Electric Load Prediction Research Articles

LSTM-Based Online Prediction Method for Building Electric Load During COVID-19

Accurate prediction of electric load is critical to optimally controlling and operating buildings. It provides the opportunities to reduce building energy consumption and to implement advanced functionalities such as demand response in the context of smart grid. However, buildings are nonstationary and it is important to consider the underlying concept changes that will affect the load pattern. In this paper we present an online learning method for predicting building electric load during concept changes such as COVID-19. The proposed methods is based on online Long Short-Term Memory (LSTM) recurrent neural network. To speed up the learning process during concept changes and improve prediction accuracy, an ensemble of multiple models with different learning rates is used. The learning rates are updated in realtime to best adapt to the new concept while maintaining the learned information for the prediction.

Electrical load prediction of healthcare buildings through single and ensemble learning

Healthcare buildings are characterized by complex energy systems and high energy usage, therefore serving as the key areas for achieving energy conservation goals in the building sector. An accurate load prediction of hospital energy consumption is of paramount importance to a successful healthcare building energy management. In this study, eight machine learning models of single learning and ensemble learning were developed for predicting healthcare facilities’ energy consumption. To validate the performance of the proposed model, an experiment was conducted on a general hospital in Shanghai, China. It was found that the two ensemble models, Extreme Gradient Boosting (XGBoost) model and Random Forest (RF) model, outperformed single models in daily electrical load prediction. A further comparison between models trained with daily and weekly temporal resolution electrical data shows that it is more likely to achieve higher accuracy with finer time granularity. Through feature importance analysis, the most influential features under the daily and weekly electrical load prediction were identified. Based on the prediction results, it is expected that hospital facility managers will be able to conveniently assess the expected energy usage of their hospitals with the machine learning models.

On the long-term density prediction of peak electricity load with demand side management in buildings

Long-term daily peak demand forecast plays an important role in the effective and economic operations and planning of power systems. However, many uncertainties and building demand variability, which are associated with climate and socio-economic changes, complicate demand forecasting and expose power system operators to the risk of failing to meet electricity demand. This study presents a new approach to provide the long-term density prediction of the daily peak demand. Specifically, we make use of temperature projections from physics-based global climate models and calibrate the projections to address possible biases. In addition, the effects of population growth and demand side management efforts in buildings are taken into consideration. Finally, the daily peak demands are modeled with the nonhomogeneous generalized extreme value distribution where the parameters are allowed to vary, depending on the predicted temperature and population. A case study using actual building use data in the south-central region in Texas demonstrates that the proposed approach can quantify the uncertainties in an integrative framework and provide useful insights into the long-term evolution of peak demand density. A well-established building demand saving strategy is predicted to buffer against the growing needs of long-term peak electricity demand.

Elements of Randomized Forecasting and Its Application to Daily Electrical Load Prediction in a Regional Power System

A randomized forecasting method based on the generation of ensembles of entropy-optimal forecasting trajectories is developed. The latter are generated by randomized dynamic regression models containing random parameters, measurement noises, and a random input. The probability density functions of random parameters and measurement noises are estimated using real data within the randomized machine learning procedure. The ensembles of forecasting trajectories are generated by the sampling of the entropy-optimal probability distributions. This procedure is used for the randomized prediction of the daily electrical load of a regional power system. A stochastic oscillatory dynamic regression model is designed. One-, two-, and three-day forecasts of the electrical load are constructed, and their errors are analyzed.

RETRACTED: Combination of cuckoo search and wavelet neural network for midterm building energy forecast

The electrical load prediction for buildings plays a critical role in the smart-grid paradigm, since accurate predictions provide efficient energy management. A synthetic approach has been used in two buildings as the case studies with wavelet neural network (WNN) as a preparation for a process from the signal assessment perspective to forecast the density of electricity requirements. In this paper, singular spectrum analysis (SSA) and WNN based forecast engine have been considered. In this model, the free parameters of WNN are tuned optimally by cuckoo search (CS) algorithm. Using parameters of ten-dimensional variables of 29 weekdays as learning samples, this technique has been performed in a hotel and a mall, where the used electricity pattern respectively denoted dynamic and static series. By comparing the proposed approach with other models, it can be claim that WNN can generally enhance the forecasting precision for the hotel, although it is not essential for the mall. Particularly, the analogous stable amount that is about 0.65 W/m2of absolute error was gotten for the mall and the hotel buildings, where ε was less than 0.1. Simultaneously, the stable quantitative magnitudes of relative errors were around 4% and 6% for the mall and the hotel, respectively. Using the brief historical measurement, real-time forecasting approach of interim electricity requirement is proposed which can be applied to have smarter energy control.

