Energy forecasting in smart grid systems: recent advancements in probabilistic deep learning

Abstract

Energy forecasting plays a vital role in mitigating challenges in data rich smart grid (SG) systems involving various applications such as demand-side management, load shedding, and optimum dispatch. Managing efficient forecasting while ensuring the least possible prediction error is one of the main challenges posed in the grid today, considering the uncertainty in SG data. This paper presents a comprehensive and application-oriented review of state-of-the-art forecasting methods for SG systems along with recent developments in probabilistic deep learning (PDL). Traditional point forecasting methods including statistical, machine learning (ML), and deep learning (DL) are extensively investigated in terms of their applicability to energy forecasting. In addition, the significance of hybrid and data preprocessing techniques to support forecasting performance is also studied. A comparative case study using the Victorian electricity consumption in Australia and American electric power (AEP) datasets is conducted to analyze the performance of deterministic and probabilistic forecasting methods. The analysis demonstrates higher efficacy of DL methods with appropriate hyper-parameter tuning when sample sizes are larger and involve nonlinear patterns. Furthermore, PDL methods are found to achieve at least 60% lower prediction errors in comparison to other benchmark DL methods. However, the execution time increases significantly for PDL methods due to large sample space and a tradeoff between computational performance and forecasting accuracy needs to be maintained.

A hybrid prediction method for short‐term load based on temporal convolutional networks and attentional mechanisms

“…At present, the power load comes from a variety of sources, and is susceptible to external factors such as weather and economy [33,34]. As a result, the increase in non-linearity and instability in power load data makes it challenging to accurately predict power loads [35]. Currently, power load forecasting methods can be divided into classical methods and in-depth learning methods.…”

Macromodeling of Local Power Supply System Balance Forecasting Using Fractal Properties of Load and Generation Schedules

“…Application of systems with fuzzy logic, i.e. sets with a set of elements of arbitrary nature, which do not clearly define belonging to a certain set, allows to eliminate the disadvantages of artificial neural networks [3].…”

Variational Regression for Multi-Target Energy Disaggregation

“…From the consumer side, NILM constitutes a vital part of intelligent home systems, providing insights into reducing energy waste, raising energy awareness [ 3 , 4 ], improving the operational efficiency of installations [ 5 , 6 , 7 ], and creating smart alert mechanisms for residents in need [ 8 , 9 , 10 ]. On the other hand, DSOs can use NILM as a building block for various applications regarding the management and efficient monitoring of the grid [ 11 , 12 ] in combination with more accurate energy consumption forecasts [ 13 , 14 ]. In a similar fashion, disaggregation can be performed in other quantities that are used in residential buildings, such as natural gas [ 15 ] and potable water [ 16 , 17 , 18 ], in order to preserve resources and reduce the overall living costs of habitats.…”