Predict Performance Degradation Research Articles

Modeling of Performance Decay Prediction Based on LSTM-SVM Flexible Circuit Board Processing Roll

The roller is the core component of the flexible material roll-to-roll equipment, and its performance will affect the processing quality, so it is necessary to predict the health of the roller. In order to improve the prediction accuracy and effectively extract the time sequence information hidden in the signal, a LSTM-SVM-based processing roll performance degradation prediction model is proposed. By collecting the bearing vibration sensor data, extracting the characteristics of the vibration signal after normalization, using the extracted features as the input of the prediction model, inputting a part of the samples as the training set into the LSTM-SVM prediction model, and inputting the model in batches for network training, and Adjustment parameters. After the model is trained, use the test set for testing. Compared with support vector machines, the SLTM-SVM model is more effective in predicting the performance degradation of roll-to-roll equipment.

Performance prediction of fuel cells using long short‐term memory recurrent neural network

Performance prediction of proton-exchange membrane fuel cell (PEMFC) under dynamic conditions, especially for vehicle applications, is increasingly become the focus of attention. This article proposes a performance prediction method of PEMFC using long short-term memory (LSTM) recurrent neural network (RNN). In this article, polarization curve (current-voltage curve) and voltage degradation curve (current-time curve) are adopted as the main performance indexes of PEMFC. Both polarization curve prediction and performance degradation prediction of PEMFC can be effectively implemented based on the LSTM method. To investigate the voltage losses law of experimental and predicted results, the paper introduces an empirical equation of polarization curve. The perfect match between the experimental and predicted polarization losses of PEMFC can further validate the prediction performance of LSTM method. The proposed prediction method is validated by the PEMFC polarization curve data obtained from the designed aging experiment of a 4 kW stack operated under dynamic loading cycling situation during 600 hours. Then, LSTM network is compared with traditional RNN and back-propagation neural network (BPNN) to prove its superiority. The minimum values of root-mean-square error (RMSE) and the mean absolute percentage error (MAPE) of LSTM network with different training data are 0.0088 and 0.0101, respectively. All the coefficient of determination (R2) of LSTM model with different training data is over 0.95, which is close to 1.0. The prediction accuracy of LSTM network is higher than that of two other networks. The result indicates that LSTM network outperforms two other networks in PEMFC performance prediction. Hence, the prediction method based on LSTM network is very suitable for PEMFC performance prediction.

Lévy Process-Based Stochastic Modeling for Machine Performance Degradation Prognosis

Accurate and reliable machine performance degradation tracking and remaining useful life (RUL) prognosis establish the foundation for predictive maintenance scheduling toward improved safety and productivity of machine operations. In general, machine performance degradation exhibits a nonlinear and nonhomogeneous pattern that arises from time-varying degradation rate and abrupt performance changes. To address this challenge and improve the generalizability of degradation modeling, in this article, we present a stochastic modeling technique based on the Lévy process, which generalizes system variations as the accumulations of successive and jump increments. The developed Lévy process model consists of two terms: a linear Brownian motion term for capturing the gradual degradation with time-varying rates and a nonhomogenous compound Poisson process term for capturing transient performance changes. By calculating the moments of the characteristic function of the Lévy model, explicit expressions for the probability distributions of predicted performance degradation and RUL are derived. To obtain the time-varying parameters in the Lévy model, Markov chain Monte Carlo is investigated. The developed technique is evaluated through simulation and run-to-failure tests of roller and ball bearings, and the good performance of the developed Lévy model is confirmed.

Three-dimensional modeling of performance degradation of planar SOFC with phosphine exposure

A three-dimensional computational model is developed to predict the performance degradation of planar SOFC anodes exposed to fuel contaminants commonly present in coal syngas. The model parameters are calibrated using data from button cell experiments. The calibrated model is then used to perform simulations to predict performance degradation of planar cells due to phosphine contamination. The results from degradation simulations show that the contaminant coverage alters the current and temperature distributions significantly. The electrochemical characteristics of the degraded cell are analyzed by performing impedance and polarization simulations. The impedance analysis is applied to the cell in three regions along the fuel flow direction to elucidate location specific degradation phenomena. The results show that the degradation rates and the impedance behavior of planar cells are very different and more complicated than those observed in button cells. In such cases, localized impedance analysis may be necessary to elucidate various degradation mechanisms.

