Electric Forecasting Research Articles

Electricity demand forecasting using fuzzy hybrid intelligence-based seasonal models

PurposeThe purpose of this paper is to propose an extensiveness intelligent hybrid model to short-term load electricity forecast that can simultaneously model the seasonal complicated nonlinear uncertain patterns in the data. For this purpose, a fuzzy seasonal version of the multilayer perceptrons (MLP) is developed.Design/methodology/approachIn this paper, an extended fuzzy seasonal version of classic MLP is proposed using basic concepts of seasonal modeling and fuzzy logic. The fundamental goal behind the proposed model is to improve the modeling comprehensiveness of traditional MLP in such a way that they can simultaneously model seasonal and fuzzy patterns and structures, in addition to the regular nonseasonal and crisp patterns and structures.FindingsEventually, the effectiveness and predictive capability of the proposed model are examined and compared with its components and some other models. Empirical results of the electricity load forecasting indicate that the proposed model can achieve more accurate and also lower risk rather than classic MLP and some other fuzzy/nonfuzzy, seasonal nonseasonal, statistical/intelligent models.Originality/valueOne of the most appropriate modeling tools and widely used techniques for electricity load forecasting is artificial neural networks (ANNs). The popularity of such models comes from their unique advantages such as nonlinearity, universally, generality, self-adaptively and so on. However, despite all benefits of these methods, owing to the specific features of electricity markets and also simultaneously existing different patterns and structures in the electrical data sets, they are insufficient to achieve decided forecasts, lonely. The major weaknesses of ANNs for achieving more accurate, low-risk results are seasonality and uncertainty. In this paper, the ability of the modeling seasonal and uncertain patterns has been added to other unique capabilities of traditional MLP in complex nonlinear patterns modeling.

Short-Term Power Forecasting Framework for Microgrids Using Combined Baseline and Regression Models

Short-term electric power forecasting is a tool of great interest for power systems, where the presence of renewable and distributed generation sources is constantly growing. Specifically, this type of forecasting is essential for energy management systems in buildings, industries and microgrids for optimizing the operation of their distributed energy resources under different criteria based on their expected daily energy balance (the consumption–generation relationship). Under this situation, this paper proposes a complete framework for the short-term multistep forecasting of electric power consumption and generation in smart grids and microgrids. One advantage of the proposed framework is its capability of evaluating numerous combinations of inputs, making it possible to identify the best technique and the best set of inputs in each case. Therefore, even in cases with insufficient input information, the framework can always provide good forecasting results. Particularly, in this paper, the developed framework is used to compare a whole set of rule-based and machine learning techniques (artificial neural networks and random forests) to perform day-ahead forecasting. Moreover, the paper presents and a new approach consisting of the use of baseline models as inputs for machine learning models, and compares it with others. Our results show that this approach can significantly improve upon the compared techniques, achieving an accuracy improvement of up to 62% over that of a persistence model, which is the best of the compared algorithms across all application cases. These results are obtained from the application of the proposed methodology to forecasting five different load and generation power variables for the Savona Campus at the University of Genova in Italy.

Short-Term Electricity Generation Forecasting Using Machine Learning Algorithms: A Case Study of the Benin Electricity Community (C.E.B)

Time series forecasting in the energy sector is important to power utilities for decision making to ensure the sustainability and quality of electricity supply, and the stability of the power grid. Unfortunately, the presence of certain exogenous factors such as weather conditions, electricity price complicate the task using linear regression models that are becoming unsuitable. The search for a robust predictor would be an invaluable asset for electricity companies. To overcome this difficulty, Artificial Intelligence differs from these prediction methods through the Machine Learning algorithms which have been performing over the last decades in predicting time series on several levels. This work proposes the deployment of three univariate Machine Learning models: Support Vector Regression, Multi-Layer Perceptron, and the Long Short-Term Memory Recurrent Neural Network to predict the electricity production of Benin Electricity Community. In order to validate the performance of these different methods, against the Autoregressive Integrated Mobile Average and Multiple Regression model, performance metrics were used. Overall, the results show that the Machine Learning models outperform the linear regression methods. Consequently, Machine Learning methods offer a perspective for short-term electric power generation forecasting of Benin Electricity Community sources.

