Long-term forecasting of multivariate time series in industrial furnaces with dynamic Gaussian Bayesian networks

Quesada, David; Valverde, Gabriel; Larrañaga, Pedro; Bielza, Concha

doi:10.1016/j.engappai.2021.104301

Cited by 21 publications

(10 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As time passes, these impurities can generate an insulating layer that will force us to progressively increase the furnace temperature to keep the fluid temperature inside the pipes constant. This process is called fouling , and it is the cause of severe costs in terms of efficiency loss 22 . Due to the inherently nonlinear nature of the physical relationships that define this phenomenon, it can be used as an example to test the effectiveness of our hybrid model.…”

Section: Resultsmentioning

confidence: 99%

“…This process is called fouling, and it is the cause of severe costs in terms of efficiency loss. 22 Due to the inherently nonlinear nature of the physical relationships that define this phenomenon, it can be used as an example to test the effectiveness of our hybrid model.…”

Section: Simulated Nonlinear Problem: Fouling Phenomenonmentioning

confidence: 99%

“…This is a process called fouling. 22 Our aim is to define a simplified version of this phenomenon, one that reflects the nonlinear relationships of the variables and respects the underlying physical process. Our first simplification is that the simulation does not have a spatial component, so heat will transfer from the heat source to the tube walls by radiation from a single point to another and from the tube walls to the fluid by convection.…”

Section: Conflict Of Interestmentioning

confidence: 99%

See 2 more Smart Citations

Piecewise forecasting of nonlinear time series with model tree dynamic Bayesian networks

Quesada

Bielza

Fontán

et al. 2022

Int J of Intelligent Sys

Self Cite

View full text Add to dashboard Cite

When modelling multivariate continuous time series, a common issue is to find that the original processes that generated the data are nonlinear or that they drift away from the original distribution as the system evolves over time. In these scenarios, using a linear model such as a Gaussian dynamic Bayesian network (DBN) can result in severe forecasting inaccuracies due to the structure and parameters of the model remaining constant without considering either the elapsed time or the state of the system. To approach this problem, we propose a hybrid model that combines a model tree with DBNs. The model first divides the original data set into different scenarios based on the splits made by the tree over the system variables and then performs a piecewise regression in each branch of the tree to obtain nonlinear forecasts. The experimental results on three different datasets show that our model outperforms standard DBN models when dealing with nonlinear processes and is competitive with state-ofthe-art time series forecasting methods.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Simulated Nonlinear Problem: Fouling Phenomenonmentioning

confidence: 99%

Section: Conflict Of Interestmentioning

confidence: 99%

See 1 more Smart Citation

Piecewise forecasting of nonlinear time series with model tree dynamic Bayesian networks

Quesada

Bielza

Fontán

et al. 2022

Int J of Intelligent Sys

Self Cite

View full text Add to dashboard Cite

show abstract

“…On the other hand, the Markovian order defines the number of time slices required to assume that the present is independent of the past. Increasing the Markovian order implies more arcs appearing from earlier lags to the present, and thus, a greater complexity when learning the network structure [11].…”

Section: Bayesian Network To Detect Apnea Episodesmentioning

confidence: 99%

Detecting Obstructive Apnea Episodes using Dynamic Bayesian Networks and ECG-based Time-Series

Romero¹,

Jané²

2022

2022 44th Annual International Conference of the IEEE Engineering in Medicine &Amp; Biology Society (EMBC)

View full text Add to dashboard Cite

In this study, we proposed an automatic detector for obstructive apnea episodes using only ECG-based timeseries from a single-ECG channel. Several obstructive apnea episodes were provoked for different separated sequences of 15 minutes in anesthetized Sprague-Dawley rats. In this recurrent obstructive sleep apnea (OSA) model, each episode lasted 15 s, while the number of total events per sequence was randomly selected. The beat-to-beat interval (RR) and the R-wave amplitude (Ra) time-series were extracted and processed for each sequence, and used to train Dynamic Bayesian Networks with different lags. An optimal trade-off between the lag (L) and RM SE values was considered to select the best model to be used when detecting apnea episodes. The selected models were then used to estimate the occurrence probability of apnea episodes, p(At), by using a filtering approach. Finally, the timeseries of the estimated probabilities were post-processed using non-overlapped 15-s epochs, to determine whether they are classified as apneic or non-apneic segments. Results showed that those lagged models with orders greater than 5, presented suitable RMSE values and become more sensitive as the order increased. A detection threshold of 0.2 seems to provide the best apnea detection performance overall, with Acc=0.81, Se=0.83 and Sp=0.79, using two ECG parameters and L = 10.Clinical relevance-Dynamic Bayesian Networks represent a powerful tool to develop personalized models for apnea detection and diagnosis in OSA patients.

show abstract

“…Many statistical methods for forecasting included ARIMA (ArunKumar et al, 2021;Fan et al, 2021;M.-D. Liu et al, 2021;Selvaraj et al, 2020;Toğa et al, 2021;Yang et al, 2021), ARIMAX (Hossain et al, 2021;Li et al, 2020), multivariate time series (Koutlis et al, 2020;X. Liu & Lin, 2021;Quesada et al, 2021;Vanhoenshoven et al, 2020;R. Zhang & Jia, 2021), GARCH (Aras, 2021;Hung et al, 2020;Z.…”

Section: Introductionmentioning

confidence: 99%

GSTARIMA Model with Missing Value for Forecasting Gold Price

Mubarak

Aslanargun

Sıklar³

2022

IJSA

View full text Add to dashboard Cite

Gold is one of the investments that be a great demand. Selecting and applying the best GSTARIMA model for gold price forecasting was the aim of this study. However, the gold price data that has been obtained missing values. Missing value data has been imputed by the last data before the missing value and moving average techniques. The GSTAR (1) and GSTARI (1, 1) models have been combined with an imputation technique solved this problem. Based on the smallest RMSE value, the GSTARI (1, 1) model which has been combined with the imputation technique that used the last value was the best method because it produced the smallest RMSE when compared to other methods. Forecasting results shown that gold prices in the United States, United Kingdom, and Indonesia increased but gold prices in Turkey actually decreased. Forecasting gold prices in each of these countries become one of the references in investing in gold. Based on the results of gold price forecasting, gold prices changed but not significantly.

show abstract

Long-term forecasting of multivariate time series in industrial furnaces with dynamic Gaussian Bayesian networks

Cited by 21 publications

References 23 publications

Piecewise forecasting of nonlinear time series with model tree dynamic Bayesian networks

Piecewise forecasting of nonlinear time series with model tree dynamic Bayesian networks

Detecting Obstructive Apnea Episodes using Dynamic Bayesian Networks and ECG-based Time-Series

GSTARIMA Model with Missing Value for Forecasting Gold Price

Contact Info

Product

Resources

About