A hybrid deep transfer learning strategy for short term cross-building energy prediction

To overcome the data shortage problem of model training in building energy prediction, this study proposes a novel hybrid deep transfer learning strategy for short term cross-building energy prediction using long short term memory (LSTM) and domain adversarial neural network (DANN). The proposed strategy can utilize knowledge learned from relevant building data to assist the energy prediction for target buildings with limited historical measurements. LSTM based feature extractor is used to extract temporal features across source and target buildings. DANN attempts to find domain invariant features between the source and target buildings via domain adaptation. Then, the domain adaptation based transfer learning model (i.e. LSTM-DANN) trained with data from source buildings can be applied to assist in predicting the target building energy without prediction performance degradation caused by domain shift. Experiments are conducted to evaluate the performance of the proposed transfer learning strategy in different models. Results demonstrate that the proposed strategy can significantly enhance the building energy prediction performance compared to models trained on the target only data, the source only data, both the target and source data, but without transfer learning. This work can provide guidance for the effective use of existing building data resources.

Atomistic investigation of GFRP composites under chloride environment

Fiber reinforced polymer composites have paved the way for the evolution of the engineering materials. Glass fiber reinforced polymer (GFRP) has become one of the most promising materials among the novel composites due to its low price and high performance. Here, the GFRP composite in investigated by fill atomistic molecular dynamics, which contains the epoxy matrix and amorphous silica substrate. The degradation in chloride environment is revealed through the interfacial interactions, the structural changes of the epoxy matrix and hydrogen bonding in the composites. Compared to dry and aqueous environment, the simulation results show that chloride environment leads to the worst deterioration of interfacial adhesion, which correlates with structural and mechanical degradation of bonded interface, as indicated by the decreased epoxy density close to interface. It is indicated by examining the nano-structures of the interface during the pulling process that the chloride environment hinders the formation of intramolecular and intermolecular hydrogen bonds in the GFRP composites. These findings indicate the deterioration towards bond and matrix is critical in the presence of the chloride environment, which provides the fundamental insight for designing and predicting performance degradation of macroscopic GFRP composites in the marine application.

Model identification of durability degradation process of concrete material and structure based on Wiener process

The degradation and damage of concrete material and structure during their service life are significantly characterized by randomness, so it is of great importance to establish a stochastic process model for the structural durability degradation. This study was based on Wiener process to identify the durability degradation of concrete structures. Firstly, two identification methods for Wiener process were introduced, including autocorrelation function method and probability statistics verification method. The former is a rough method for preliminary identification, and the latter is a strict method for accurate identification. According to the degradation processes and damage mechanisms of dynamic elastic modulus of concrete and section stiffness of reinforced concrete girder, and based on relevant data, autocorrelation function method and probability statistics verification method were used to identify whether the degradation processes of concrete dynamic elastic modulus and girder section stiffness conform to Wiener process. The results showed that the degradation process of concrete dynamic elastic modulus and the degradation process of section stiffness of reinforced concrete girder can be modeled with Wiener process. The conclusions contribute to the prediction of performance degradation and the maintenance for concrete structures in service.

Performance degradation prediction of proton exchange membrane fuel cell using a hybrid prognostic approach

The proton exchange membrane fuel cell has been widely used for industrial systems; however, its performance gradually degrades during use. Therefore, the study on the performance degradation prediction of fuel cells is helpful to extend its lifespan. In this paper, a novel hybrid approach using a combination of model-based adaptive Kalman filter and data-driven NARX neural network is proposed to predict the degradation of fuel cells. The overall degradation trend (i.e., irreversible degradation process) is captured by an empirical aging model and adaptive Kalman filter. Meanwhile, the detail degradation information (i.e., reversible degradation process) is depicted by the NARX neural network. Moreover, the correlation analysis of the reversible voltage time series is carried out to obtain the number of delays of the NARX neural network based on the autocorrelation function and the partial autocorrelation function. Then, the total degradation prediction is the sum of the overall degradation prediction and the detail degradation prediction. Finally, the prognostic capability of the proposed method is verified by two aging datasets, and the results show the effectiveness and superiority of the proposed method which can provide accurate degradation forecasting and remaining useful life.

