Early and robust remaining useful life prediction of supercapacitors using BOHB optimized Deep Belief Network

Haris, Muhammad; Hasan, Muhammad Noman; Qin, Shiyin

doi:10.1016/j.apenergy.2021.116541

Cited by 54 publications

(25 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Regarding the convolutional filter sizes, i.e., 𝑥 (single-head) or 𝑦 and 𝑦 (doublehead) variables, odd square dimensions ranging from 3 × 3 to 29 × 29 were evaluated by considering all possible combinations taking 𝐾 (𝑖 0,1,2) at a time. For example, in the case of 𝐾 4 , 𝑥 1 corresponded to 𝑆 , 𝑆 , 𝑆 , 𝑆 3,5,7,9 , 𝑥 2 to 𝑆 , 𝑆 , 𝑆 , 𝑆 3,5,7,11 , 𝑥 3 to 𝑆 , 𝑆 , 𝑆 , 𝑆 3,5,7,13 , etc., and 𝑥 to (23,25,27,29); after that, it continued in reverse order, i.e., 𝑥 corresponded to (29,27,25,23), 𝑥 corresponded to (29,27,25,21), and so on. Regarding the number of filters, i.e., 𝑥 (single-head) or 𝑦 and 𝑦 (double-head) variables, power of two between 8 and 256 were considered in incremental order.…”

Section: Genetic Optimizationmentioning

confidence: 99%

“…Haris et al [21] addressed the problem of finding optimal hyperparameters for a deep belief network (DBN), which is a generative model composed of multiple RMB layers, with the purpose of predicting the RUL of supercapacitors. To this end, they proposed a combination of Bayesian and HyperBand optimization and showed the universality of their model by training it on different degradation profiles with the same hyperparameters.…”

Section: Related Workmentioning

confidence: 99%

“…27,25,23,21,17), S2 = (29,25,11,5) N1 =(16,32,32,32,64), N2 = (32,32,32,32) P1 = (0,0,0,0,0), P2 = (0,0,0,0) d1 = 0.598, d1 = 09,11,13,25,27), S2 =(7,11,17,19,21,27) N1 =(16,16,32,64,64), N2 =(16,32,64,64,64,64) …”

mentioning

confidence: 99%

See 2 more Smart Citations

Remaining Useful Life Prediction from 3D Scan Data with Genetically Optimized Convolutional Neural Networks

Diraco

Siciliano

Leone

2021

Sensors

View full text Add to dashboard Cite

In the current industrial landscape, increasingly pervaded by technological innovations, the adoption of optimized strategies for asset management is becoming a critical key success factor. Among the various strategies available, the “Prognostics and Health Management” strategy is able to support maintenance management decisions more accurately, through continuous monitoring of equipment health and “Remaining Useful Life” forecasting. In the present study, convolutional neural network-based deep neural network techniques are investigated for the remaining useful life prediction of a punch tool, whose degradation is caused by working surface deformations during the machining process. Surface deformation is determined using a 3D scanning sensor capable of returning point clouds with micrometric accuracy during the operation of the punching machine, avoiding both downtime and human intervention. The 3D point clouds thus obtained are transformed into bidimensional image-type maps, i.e., maps of depths and normal vectors, to fully exploit the potential of convolutional neural networks for extracting features. Such maps are then processed by comparing 15 genetically optimized architectures with the transfer learning of 19 pretrained models, using a classic machine learning approach, i.e., support vector regression, as a benchmark. The achieved results clearly show that, in this specific case, optimized architectures provide performance far superior (MAPE = 0.058) to that of transfer learning, which, instead, remains at a lower or slightly higher level (MAPE = 0.416) than support vector regression (MAPE = 0.857).

show abstract

Section: Genetic Optimizationmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Remaining Useful Life Prediction from 3D Scan Data with Genetically Optimized Convolutional Neural Networks

Diraco

Siciliano

Leone

2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Haris et al [21] addressed the problem of find optimal hyperparameters for a Deep Belief Network (DBN), which is a generative model composed of multiple RMB layers, at the purpose to predict the RUL of supercapacitors. To this end, they proposed a combination of Bayesian and HyperBand optimization, and showed the universality of their model by training it on different degradation profiles with the same hyperparameters.…”