An Efficient Short-Term Electricity Forecasting Approach Based on EEMD-LSTM Model with Feature Factors

The forecasting of electricity consumption data plays an important role in the operation, planning, and security of the power grid. However, electricity data is affected by multiple factors and large fluctuations, which makes it difficult to accurately forecast. Traditionally, ARIMA and SVM are widely used for electricity forecasting based on historical consumption data. However, for non-stationary multi-feature data, traditional schemes cannot achieve deep feature mining of them, and the forecast results are inaccurate. To address this problem, this paper proposes an efficient short-term electricity forecasting approach based on EEMD-LSTM model. Firstly, we perform Savitzky-golay (SG) smoothing on the original data, and then introduce feature factors to the feature analysis. In particular, the proposed approach can reduce the random noise in data, as well as reduce the impact of data fluctuations, and effectively learn the long-term characteristics of the data. The simulation results show that, compared with ARIMA, LSTM, EMD-SVM, EMD-LSTM, the proposed approach can achieve better accuracy in the electricity forecasting.

A Short-Term Electricity Forecasting Scheme Based on Combined GRU Model with STL Decomposition

With development of smart grid, the stable operation of grid has put forward higher requirements for system dispatch. In particular, short-term load forecasting of power systems is a key factor of power grid management systems, which is related to the safety, economy, and stable operation of the smart grid. However, short-term electricity forecasting is affected by many external factors. It has complex characteristics, especially non-linear relationships, so it cannot be accurately predicted. Recently, Recurrent Neural Network based models have good performance in electricity forecasting because of their excellent ability to capture non-linear relationships. However, they cannot fully capture historical information, especially local historical information, which has an impact on prediction accuracy. In order to address these problems, we propose a scheme by combining STL decomposition and GRU model. Specifically, we first decompose the original time series into three different components by STL. The decomposition results are separately imported into the main prediction module, which uses two GRU networks with different structures to obtain the local and global dependencies of the data. We also add an autoregressive method to make the model more robust. The proposed scheme is validated based on real-world data, and the simulation results show that our proposed method can perfectly capture local and global information and achieve higher prediction accuracy than traditional models.

Long-term forecasting of electricity in East Kalimantan Province using the fuzzy logic approach

The increase in the number of residents has an impact on the increasing number of electricity customers each year, resulting in an increase in electricity consumption. In order to achieve the adjustment of generation and demand for electrical energy needs electricity system planning is needed, one of which is forecasting the consumption of electrical energy in the future. This aims to realize optimization in the process of supplying electrical energy. In this study, the application of the fuzzy logic method is used to process the rules of the input variables and produce forecasting of long-term annual electricity consumption in East Kalimantan Province until 2028. In addition, the Exponential Smoothing method is also used as a comparison method to determine the accuracy of forecasting calculation models. Electrical energy forecasting has characteristics that are usually influenced by factors such as population, number of customers, GRDP and so on. Input data used for learning fuzzy logic in this forecasting are the actual data for 2010-2018 and the projected input data for 2019-2028 are used to obtain forecasting results. Then the forecasting data for 2019-2028 is compared with the PLN RUPTL projection data. From the results of forecasting, the best level of accuracy with the smallest error value is obtained using the fuzzy logic method that is with a MAPE value of 3.80%. This research is expected to be a consideration for forecasting electricity consumption models for the planned transfer of Indonesia’s new capital in the area of East Kalimantan Province.