Improving robot dual-system motor learning with intrinsically motivated meta-control and latent-space experience imagination

Combining model-based and model-free learning systems has been shown to improve the sample efficiency of learning to perform complex robotic tasks. However, dual-system approaches fail to consider the reliability of the learned model when it is applied to make multiple-step predictions, resulting in a compounding of prediction errors and performance degradation. In this paper, we present a novel dual-system motor learning approach where a meta-controller arbitrates online between model-based and model-free decisions based on an estimate of the local reliability of the learned model. The reliability estimate is used in computing an intrinsic feedback signal, encouraging actions that lead to data that improves the model. Our approach also integrates arbitration with imagination where a learned latent-space model generates imagined experiences, based on its local reliability, to be used as additional training data. We evaluate our approach against baseline and state-of-the-art methods on learning vision-based robotic grasping in simulation and real world. The results show that our approach outperforms the compared methods and learns near-optimal grasping policies in dense- and sparse-reward environments.

Performance Degradation Prediction Based on a Gaussian Mixture Model and Optimized Support Vector Regression for an Aviation Piston Pump.

Performance degradation prediction plays a key role in realizing aviation pump health management and condition-based maintenance. Thus, this paper proposes a new approach that combines a Gaussian mixture model (GMM) and optimized support vector regression (SVR) to predict aviation pumps’ degradation processes based on the pump outlet pressure signals. Different from other feature extraction methods in which the information of intrinsic mode functions (IMFs) is not fully utilized, some useful IMF components are firstly chosen, and the corresponding multi-domain features are extracted from each selected component. Considering that it is not the case that all features are equally sensitive to degradation assessment, PCA is used to select more sensitive degradation features. Since the distribution of these extracted features is a stochastic process in feature space, meanwhile, self-information quantity can describe the uncertainty of system by measuring the average information quantity contained in the probability distribution, self-information quantity based on GMM is defined as degradation index (DI) to describe the degradation degree of the pump quantitatively. Finally, an SVR model is constructed to predict the degradation status of the pump. To achieve higher prediction accuracy, phase space reconstruction theory is first employed to determine the number of the inputs of the SVR model, then a new method combining particle swarm optimization (PSO) with grid search (GS) is developed to optimize the parameters of the SVR model. Finally, both the online data and historical data are utilized for the construction of the SVR model, respectively. The effectiveness of the proposed approach is validated by full life cycle data collected from an aviation pump test rig. The results demonstrate that the DI extracted from pump outlet pressure signals can effectively identify and track the current deterioration stage, and the established SVR model has better prediction ability when compared with previously published methods.

Degradation prediction using displacement damage dose method for AlInGaP solar cells by changing displacement threshold energy under irradiation with low-energy electrons

Performance degradation prediction for space solar cells under irradiation with low-energy electrons is greatly affected by displacement threshold energy (Ed) when a displacement damage dose (DDD) model is used. According to recent studies, the Ed of P atoms is much lower than the conventional Ed value in InP-type solar cells irradiated with low-energy electrons. This indicates that the value of Ed typically used in DDD model leads to significant error in performance degradation prediction. In this study, degradation of AlInGaP solar cells is observed after irradiation with 60–500 keV electrons. The results suggest that the Ed of P atoms in AlInGaP solar cells is much smaller than the conventionally used Ed value. By using the DDD model with the Ed value obtained in this study (Ed = 3.5 eV for P), we demonstrated that the performance degradation predicted by the DDD model agrees well with the experimental results.

Adaptive detection and prediction of performance degradation in off-shore turbomachinery

Performance-based maintenance of machinery relies on detection and prediction of performance degradation. Degradation indicators calculated from process measurements need to be approximated with degradation models that smooth the variations in the measurements and give predictions of future values of the indicator. Existing models for performance degradation assume that the performance monotonically decreases with time. In consequence, the models yield suboptimal performance in performance-based maintenance as they do not take into account that performance degradation can reverse itself. For instance, deposits on the blades of a turbomachine can be self-cleaning in some conditions. In this study, a data-driven algorithm is proposed that detects if the performance degradation indicator is increasing or decreasing and adapts the model accordingly. A moving window approach is combined with adaptive regression analysis of operating data to predict the expected value of the performance degradation indicator and to quantify the uncertainty of predictions. The algorithm is tested on industrial performance degradation data from two independent offshore applications, and compared with four other approaches. The parameters of the algorithm are discussed and recommendations on the optimal choices are made. The algorithm proved to be portable and the results are promising for improving performance-based maintenance.

