Deep learning in electrical utility industry: A comprehensive review of a decade of research

Mishra, Manohar; Nayak, Janmenjoy; Naik, Bighnaraj; Abraham, Ajith

doi:10.1016/j.engappai.2020.104000

Cited by 85 publications

(35 citation statements)

References 143 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…These models translate a linear relationship with the observed output data and rely on the assumption that future observations can be approximated based on past observations. Models that can be written using a single equation as expressed in Equation (1), with an autoregressive term, AR(p), where p is the autoregressive polynomial order and a moving average term, MA(q), where q is the moving average polynomial order, hence translating a linear combination of the preceding error terms up to lag q. Thus, the polynomial orders or degrees, p, q ∈ Z + , represent the maximum value for the AR and MA terms, respectively.…”

Section: Box-jenkins Methodsmentioning

confidence: 99%

“…The large-scale and complex nature of Power Systems Operation in a constantly changing environment poses a series of challenges for policy makers, regulators, market operators and participants (both the generation and demand sides) and transmission and distribution operators, among others [1]. One of these challenges is the task of electrical power load forecasting.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Stacking Ensemble Methodology Using Deep Learning and ARIMA Models for Short-Term Load Forecasting

et al. 2021

View full text Add to dashboard Cite

Short-Term Load Forecasting is critical for reliable power system operation, and the search for enhanced methodologies has been a constant field of investigation, particularly in an increasingly competitive environment where the market operator and its participants need to better inform their decisions. Hence, it is important to continue advancing in terms of forecasting accuracy and consistency. This paper presents a new deep learning-based ensemble methodology for 24 h ahead load forecasting, where an automatic framework is proposed to select the best Box-Jenkins models (ARIMA Forecasters), from a wide-range of combinations. The method is distinct in its parameters but more importantly in considering different batches of historical (training) data, thus benefiting from prediction models focused on recent and longer load trends. Afterwards, these accurate predictions, mainly the linear components of the load time-series, are fed to the ensemble Deep Forward Neural Network. This flexible type of network architecture not only functions as a combiner but also receives additional historical and auxiliary data to further its generalization capabilities. Numerical testing using New England market data validated the proposed ensemble approach with diverse base forecasters, achieving promising results in comparison with other state-of-the-art methods.

show abstract

Section: Box-jenkins Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Stacking Ensemble Methodology Using Deep Learning and ARIMA Models for Short-Term Load Forecasting

et al. 2021

View full text Add to dashboard Cite

show abstract

“…However, in production related reviews, deep RL has often been considered only in the context of other machine learning techniques as in Kang, Catal, and Tekinerdogan (2020) or Arinez et al (2020) and is not mentioned in an industrial intelligence context in Peres et al (2020), lacking in consolidation of the already obtained results. This is also apparent in other technology fields such as energy (Mishra et al 2020), process industry (Lee, Shin, and Realff 2018), or tool condition monitoring (Serin et al 2020).…”

Section: Introductionmentioning

confidence: 93%

Deep reinforcement learning in production systems: a systematic literature review

Panzer

Bender

2021

International Journal of Production Research

103

View full text Add to dashboard Cite

Shortening product development cycles and fully customisable products pose major challenges for production systems. These not only have to cope with an increased product diversity but also enable high throughputs and provide a high adaptability and robustness to process variations and unforeseen incidents. To overcome these challenges, deep Reinforcement Learning (RL) has been increasingly applied for the optimisation of production systems. Unlike other machine learning methods, deep RL operates on recently collected sensor-data in direct interaction with its environment and enables real-time responses to system changes. Although deep RL is already being deployed in production systems, a systematic review of the results has not yet been established. The main contribution of this paper is to provide researchers and practitioners an overview of applications and to motivate further implementations and research of deep RL supported production systems. Findings reveal that deep RL is applied in a variety of production domains, contributing to data-driven and flexible processes. In most applications, conventional methods were outperformed and implementation efforts or dependence on human experience were reduced. Nevertheless, future research must focus more on transferring the findings to real-world systems to analyse safety aspects and demonstrate reliability under prevailing conditions.

show abstract

“…In recent years, artificial intelligence technology is considered to provide a new solution to the problems faced by the electricity industry (Khargonekar and Dahleh, 2018;Mishra et al, 2020). The promise of artificial intelligence technology in power systems stems from the great results it has already achieved in other areas.…”

Section: Introductionmentioning

confidence: 99%

Selective Catalytic Reduction System Ammonia Injection Control Based on Deep Deterministic Policy Reinforcement Learning

et al. 2021

View full text Add to dashboard Cite

The control of flue gas emission in thermal power plants has been a topic of concern. Selective catalytic reduction technology has been widely used as an effective flue gas treatment technology. However, precisely controlling the amount of ammonia injected remains a challenge. Too much ammonia not only causes secondary pollution but also corrodes the reactor equipment, while too little ammonia does not effectively reduce the NOx content. In recent years, deep reinforcement learning has achieved better results than traditional methods in decision making and control, which provides new methods for better control of selective catalytic reduction systems. The purpose of this research is to design an intelligent controller using reinforcement learning technology, which can accurately control ammonia injection, and achieve higher denitrification effect and less secondary pollution. To train the deep reinforcement learning controller, a high-precision virtual denitration environment is first constructed. In order to make the virtual environment more realistic, this virtual environment was designed as a special structure with two decoders and a unique approach was used in fitting the virtual environment. A deep deterministic policy agent is used as an intelligent controller to control the amount of injected ammonia. To make the intelligent controller more stable, the actor-critic framework and the experience pool approach were adopted. The results show that the intelligent controller can control the emissions of nitrogen oxides and ammonia at the outlet of the reactor after training in virtual environment.

show abstract

Deep learning in electrical utility industry: A comprehensive review of a decade of research

Cited by 85 publications

References 143 publications

Stacking Ensemble Methodology Using Deep Learning and ARIMA Models for Short-Term Load Forecasting

Stacking Ensemble Methodology Using Deep Learning and ARIMA Models for Short-Term Load Forecasting

Deep reinforcement learning in production systems: a systematic literature review

Selective Catalytic Reduction System Ammonia Injection Control Based on Deep Deterministic Policy Reinforcement Learning

Contact Info

Product

Resources

About