A Multi-Scale Electricity Consumption Prediction Algorithm Based on Time-Frequency Variational Autoencoder

Accurate electricity consumption forecasting can be treated as a reliable guidance for power production. However, traditional electricity forecasting models suffer from simultaneously capturing the periodicity and the volatility of sequential electricity consumption data, while the periodicity and the volatility are important for electricity forecasting. In order to effectively model this sequential data and predict electricity consumption accurately, we propose a multi-scale prediction (Long Short Term Memory, LSTM) algorithm based on Time-Frequency Variational Autoencoder (TFVAE-LSTM). The proposed algorithm treats the sequential data as a superposition of data in different frequencies, it defines an encoder in frequency domain to extract frequency features to model the periodicity and volatility, and defines a decoder in time domain to capture the sequential features of data. Based on the extracted Time-Frequency features in a TFVAE, a multi-scale LSTM model is defined to further extract sequential features from different scales to predict electricity consumption. Experiments show the effectiveness of the proposed TFVAE-LSTM for electricity consumption forecasting tasks.

Efficient Electricity Forecasting in Multiple Residential Buildings considering Demand-Side Management

Gateway based Home Area Network (HAN ) to Neighbourhood Area Network (NAN); NAN to HAN improved the communication in the Smart grid. The gateway reduces the Load dispatch centre (LDC) work in varying power consumption in a short time interval. The proposed work explains the working of gateway, reducing the work of LDC using the load scheduling procedure. Deep learning methods incorporated gateway will aid in achieving the requirement. It reduces black start operation and it may be prevented by indulging consumers in the supply automation of the grid. It will produce the grid to maintain the operating frequency, avoiding the substations' disciplinary charges. A variety of types of abrupt load variation and load kinds has been taken in this function. The analysis shows that the gateway achieves a decrease in complexity in the proposed work. This method provides the detail of employment of deep learning for predicting the load forecasting performances in smart grids that can be made better through a gateway between SMs-DCU. The Proposed work compares with systems that employ the predictable profound estimating methods for load forecasting, which has provided better performance. Load Prediction, Deep Learning, Gateway, Smart Grid, Estimation

A Review on Deep Learning with Focus on Deep Recurrent Neural Network for Electricity Forecasting in Residential Building

The rapid increase in urbanization has resulted in a significant rise in electricity consumption, which resulted in a wide gap between the amount of electricity generated and the consumer’s demand. Literature shows that 40% of the generated electricity is consumed by building sectors. To address the gap between demand and supply, there is a need to develop novel prediction models that adopt automated techniques, to dynamically predict buildings energy consumption. An efficient load forecast can assist in efficient power generation and distribution among users. Different Machine Learning techniques have been applied in future electricity consumption forecasts with the need for a more ideal solution. This paper reviews methods for building energy consumption forecasts that use Machine Learning algorithms like Artificial Neural Network, Deep Belief Network, Recurrent Neural Network, Elman Neural Network, Deep Recurrent Neural Network, Convolutional Neural Network and Nonlinear Autoregressive Network. The review explores existing research gaps and research directions for future work. Finally, a novel Deep Learning framework was suggested for future work on enhancing prediction performance and reliability using occupancy profile and distinct climatic scenarios based on Transfer Learning and LSTM algorithms (Trans-LSTM) for medium to long term electricity consumption forecast.

A hybrid multi-objective optimizer-based model for daily electricity demand prediction considering COVID-19

Electricity consumption has been affected due to worldwide lockdown policies against COVID-19. Many countries have pointed out that electricity supply security during the epidemic is critical to ensuring people’s livelihood. Accurate prediction of electricity demand would act a more important role in ensuring energy security for all the countries. Although there have been many studies on electricity forecasting, they did not consider the pandemic, and many works only considered the prediction accuracy and ignored the stability. Driven by the above reasons, it is necessary to develop an electricity consumption prediction model that can be well applied in the pandemic. In this work, a hybrid prediction system is proposed with data processing, modelling, and optimization. An improved complete ensemble empirical mode decomposition with adaptive noise is used for data preprocessing, which overcomes the shortcomings of the original method; a multi-objective optimizer is adopted for ensuring the accuracy and stability; support vector machine is used as the prediction model. Taking daily electricity demand of US as an example, the results prove that the proposed hybrid models are superior to benchmark models in both prediction accuracy and stability. Moreover, selection of input parameters is discussed, and the results indicate that the model considering the daily infections has the highest prediction accuracy and stability, and it is proved that the proposed model has great potential in real-world applications.

