreover, it has been demonstrated that constructing an effective health index can increase the performance of a predictive model [15,16]. However, to our best knowledge, very few studies have been conducted to define or construct a health index of a battery.…”

Section: Introductionmentioning

confidence: 99%

Prediction of State of Health of Lithium-Ion Battery Using Health Index Informed Attention Model

Wei

2023

Sensors

Self Cite

View full text Add to dashboard Cite

State-of-health (SOH) is a measure of a battery’s capacity in comparison to its rated capacity. Despite numerous data-driven algorithms being developed to estimate battery SOH, they are often ineffective in handling time series data, as they are unable to utilize the most significant portion of a time series while predicting SOH. Furthermore, current data-driven algorithms are often unable to learn a health index, which is a measurement of the battery’s health condition, to capture capacity degradation and regeneration. To address these issues, we first present an optimization model to obtain a health index of a battery, which accurately captures the battery’s degradation trajectory and improves SOH prediction accuracy. Additionally, we introduce an attention-based deep learning algorithm, where an attention matrix, referring to the significance level of a time series, is developed to enable the predictive model to use the most significant portion of a time series for SOH prediction. Our numerical results demonstrate that the presented algorithm provides an effective health index and can precisely predict the SOH of a battery.

show abstract

“…One of the most important areas of HI generation is empirically constructed models, both in the form of classical equations [16] and deep learning approaches [12,17]. There are several Ways of constructing an HI index by the example of statistical approaches, e.g., in [18] there are methods based on Dempster-Shafer evidence theory.…”

Section: Auxiliary Health Characteristicmentioning

confidence: 99%

Interaction models for remaining useful life estimation

Zhevnenko¹,

Kazantsev²,

Makarov³

2023

Preprint

View full text Add to dashboard Cite

The paper deals with the problem of controlling the state of industrial devices according to the readings of their sensors. The current methods rely on one approach to feature extraction in which the prediction occurs. We proposed a technique to build a scalable model that combines multiple different feature extractor blocks. A new model based on sequential sensor space analysis achieves state-of-the-art results on the C-MAPSS benchmark for equipment remaining useful life estimation. The resulting model performance was validated including the prediction changes with scaling.

show abstract

Learning the health index of complex systems using dynamic conditional variational autoencoders

Cited by 35 publications

References 18 publications

An interpretable RUL prediction method of aircraft engines under complex operating conditions using spatio-temporal features

An interpretable RUL prediction method of aircraft engines under complex operating conditions using spatio-temporal features

Prediction of State of Health of Lithium-Ion Battery Using Health Index Informed Attention Model

Interaction models for remaining useful life estimation

Contact Info

Product

Resources

About