Section: Related Workmentioning

confidence: 99%

“…Regarding the convolutional filter sizes, i.e., (single-head) or and (doublehead) variables, odd square dimensions ranging from 3×3 to 29×29 were evaluated by considering all possible combinations taken ( = 0,1,2) at a time. For example, in the case of = 4, = 1 corresponded to ( , , , ) = (3,5,7,9), = 2 to ( , , , ) = (3,5,7,11), = 3 to ( , , , ) = (3,5,7,13), etc., to (23,25,27,29), after that it continued in reverse order, i.e., corresponded to (29,27,25,23), corresponded to (29,27,25,21), and so on. Regarding the number of filters, i.e., (single-head) or and (double-head) variables, power of two between 8 and 256 were considered in incremental order.…”

Section: Genetic Optimizationmentioning

confidence: 99%

Remaining Useful Life Prediction from 3D Scan Data with Genetically Optimized Convolutional Neural Networks

Diraco¹,

Siciliano²,

Leone³

2021

Preprint

View full text Add to dashboard Cite

In the current industrial landscape, increasingly pervaded by technological innovations, the adoption of optimized strategies for asset management is becoming a critical key success factor. Among the various strategies available, the “Prognostics and Health Management” strategy is able to support maintenance management decisions more accurately, through continuous monitoring of equipment health and “Remaining Useful Life” forecasting. In the present study, Convolutional Neural Network-based Deep Neural Network techniques are investigated for the Remaining Useful Life prediction of a punch tool, whose degradation is caused by working surface deformations during the machining process. Surface deformation is determined using a 3D scanning sensor capable of returning point clouds with micrometric accuracy during the operation of the punching machine, avoiding both downtime and human intervention. The 3D point clouds thus obtained are transformed into bidimensional image-type maps, i.e., maps of depths and normal vectors, to fully exploit the potential of convolutional neural networks for extracting features. Such maps are then processed by comparing 15 genetically optimized architectures with the transfer learning of 19 pre-trained models, using a classic machine learning approach, i.e., Support Vector Regression, as a benchmark. The achieved results clearly show that, in this specific case, optimized architectures provide performance far superior (MAPE=0.058) to that of transfer learning which, instead, remains at a lower or slightly higher level (MAPE=0.416) than Support Vector Regression (MAPE=0.857).

show abstract

State of charge estimation of supercapacitors based on multi‐innovation unscented Kalman filter under a wide temperature range

Zhang

Yang

et al. 2022

Intl J of Energy Research

View full text Add to dashboard Cite

Supercapacitors are characterized by a long service lifetime and high power density, which can meet the instantaneous high-power demand during the acceleration of electric vehicles. In this study, a fractional-order model is developed to simulate the polarization effect and charging/discharging characteristics of supercapacitors, considering the precision of the electrochemical model and the amount of calculation of the equivalent circuit model and using the adaptive genetic algorithm to identify the parameters. The accurate prediction of the state of charge (SOC) can improve efficiency, prolong the service lifetime, and ensure the safety of supercapacitors. This study proposes a multiinnovation unscented Kalman filter algorithm based on the fractional-order model to improve the SOC estimation accuracy. The proposed algorithm is compared with other algorithms and analyzed under different temperatures and operating conditions to verify the accuracy and effectiveness of the proposed algorithm in estimating the SOC and tracking the terminal voltage.Experimental results show that the root mean squared error and mean absolute error of the proposed algorithm are less than those of the other algorithms. The proposed algorithm accurately estimates the SOC and tracks the terminal voltage. The maximum root mean squared error and mean absolute error of SOC estimation error are 1.8% and 1.78%, respectively.

show abstract

Early and robust remaining useful life prediction of supercapacitors using BOHB optimized Deep Belief Network

Cited by 54 publications

References 18 publications

Remaining Useful Life Prediction from 3D Scan Data with Genetically Optimized Convolutional Neural Networks

Remaining Useful Life Prediction from 3D Scan Data with Genetically Optimized Convolutional Neural Networks

Remaining Useful Life Prediction from 3D Scan Data with Genetically Optimized Convolutional Neural Networks

State of charge estimation of supercapacitors based on multi‐innovation unscented Kalman filter under a wide temperature range

Contact Info

Product

Resources

About