ELECTRICITY PRICE FORECASTING MODELS

New era has begun two decades ago in the most power sectors of the globe. As electricity itself shifted from being a commodity only to a competitive trading market instrument, forecasting of power load and prices started to be another important factor. With the introduction of liberal power markets, independent players like producers, operators, traders became an important factors of liberalized power markets. New era in this secotr also increased competition among companies and countries. Thus, as new players came into scene, electricity became to be a tradable product and similar to another commodities needs to account, forecast plan of electiricity brought new challenges. With its obstacles, new atmosphere of electiricity repormation brought benefits such as lower price to end users and more utilized energy systems among the countries. Different forecasting models have been developed in order to forsee future operations. Moreover, popular modelts from economics, such as game theories, cournot model, Bertrand model, nash equilibrium plays a big role in electricity price forecasting. Simulation modelts are another methods heavily used by producers, operators and power traders in market. Additionally, statistical models, such as moving averages, Root Mean Square Error, Mean Absolute Error, mean Absolute Percentage Error, Theil's inequality coefficient are frequesntly used to determine price movements of electiricity. As its broadly used in several different industries, Time Series models using historical information and adding updated information helps to model future movement of prices. This article discusses the several methods of price forecasting of electricity in liberal power markets as high volatility of the market raises a big risk for all participants. Forecasting models explains time series analysis and briefly discusses about autoregressive, moving average, autoregressive moving average, and seasonal autoregressive moving average models. Moreover, the article illustrates examples from electricity price and load forecasts and their comparisons with actual results from trades in Turkish electricity market.

Regression modeling for enterprise electricity consumption: A comparison of recurrent neural network and its variants

Effective electricity consumption forecasting is extremely significant for enterprises’ electricity planning which can provide data support for production decision, thus improving the level of enterprises' clean production. In recent years, recurrent neural network (RNN) and its variants have led to extensive research for time series forecasting. However, the performance and selection of these models in enterprise electricity forecasting have not been reported. With this study, we attempted to back some of these solutions with experimental results. This paper focused on a comparison for daily enterprise electricity consumption forecasting using different RNN models, i.e, standard RNN, long short-term memory-based RNN (LSTM), and gated recurrent unit-based RNN (GRU). To test their regression performance, three Chinese enterprises with different scales of electricity consumption are investigated. The comparison results show that the LSTM and the GRU models are slightly better than that of the RNN in terms of normalized root-mean-square error, mean absolute percentage error and threshold statistic. Moreover, the GRU model with the simplest structure is significantly different from the RNN, but not from LSTM in terms of Friedman testing. Hence the GRU model can be regarded as the first candidate for the enterprise electricity consumption forecasting in the future work.

Short-Term Electricity Price Forecasting with Recurrent Regimes and Structural Breaks

This paper develops a new approach to short-term electricity forecasting by focusing upon the dynamic specification of an appropriate calibration dataset prior to model specification. It challenges the conventional forecasting principles which argue that adaptive methods should place most emphasis upon recent data and that regime-switching should likewise model transitions from the latest regime. The approach in this paper recognises that the most relevant dataset in the episodic, recurrent nature of electricity dynamics may not be the most recent. This methodology provides a dynamic calibration dataset approach that is based on cluster analysis applied to fundamental market regime indicators, as well as structural time series breakpoint analyses. Forecasting is based upon applying a hybrid fundamental optimisation model with a neural network to the appropriate calibration data. The results outperform other benchmark models in backtesting on data from the Iberian electricity market of 2017, which presents a considerable number of market structural breaks and evolving market price drivers.