A hybrid transfer learning model for short-term electric load forecasting

Transfer learning approach can be applied to electric load forecasting because electric load data from nearby locations are significantly correlated. However, ordinary transfer learning methods may bring negative transfer into load forecasting as time series prediction is not exactly the same as traditional data regression problem. Consequently, this paper proposes a novel hybrid transfer learning model based on time series decomposition. Firstly, trend and seasonal components are handled by standard machine learning approach so that seasonal cycles of electric load data can be interpreted better. Secondly, two-stage transfer regression model is established to forecast the irregular component in order to improve the forecasting accuracy. The negative transfer is successfully avoided, and the prediction accuracies are significantly improved because of time series decomposing and the additional information provided by the related dataset. The case study presented by two real-world power load datasets illustrates that the proposed approach can improve electric load prediction for a location by 30% at most by using additional data from another location.

Multimodal Feature Extraction and Fusion Deep Neural Networks for Short-Term Load Forecasting

Precise and reliable forecasting of short-term electricity load is essential to the development of smart grids. Particularly, deep neural networks (DNNs) are widely utilized for the prediction of short-term electricity load due to their automatic feature extraction ability. However, these available stacked deep-learning models may lose some temporal features or spatial features of original input data. To capture more comprehensive information, in this article, we present an integration scheme based on empirical mode decomposition (EMD), similar day methods, and DNNs to perform short-term load forecasting. It is especially worth noting that the electricity price is also an important factor for load variation, which is considered in our proposed scheme. Specifically, there are two primary layers: a feature extraction layer and a forecasting layer. In the feature extraction layer, EMD is applied to decompose load time series into several components, which are arranged into the 2-D input matrix of the convolutional neural network (CNN). Both the output vectors of the CNN and the raw load sequences are fed into the long short-term memory (LSTM) layer. Therefore, the whole EMD based CNN-LSTM approach extracts multimodal spatial-temporal features from input data. Meanwhile, the electricity price data is utilized to obtain multimodal spatial-temporal features in the same way. Additionally, the day and hour information and loads of similar days are to augment extra features for prediction. In the forecasting layer, the forecasting task is accomplished through a fully-connected neural network based on the outputs of the feature extraction layer. Leveraging these techniques enables our proposed scheme to extract more latent features, which significantly improve the accuracy. In order to demonstrate the performance of our proposed scheme, related experiments are conducted on actual data from the electricity market in Singapore. Compared to other available models, our proposed scheme is superior in graphic and numerical results.

Probabilistic prediction of electric power extreme load

For load forecasting, the difficulty lies in how to accurately grasp some key points in the process of load fluctuation, such as the highest load and the lowest load. It is advisable to call the forecast of daily maximum load and daily minimum load as “extreme load forecast”. On one hand, we can use the whole day load curve forecasting technology to directly get the forecasting results of extreme load; on the other hand, we need to study the special method of directly forecasting extreme load. At the same time, the conventional extreme load forecasting results are generally deterministic. Only a precise value is given, which cannot estimate the probability of the load value and determine the possible fluctuation range of the forecasting results, ignoring the probability characteristics of the forecasting results themselves. In fact, because of the advance of the prediction problem, the realization of the uncertainty prediction is more in line with the objective needs. According to the probability characteristics of the prediction results, it is helpful for the decision-makers to better grasp the objective laws of the research objects in the aspects of risk analysis and reliability evaluation, so as to achieve more reliable and scientific analysis and evaluation. Therefore, it is of great significance to introduce the idea of uncertainty analysis to realize the probability prediction of extreme load.

Electrical Load Prediction for Short Term using Support Vector Machine Techniques

The electrical load prediction during an interval of a week or a day plays an important role for scheduling and controlling operations of any power system. The techniques which are presently being used and are used for Short Term Load Forecasting (STLF) by utilizing various prediction models try for the performance improvement. The prediction models and their performance mainly depend upon the training data and its quality. The different forecasting approaches using Support Vector Machine (SVM) depending on several performance indices has been discussed. The accuracy of the forecasting approaches is measured by Mean Absolute Error (MAE), Root Mean Square Error (RMSE), prediction speed and training time. The approach with least RMSE reveals as the best among the SVM methods for short term load forecasting.

