Load Forecasting Techniques and Their Applications in Smart Grids

This review offers an in-depth examination of Deep Learning (DL) and Machine Learning (ML) techniques for smart grid load forecasting, emphasizing language precision, methodological rigor, and the exploration of novel contributions. The language used in this review is both technical and accessible, balancing complex concepts with clear explanations to cater to both specialists and general readers. It meticulously dissects contemporary DL models, including neural networks and ensemble methods, and evaluates their effectiveness through a detailed review of algorithms and frameworks. The methodology section systematically compares these techniques against traditional forecasting methods using performance metrics such as MAPE, RMSE, and MSE, ensuring a comprehensive assessment of their accuracy and scalability. A significant contribution of this review is its examination of real-world applications and case studies, which demonstrate how ML and DL techniques address practical challenges in energy management, such as grid stability and demand forecasting. Furthermore, the review introduces novel perspectives on the integration of probabilistic forecasting and ensemble methods, which offer innovative approaches for managing energy demand uncertainties. By identifying current limitations and proposing future research directions, this review not only advances the understanding of DL and ML applications in smart grids but also provides a foundation for future developments in this evolving field.

An effective dimensionality reduction approach for short-term load forecasting

Predicting stock market (SM) trends is an issue of great interest among researchers, investors and traders since the successful prediction of SMs’ direction may promise various benefits. Because of the fairly nonlinear nature of the historical data, accurate estimation of the SM direction is a rather challenging issue. The aim of this study is to present a novel machine learning (ML) model to forecast the movement of the Borsa Istanbul (BIST) 100 index. Modeling was performed by multilayer perceptron–genetic algorithms (MLP–GA) and multilayer perceptron–particle swarm optimization (MLP–PSO) in two scenarios considering Tanh (x) and the default Gaussian function as the output function. The historical financial time series data utilized in this research is from 1996 to 2020, consisting of nine technical indicators. Results are assessed using Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE) and correlation coefficient values to compare the accuracy and performance of the developed models. Based on the results, the involvement of the Tanh (x) as the output function, improved the accuracy of models compared with the default Gaussian function, significantly. MLP–PSO with population size 125, followed by MLP–GA with population size 50, provided higher accuracy for testing, reporting RMSE of 0.732583 and 0.733063, MAPE of 28.16%, 29.09% and correlation coefficient of 0.694 and 0.695, respectively. According to the results, using the hybrid ML method could successfully improve the prediction accuracy.

This study aims to create a monthly sales quantity budget by making use of the previous income data of an enterprise operating within the construction sector, which is considered the locomotive of the economy. For estimating time-series of sales as a linear model ARIMA (Auto-Regressive Integrated Moving Average), as nonlinear model LSTM (Long ShortTerm Memory) and a HYBRID (LSTM and ARIMA) model built to improve system performance compared to a single model was used. As a result of the study, Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE) and Mean Absolute Error (MAE) values obtained from each of the methods used in the application were compared, and a monthly sales volume budget was created for 2017 with all the methods used. When the MAPE and MSE values obtained from each of these methods were compared, the best performance was the Hybrid model that gave the lowest error, and in addition, the fact that all of the application models got very realistic results by using the historical data showed the success of the predictions.

A novel decomposition-ensemble forecasting system for dynamic dispatching of smart grid with sub-model selection and intelligent optimization

This study aims to design, implement, and compare the performance of the Backpropagation algorithm from Artificial Neural Networks and the Seasonal Autoregressive Integrated Moving Average (SARIMA) model in predicting the optimal daily rice requirement at Grillme Restaurant in Pontianak. The main problem faced by the restaurant is the uncertainty in determining the required daily rice stock, which periodically results in either understocking (shortage) or overstocking (wastage), leading to operational losses. To address this, the study utilizes historical daily rice sales data from January 2023 to April 2025 as the database for training and testing both predictive models. The SARIMA approach is employed to capture time series components (trend and seasonality), while Backpropagation is utilized to model non-linear patterns. Comparative test results indicate that the SARIMA model achieved superior accuracy compared to the Backpropagation model. This is confirmed by the Mean Absolute Percentage Error (MAPE) value of the SARIMA algorithm being 17.35%, which is lower than the MAPE value of Backpropagation at 19.62%. The MAPE values obtained by both models demonstrate good predictive capability, but it is concluded that SARIMA is more recommended for a more efficient and planned management of rice stock at Grillme Restaurant in Pontianak.