Economic implications of forecasting electricity generation from variable renewable energy sources

Short-term forecasting of electricity generation from variable renewable energy sources is not an end in itself but should provide some net benefit to its user. In the case of electricity trading, which is in the focus of this paper, the benefit can be quantified in terms of an improved economic outcome. Although some effort has been made to evaluate and to improve the profitability of electricity forecasts, the understanding of the underlying effects has remained incomplete so far. In this paper, we develop a more comprehensive theoretical framework of the connection between the statistical and the economic properties of day-ahead electricity forecasts. We find that, apart from the accuracy and the bias, which have already been extensively researched, the correlation between the forecast errors and the market price spread determines the economic implications - a phenomenon which we refer to as ‘correlation effect’. Our analysis is completed by a case study on solar electricity forecasting in Germany which illustrates the relevance and the limits of both our theoretical framework and the correlation effect. • The economics of short-term electricity forecasts are studied analytically and empirically. • The value and the risk of forecast-based electricity trading are evaluated. • We find that the economics depend on the correlation of forecast errors with prices. • Hence, the most accurate forecast is not necessarily the most accurate.

Issues of an improving the accuracy of energy carriers production forecasting in a computer-aided system for monitoring the operation of a gas-fired cogeneration plant

The article presents issues related to improving the accuracy of forecasting of combined electricity and heat production in a selected gas-fired CHP plant equipped with two gas turbines with heat exchangers and a heat accumulator. To improve the accuracy of forecasting the production of energy carriers in the considered CHP plant, it has been proposed to extend the existing computer system for supervising operation with an additional calculation module. The task of this module is to correct the forecasts for electricity and heat production obtained using gas turbine model based on correction curves. In the additional computational module of the computer system, it is proposed to use empirical functions developed with the use of operational measurements of the analysed gas turbines. In the process of preparing operational measurement data for the construction of static empirical functions, the GLR (Generalized Likelihood Ratio) algorithm was used to detect the process changes and identify the periods of operation of gas turbines in a close to steady-state. The obtained results of the analyses confirm the possibility of their use in the existing computer system of operation supervision of the CHP plant. The developed empirical models are characterized by the high quality of prediction, which is confirmed by the obtained high values of determination coefficients and low error values of the model accuracy measures. The values of the coefficient of determination R^2 are within the following ranges: from 0.9754 to 0.9936 (for gas turbine GT-1) and from 0.9138 to 0.9939 (for gas turbine GT-2). The developed empirical functions can be implemented in an additional calculation module of the computer system, increasing the accuracy of planning the production of energy carriers in a gas-fired power plant.

Towards Efficient Electricity Forecasting in Residential and Commercial Buildings: A Novel Hybrid CNN with a LSTM-AE based Framework.

Due to industrialization and the rising demand for energy, global energy consumption has been rapidly increasing. Recent studies show that the biggest portion of energy is consumed in residential buildings, i.e., in European Union countries up to 40% of the total energy is consumed by households. Most residential buildings and industrial zones are equipped with smart sensors such as metering electric sensors, that are inadequately utilized for better energy management. In this paper, we develop a hybrid convolutional neural network (CNN) with an long short-term memory autoencoder (LSTM-AE) model for future energy prediction in residential and commercial buildings. The central focus of this research work is to utilize the smart meters’ data for energy forecasting in order to enable appropriate energy management in buildings. We performed extensive research using several deep learning-based forecasting models and proposed an optimal hybrid CNN with the LSTM-AE model. To the best of our knowledge, we are the first to incorporate the aforementioned models under the umbrella of a unified framework with some utility preprocessing. Initially, the CNN model extracts features from the input data, which are then fed to the LSTM-encoder to generate encoded sequences. The encoded sequences are decoded by another following LSTM-decoder to advance it to the final dense layer for energy prediction. The experimental results using different evaluation metrics show that the proposed hybrid model works well. Also, it records the smallest value for mean square error (MSE), mean absolute error (MAE), root mean square error (RMSE) and mean absolute percentage error (MAPE) when compared to other state-of-the-art forecasting methods over the UCI residential building dataset. Furthermore, we conducted experiments on Korean commercial building data and the results indicate that our proposed hybrid model is a worthy contribution to energy forecasting.