PTS-FNN-Based Health Prediction Method for Flexible Photoelectric Film Material Processing Equipment

Roll-to-Roll (R2R) processing is a common processing method for flexible photoelectric film materials. Due to the physical properties of the materials, the change in the performance of the R2R processing equipment can easily cause deformation of the flexible film material, it is particularly important to predict the performance degradation of the processing equipment. Based on the accuracy and real-time requirements of performance degradation prediction, a PTS-FNN model for performance degradation prediction was proposed in this paper, which combines the Possibilistic C-Means (PCM) fuzzy clustering and Takagi–Sugeno Fuzzy Neural Network (TS-FNN). We also studied the PCM classification algorithm of input data of PTS-FNN model, the predecessor network of TS-FNN prediction model and the construction method of post-component network. Finally, the implementation process of PCM classification algorithm and TS-FNN prediction model were given. The R2R processing equipment health prediction experiment system was built and the PTS-FNN model experiment was carried out. The experimental results showed that the training time of PTS-FNN model was 50.37% less than the standard TS-FNN prediction model. The prediction accuracy increased by 5.48%, and the PTS-FNN had no error in the judgment of state 1 and state 4.

Predicting and reining in application-level slowdown on spatial multitasking GPUs

Predicting performance degradation of a GPU application at co-location on a spatial multitasking GPU without prior application knowledge is essential in public Clouds. Prior work mainly targets CPU co-location, and is inaccurate and/or inefficient for predicting performance of applications at co-location on spatial multitasking GPUs. Our investigation shows that hardware event statistics caused by co-located applications strongly correlate with their slowdowns. Based on this observation, we present Themis with a kernel slowdown model (Themis-KSM), which performs precise and efficient online application slowdown prediction without prior application knowledge. The kernel slowdown model is trained offline. When new applications co-run, Themis-KSM collects event statistics and predicts their slowdowns simultaneously. In addition, we also propose a two-stage slowdown prediction mechanism (Themis-TSP) for real-system GPUs without any hardware modification. Our evaluation shows that Themis has negligible runtime overhead, and both Themis-KSM and Themis-TSP can precisely predict application-level slowdown with prediction error smaller than 9.5% and 12.8%, respectively. Based on Themis, we also implement an SM allocation engine to rein in application slowdown at co-location. Case studies show that the engine successfully enforces fair sharing and QoS.

HKF-SVR Optimized by Krill Herd Algorithm for Coaxial Bearings Performance Degradation Prediction.

In real industrial applications, bearings in pairs or even more are often mounted on the same shaft. So the collected vibration signal is actually a mixed signal from multiple bearings. In this study, a method based on Hybrid Kernel Function-Support Vector Regression (HKF–SVR) whose parameters are optimized by Krill Herd (KH) algorithm was introduced for bearing performance degradation prediction in this situation. First, multi-domain statistical features are extracted from the bearing vibration signals and then fused into sensitive features using Kernel Joint Approximate Diagonalization of Eigen-matrices (KJADE) algorithm which is developed recently by our group. Due to the nonlinear mapping capability of the kernel method and the blind source separation ability of the JADE algorithm, the KJADE could extract latent source features that accurately reflecting the performance degradation from the mixed vibration signal. Then, the between-class and within-class scatters (SS) of the health-stage data sample and the current monitored data sample is calculated as the performance degradation index. Second, the parameters of the HKF–SVR are optimized by the KH (Krill Herd) algorithm to obtain the optimal performance degradation prediction model. Finally, the performance degradation trend of the bearing is predicted using the optimized HKF–SVR. Compared with the traditional methods of Back Propagation Neural Network (BPNN), Extreme Learning Machine (ELM) and traditional SVR, the results show that the proposed method has a better performance. The proposed method has a good application prospect in life prediction of coaxial bearings.

