A novel short receptive field based dilated causal convolutional network integrated with Bidirectional LSTM for short-term load forecasting

Javed, Umar; Ijaz, Khalid; Jawad, Muhammad; Khosa, Ikramullah; Ansari, Ejaz A.; Zaidi, Shabih H.; Rafiq, Muhammad; Shabbir, Noman

doi:10.1016/j.eswa.2022.117689

Cited by 31 publications

(13 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Dilated convolution is a phenomenal progress made in CNN. Inevitably the dilated causal convolution will deepen the neural network . In a 1-D sequence with input x ∈ double-struckR n , and filter normalf : { 0 , ... , k − 1 } → double-struckR the dilated causal convolution ( D c ) on an element s of the sequence is expressed as given below. D c ( s ) = ( x * d f ) ( s ) = prefix∑ i = 0 k − 1 normalf ( i ) . x s − d i d is the dilation factor of the residual blocks in the stack.…”

Section: Proposed Tcn Methodologymentioning

confidence: 99%

See 1 more Smart Citation

Deep Learning Based System Identification and Nonlinear Model Predictive Control of pH Neutralization Process

Durairaj

Xavier

Patnaik

et al. 2023

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

An essential step in the progression of nonlinear system identification is the inception of recurrent and convolutiontype deep learning methods in industrial units. Many chemical/ pharmaceutical/wastewater process units employ pH neutralization schemes to check the acidity and alkalinity of the product before bringing them for industrial production. The inherent nonlinear dynamics, especially during the neutralization of strong acid by a strong base, pose rigorous difficulties to model uncertainties, making it highly challenging to implement automatic pH control. This research endeavor focuses on building a deep Temporal Convolution Network (TCN) with a larger receptive field to learn the dynamics of the pH neutralization process. This method uses a 1-D causal convolution strategy with a residual learning framework to perform dilated causal convolutions. The simulation studies of the proposed TCN-based identification-scheme are adopted and executed in a Python environment for two important case studies, namely, (a) single tank pH process and (b) ETP (effluent treatment plant)-pH process. The proposed TCN framework manifested supremacy over predicted model responses of LSTM (long short-term memory) and MLP (multilayer-perceptron) architectures in terms of accuracy while requiring less training time. Furthermore, in this research, a state-of-art TCN-based NMPC (nonlinear model predictive control) and LSTM-based NMPC schemes are designed, implemented, and investigated where the control performance revealed the precision of the former.

show abstract

Section: Proposed Tcn Methodologymentioning

confidence: 99%

“…Inevitably the dilated causal convolution will deepen the neural network. 32 In a 1-D sequence with input x n , and filter…”

Section: Proposed Tcn Methodologymentioning

confidence: 99%

Deep Learning Based System Identification and Nonlinear Model Predictive Control of pH Neutralization Process

Durairaj

Xavier

Patnaik

et al. 2023

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

show abstract

“…Subsequently, the contribution of these variables to the prediction accuracy improvement was examined by trial and error, sometimes leading to shorter input sets for some of the horizons. Alternatively, other approaches, such as gradient boosting decision tree and Pearson correlation coefficient [ 102 ], attention mechanism [ 103 ], or Exploratory Data Analysis [ 104 ], are considered to have an effective contribution during input features reduction and selection. However, it is important to note that for each horizon, inputs remain the same for all machine-learning methods used in the present study.…”

Section: Case Studymentioning

confidence: 99%

A Machine Learning Model Ensemble for Mixed Power Load Forecasting across Multiple Time Horizons

Γιαμαρέλος

Papadimitrakis

Stogiannos

et al. 2023

Sensors

View full text Add to dashboard Cite

The increasing penetration of renewable energy sources tends to redirect the power systems community’s interest from the traditional power grid model towards the smart grid framework. During this transition, load forecasting for various time horizons constitutes an essential electric utility task in network planning, operation, and management. This paper presents a novel mixed power-load forecasting scheme for multiple prediction horizons ranging from 15 min to 24 h ahead. The proposed approach makes use of a pool of models trained by several machine-learning methods with different characteristics, namely neural networks, linear regression, support vector regression, random forests, and sparse regression. The final prediction values are calculated using an online decision mechanism based on weighting the individual models according to their past performance. The proposed scheme is evaluated on real electrical load data sensed from a high voltage/medium voltage substation and is shown to be highly effective, as it results in R2 coefficient values ranging from 0.99 to 0.79 for prediction horizons ranging from 15 min to 24 h ahead, respectively. The method is compared to several state-of-the-art machine-learning approaches, as well as a different ensemble method, producing highly competitive results in terms of prediction accuracy.

show abstract

“…Another similar model CNN-BiGRU was adopted by Xuan et al, [46], which gives a higher prediction as GRU has lower parameters affecting the prediction than LSTM. To deal with the limitations of multivariate load forecasting issues in LSTM, Javed et al [73] proposed a hybrid SRDCC-BiLSTM with improved generalization capability for multi-step and multivariate STLF. Compared to CNN-LSTM, the proposed approach exhibits 35% greater accuracy and can capture local trends in electrical load patterns.…”

Section: B: Recurrent Neural Network (Rnn Models)mentioning

confidence: 99%

Energy Demand Forecasting and Optimizing Electric Systems for Developing Countries

et al. 2023

View full text Add to dashboard Cite

under Grant 56.00.0000. 028.20.0005.19-440; and in part by the Innovation Design and Entrepreneurship Academy, (iDEA) project of Bangladesh Computer Council (BCC), Bangladesh under Grant Mujib1002021/680. We thank our colleagues from Advanced Intelligent Multidisciplinary Systems (AIMS) Laboratory, United International University (UIU), Bangladesh who provided insights and expertise that greatly assisted the research.

show abstract

A novel short receptive field based dilated causal convolutional network integrated with Bidirectional LSTM for short-term load forecasting

Cited by 31 publications

References 43 publications

Deep Learning Based System Identification and Nonlinear Model Predictive Control of pH Neutralization Process

Deep Learning Based System Identification and Nonlinear Model Predictive Control of pH Neutralization Process

A Machine Learning Model Ensemble for Mixed Power Load Forecasting across Multiple Time Horizons

Energy Demand Forecasting and Optimizing Electric Systems for Developing Countries

Contact Info

Product

Resources

About