This paper delves into the engineering applications of lime-stabilized red clay, a highly water-sensitive material, particularly in the context of the climatic conditions prevalent in the Dalian region. We systematically investigate the impact of water content, dry density, and freeze-thaw cycles (with a freezing temperature set at -10°C) on the thermal conductivity of stabilized soil, a crucial parameter for analyzing soil temperature fields that is influenced by numerous factors. By developing and validating both empirical and machine learning prediction models, we unravel the evolution of thermal conductivity in response to these factors: within the range of influencing variables, thermal conductivity exhibits an exponential or linear increase with rising water content and dry density, while it decreases exponentially with increasing freeze-thaw cycles. Furthermore, we quantitatively analyze the specific influence of water content and other factors on the thermal conductivity of stabilized soil and construct a comprehensive prediction model encompassing BP neural network, gradient boosting decision tree, and linear regression models. Comparative analysis highlights the significant enhancement in prediction accuracy achieved by the proposed ensemble model over single machine learning models, with root mean square error (RMSE) values below 0.05 and mean absolute percentage error (MAPE) values remaining under 2.5% in both frozen and unfrozen states. Additionally, a secondary validation using experimental data from other researchers confirms the model's good agreement with previous results, demonstrating its robust generalization ability. Our findings provide valuable insights for engineering studies in the Dalian region and red clay areas subjected to extreme climatic conditions.

Stocks have high-profit potential but also have high risk. Many people have ways to forecast stock prices. The Geometric Brownian Motion (GBM) method forecasts stock prices. The data used in this study are closing stock price data from July 1, 2021 to August 31, 2021 taken from Yahoo! Finance. The stocks used in this research are Bank Rakyat Indonesia (BBRI), Indofood Sukses Makmur (INDF), and Telkom Indonesia (TLKM). A strategy is carried out to improve prediction accuracy by utilising the Kalman Filter (KF). This research will compare the mean absolute percentage error (MAPE) value between GBM-KF, which was manually computed and computed using the Python library. As an example of this research, for BBRI stock, the high GBM MAPE value of 9.02% can be reduced to 3.52% with manually computed GBM-KF and 3.68% with Python library computed GBM-KF. Similarly, INDF and TLKM stocks are showing a significant reduction in MAPE values to deficient levels in some cases. The GBM-KF method employing manual computing may enhance the overall precision of stock price forecasting. Future research may enhance this study by using the GBM-KF model on alternative financial instruments, integrating supplementary market data, or evaluating its efficacy under extreme market conditions.

Forecasting energy consumption is highly essential for strategic and operational planning. This study uses the Adaptive-Neuro-Fuzzy Inference System (ANFIS) for a midterm forecast of electricity consumption. The model comprises of three meteorological variables as inputs and electricity consumption as output. Two ANFIS models with two clustering techniques (Fuzzy c-Means (FCM) and Grid Partitioning (GP) were developed (ANFIS-FCM and ANFIS-GP) to forecast monthly energy consumption based on meteorological variables. The performance of each model was determined using known statistical metrics. This compares the predicted electricity consumption with the observed and a statistical significance between the two reported. ANFIS-FCM model recorded a better mean absolute deviation (MAD), root mean square (RMSE), and mean absolute percentage error (MAPE) values of 0.396, 0.738, and 8.613 respectively compared to the ANFIS-GP model, which has MAD, RMSE, and MAPE values of 0.450, 0.762, and 9.430 values respectively. The study established that FCM is a good clustering technique in ANFIS compared to GP and recommended a comparison between the two techniques on hybrid ANFIS model.

A survey on applications of Alternating Direction Method of Multipliers in smart power grids

Drought is a complex and frequently occurring natural hazard in many parts of the world. Therefore, accurate drought forecasting is essential to mitigate its adverse impacts. This research has inferred the implication and the appropriateness of the extreme learning machine (ELM) algorithm for drought forecasting. For numerical evaluation, time series data of the Standardized Precipitating Temperature Index (SPTI) are used for nine meteorological stations located in various climatological zones of Pakistan. To assess the performance of ELM, this research includes parallel inferences of multilayer perceptron (MLP) and autoregressive integrated moving average (ARIMA) models. The performance of each model is assessed using root mean square error (RMSE), mean absolute error (MAE), mean absolute percent error (MAPE), Kling‐Gupta efficiency (KGE), Willmott index (WI), and Karl Pearson’s correlation coefficient. Generally, graphical results illustrated an excellent performance of the ELM algorithm over MLP and ARIMA models. For training data of SPTI‐1, ELM’s best performance has observed at Chitral station (RMSE = 0.374, KGE = 0.838, WI = 0.960, MAE = 0.272, MAPE = 259.59, R = 0.93). For SPTI‐1 at Astore station, the numerical results are (RMSE = 0.688, KGE = 0.988, WI = 0.997, MAE = 0.798, MAPE = 247.35). The overall results indicate that the ELM outperformed by producing the smallest RMSE, MAE, and MAPE values and maximum values for KGE, WI, and correlation coefficient values at almost all the selected meteorological stations for (1, 3, 6, 9, and 12) month time scales. In summary, this research endorses the use of ELM for accurate drought forecasting.