A novel composite electricity demand forecasting framework by data processing and optimized support vector machine

Reliable forecast of electricity can encourage accessible and responsible information for scholars, policymakers, end-consumers and managers of the electricity market. Numerous electricity forecasting methods have been achieved commendably, the performance of which varies depending on numerical characteristics and operational conditions. In this study, the composite forecasting concept is introduced and implemented to show the potential of forecasting performance. This modeling concept is a remarkable ability to identify and measure any seasonal relationship that exists in electricity demand data. Moreover, it is available as a toolbox in many of the programming operation research. In the module of nonlinear time series decomposition, the noise disturbance is initially considered before extracting the seasonal variation to support the condition that the linear and stationary time series should be used for the seasonality identifying method. Also, we further provide a new insight of prediction intervals estimation to better reflect the uncertainties of the underlying challenging power system plan and operation. The results show that the proposed model can generate promising forecasts compared to the other combination schemata and it can be useful for both policy-makers and public agencies to guarantee the security and regulation of the power system.

Integrating Blockchain and Broad Learning for Smart Energy Innovation: Design and Experiment

With the rapid development in recent years, the architecture and data sources of energy internet (EI) becomes increasingly complicated. It is a good opportunity and trend to combine energy equipment and information based on blockchain technology. However, high computational complexity inherently due to the blockchain must be addressed to achieve rapidly user electricity prediction. In this paper, an algorithm model of a blockchain technology combined with a broad learning data fusion algorithm is proposed. A novel and detailed combination model is introduced, mainly including CNN and LSTM model. Combination model considers energy data and blockchain data as parameter, and defines selection logic to guarantee user electricity forecast accuracy. Our proposed solution takes the advantage of CNN model in prediction accuracy and computation complexity, and advantage of LSTM model in long term memory. Further experiments illustrate that our solution achieves higher prediction accuracy of 94.1% (without considering the influence of real-time data), with the best result achieved ever. The experimental prediction results provide underlying insights to guide the direction of system resource planning.

Forecasting the electrical energy needs in Bangka Island

Bangka Island is one of the two largest islands in Bangka Belitung Islands Province. The total population in Bangka Island in 2018 was 1146654 persons. The growth of the population every year is accompanied by an increase in electricity demand. The increase in electricity demand is also caused by economic factors, namely GDP growth and regional development factor. By using LEAP software, forecasting of electricity needs for the year 2019-2028 are obtained. The number of electricity customers grows by an average of 1.83%/year while average growth of electricity consumption is 11.697 %/year. The total electricity production in 2028 is estimated to reach 2.47GWh with peak load of 394.07 MW. The average peak load growth during 2019-2028 is 10.48%/year. Fulfillment of electricity demand will be met by the construction of 4 new types of power plants and the addition of submarine cables from Substation Tanjung Api-Api to Substation Mentok in 2020.

Electric Forecasting using Nature Inspired Optimization Techniques

Now a day, there exists huge competition among power industries in terms of fulfilling various customers’ electrical needs. Reliable and quality power supply is no doubt a basic need for all power consumers. Moreover, planning & operation engineers also targets for proper unit commitment, economic power dispatch, etc., and highly depend upon good power system planning. Therefore, Electric Forecasting (EF) is a major criterion for power engineers. In this manuscript, Artificial Neural Network (ANN), being a well established tool for modeling non-linear and black box systems, is used to forecast hydro generation power plant, energy met and peak demand of India. Furthermore, in this competitive world, ANN model is further optimized using genetic algorithm (GA) and particle swarm optimization (PSO) to explore accurate forecasting model with minimal amount of error. These optimization methods explore highly diversified search area, resulting in more accurate forecasting results in comparison to ANN when trained with standard back propagation training algorithm.