A Novel Deep Belief Network and Extreme Learning Machine Based Performance Degradation Prediction Method for Proton Exchange Membrane Fuel Cell

Lifetime and reliability seriously affect the applications of proton exchange membrane fuel cell (PEMFC). Performance degradation prediction of PEMFC is the basis for improving the lifetime and reliability of PEMFC. To overcome the lower prediction accuracy caused by uncertainty and nonlinearity characteristics of degradation voltage data, this article proposes a novel deep belief network (DBN) and extreme learning machine (ELM) based performance degradation prediction method for PEMFC. A DBN based fuel cell degradation features extraction model is designed to extract high-quality degradation features in the original degradation data by layer-wise learning. To tackle the issues of overfitting and instability in fuel cell performance degradation prediction, an ELM with good generalization performance is introduced as a nonlinear prediction model, which can get some enhancement of prediction precision and reliability. Based on the designed DBN-ELM model, the particle swarm optimization (PSO) algorithm is used in the model training process to optimize the basic network structure of DBN-ELM further to improve the prediction accuracy of the hybrid neural network. Finally, the proposed prediction method is experimentally validated by using actual data collected from the 5-cells PEMFC stack. The results demonstrate that the proposed approach always has better prediction performance compared with the existing conventional methods, whether in the cases of various training phase or the cases of multi-step-ahead prediction.

Performance Prediction Using High-Order Differential Mathematical Morphology Gradient Spectrum Entropy and Extreme Learning Machine

Performance prediction is significant to monitor the health status of rolling bearings, which can greatly reduce the loss caused by potential faults in the whole life cycle of rolling bearings. It is a very important part of Prognostic and Health Management (PHM). In this article, a new performance degradation prediction (HMEPEM) method based on high-order differential mathematical morphology gradient spectrum entropy (HOMMSE), phase space reconstruction, and extreme learning machine (ELM) is proposed to predict the performance degradation trend of rolling bearings. In the proposed HMEPEM method, the HOMMSE method is used to extract the initial features of performance degradation from the raw bearing vibration signals and divide working stages. Then the phase space reconstruction is used to further extract more useful features from the initial features of performance degradation in order to construct a feature matrix, which is input into the ELM in order to build the performance degradation prediction model for predicting the performance degradation trend of rolling bearings. The proposed HMEPEM method is validated on the performance degradation data of rolling bearings provided by the PRONOSTIA platform. The results show that the proposed HMEPEM method can efficiently track the evolution of degradation and predict the performance degradation trend of rolling bearings.

Performance degradation prediction of mechanical equipment based on optimized multi-kernel relevant vector machine and fuzzy information granulation

An operation condition evaluation and prediction method based on wavelet packet information entropy and multi-kernel relevance vector machine is proposed for mechanical equipment in this paper. Firstly, the vibration signals of mechanical equipment are decomposed into high-frequency signals and low-frequency signals with different scales by using the theory of wavelet packet decomposition. The complexity and irregularity of the frequency-varying signals are represented by information entropy. Then, these constructed characteristic indicators are input to the relevance vector machine for performance degradation prediction. Finally, two kinds of roller bearings used in the experiment are taken as an example to verify the effectiveness of the method. The results show that the proposed method for evaluating the operational state of mechanical equipment can fully extract the characteristic information representing the operational state of roller bearings. It provides a new theoretical and practical support for mechanical equipment operational state evaluation and prediction.

기계학습을 이용한 필터의 성능 저하 예측

기계학습을 이용한 필터의 성능 저하 예측

Performance Supervised Fault Detection Schemes for Industrial Feedback Control Systems and their Data-Driven Implementation

This article addresses performance supervised fault detection (PSFD) issues for industrial feedback control systems based on performance degradation prediction. To be specific, three performance indicators are first introduced based on Bellman equation to predict system performance degradations for industrial processes with the aid of machine learning techniques. Based on them, three PSFD schemes are proposed by embedding the performance indicators as supervising information. In this context, the data-driven implementation of PSFD schemes are investigated for linear systems with unmeasurable state variables. A case study on rolling mill process, a typical benchmark in the steel manufacturing processes, is given at the end of this article to illustrate the applications of the proposed fault detection schemes.