Smart grid technologies and applications

Fisheries is one of the sources of food security in Indonesia which plays an important role in the rich biodiversity of Indonesian coastal waters. This research uses the Chen Fuzzy Time Series Model forecasting method. The samples used in this study are data on the number of fishery commodity products with specific Tilapia and Catfish monthly Aceh Province as much as 24 data, starting from January 2021 to December 2022. The results of forecasting the number of tilapia in January 2023 were 1,320,815, with a MAPE (Mean Absolute Percentage Error) value of 12.93%. And the result of forecasting the number of catfish in January 2023 was 2,339,568, with a MAPE (Mean Absolute Percentage Error) value of 22.9%. This shows that forecasting using the Chen model Fuzzy Time Series method is good enough to forecast the number of tilapia and catfish in Aceh Province.

<strong class="journal-contentHeaderColor">Abstract.</strong> A long short-term memory (LSTM) neural network is proposed to predict hurricane-forced significant wave heights (SWHs) in the Caribbean Sea (CS) based on a dataset of 20 CS, Gulf of Mexico, and western Atlantic hurricane events collected from 10 buoys from 2010–2020. SWH nowcasting and forecasting are initiated using LSTM on 0, 3, 6, 9, and 12 h horizons. Through examining study cases Hurricanes Dorian (2019), Sandy (2012), and Igor (2010), results illustrate that the model is well suited to forecast hurricane-forced wave heights much more rapidly at a significantly cheaper computational cost compared to numerical wave models, with much less required expertise. Forecasts are highly accurate with regards to observations. For example, Hurricane Dorian nowcasts had correlation (<span class="inline-formula"><i>R</i></span>), root mean square error (RMSE), and mean absolute percentage error (MAPE) values of 0.99, 0.16 m, and 2.6 %, respectively. Similarly, on the 3, 6, 9, and 12 h forecasts, results produced <span class="inline-formula"><i>R</i></span> (RMSE; MAPE) values of 0.95 (0.51 m; 7.99 %), 0.92 (0.74 m; 10.83 %), 0.85 (1 m; 13.13 %), and 0.84 (1.24 m; 14.82 %), respectively. In general, the model can provide accurate predictions within 12 h (<span class="inline-formula"><i>R</i>≥0.8</span>) and errors can be maintained at under 1 m within 6 h of forecast lead time. However, the model also consistently overpredicted the maximum observed SWHs. From a comparison of LSTM with a third-generation wave model, Simulating Waves Nearshore (SWAN), it was determined that when using Hurricane Dorian as a case example, nowcasts were far more accurate with regards to the observations. This demonstrates that LSTM can be used to supplement, but perhaps not replace, computationally expensive numerical wave models for forecasting extreme wave heights. As such, addressing the fundamental problem of phase shifting and other errors in LSTM or other data-driven forecasting should receive greater scrutiny from Small Island Developing States. To improve models results, additional research should be geared towards improving single-point LSTM neural network training datasets by considering hurricane track and identifying the hurricane quadrant in which buoy observations are made.

Path loss prediction in a multi-urban environment is presented in this paper. The multi-urban environments comprise seven identified environment for the study. Forecasting of the path loss in rural, and other areas such as urban, dense urban and so on performs a major part in order to effectively design wireless networks and mobile systems. Many researchers have exploited soft computing techniques to achieve accurate results over the years. However, in this paper, a unique novel approach of determining path loss employing ellipsoidal fuzzy inference set up is applied to the classified propagation environment. Adequate and reasonable results were obtained with superior Root Mean Square Error(RMSE) and Mean Absolute Percentage Error(MAPE) values for both 900MHz band and 1800MHz bands. Also, the optimal implementation is provided by both fuzzy logic type 2 and EFIS, which examined the path attenuation with an RMSE and a MAPE less than 2.17% for the 900MHz band and 2.64% for the 1800MHz band. It reveals again that the number of rules used in EFIS is lesser compared to the number of rules used in a fuzzy system, making EFIS a more preferable model.