Study on deep reinforcement learning techniques for building energy consumption forecasting

Reliable and accurate building energy consumption prediction is becoming increasingly pivotal in building energy management. Currently, data-driven approach has shown promising performances and gained lots of research attention due to its efficiency and flexibility. As a combination of reinforcement learning and deep learning, deep reinforcement learning (DRL) techniques are expected to solve nonlinear and complex issues. However, very little is known about DRL techniques in forecasting building energy consumption. Therefore, this paper presents a case study of an office building using three commonly-used DRL techniques to forecast building energy consumption, namely Asynchronous Advantage Actor-Critic (A3C), Deep Deterministic Policy Gradient (DDPG) and Recurrent Deterministic Policy Gradient (RDPG). The objective is to investigate the potential of DRL techniques in building energy consumption prediction field. A comprehensive comparison between DRL models and common supervised models is also provided.The results demonstrate that the proposed DDPG and RDPG models have obvious advantages in forecasting building energy consumption compared to common supervised models, while accounting for more computation time for model training. Their prediction performances measured by mean absolute error (MAE) can be improved by 16%-24% for single-step ahead prediction, and 19%-32% for multi-step ahead prediction. The results also indicate that A3C performs poor prediction accuracy and shows much slower convergence speed than DDPG and RDPG. However, A3C is still the most efficient technique among these three DRL methods. The findings are enlightening and the proposed DRL methodologies can be positively extended to other prediction problems, e.g., wind speed prediction and electricity load prediction.

ARIMAX Model for Short-Term Electrical Load Forecasting

The scope for ARIMAX approach to forecast short term load has gained a lot of significant importance.In this paper, ARIMAXmodel which is an extension of ARIMA model with exogenous variables is used for STLF on a time series data of Karnataka State Demand pattern. The forecasting accuracy of ARIMA model is enhanced by taking into consideration hour of the day and day of the week as exogenous variables for ARIMAX model. Forecasting performance is thus improved by considering these significant load dependent factors. The forecasted results indicate that the proposed model is more accurate according to mean absolute percentage error (MAPE) obtained during the testing period of the model. As the historical load data are available on the databases of the utility, researches in the areas of time series modelling are ongoing for electrical load forecasting. In the proposed paper real time demand data available on Karnataka Power Transmission Corporation Ltd. (KPTCL) website is taken to develop and test the proposedload forecasting model.The power utility system operational costs and its securitydepend on the load forecasting for next few hours. Regional load forecasting helps in the accurate management performance of generation of power plant. Today’s deregulated markets have great demand for prediction of electrical loads, required for generating companies. There has been tremendous growth in electric power demand and hence it is very much essentialfor the utility sectors to have theirdemand information in advance.

Stacked autoencoders and extreme learning machine based hybrid model for electrical load prediction

Stacked autoencoders and extreme learning machine based hybrid model for electrical load prediction

Hybrid teaching—learning artificial neural network for city-level electrical load prediction

Hybrid teaching—learning artificial neural network for city-level electrical load prediction

On Linear Regression and other advanced algorithms for electrical load forecast using weather and time data

Energy field plays an important role in commercial world and makes a vital part of humanity. The paper considers the prediction problem concerning spent electrical energy in the Republic of Tatarstan, Russia. In fact, the task of electrical load prediction was set for each hour of the specified period. We solve the problem using Machine learning methods. Four models are considered. These are Linear Regression, Decision Tree, Random Forest and Gradient Tree Boosting. Linear Regression method shows good results, the average relative error is 3.98%. Decision Tree and Random Forest show the worst result, the average relative error is 10.44%. Gradient Tree Boosting show the best result, the average relative error made 2.17%. At the same time, Linear Regression model is much faster than other ones and more useful in industry. In the paper we show that using several techniques we can improve results for Linear Regression, such that it will be close to another advanced algorithms. The average relative error that is less than 5% is considered as a high enough result. The solution of the problem with a small error allows us to prevent the accidents related to electric overload. We assume that load depends on date, time and temperature. Then input variable models were obtained from these data. It is common approach to use that parameters as input data but let us note that the results obtained by other researchers are not suitable for this area, since each area has individual climatic, geological and social features.

Short-Term Load Forecasting for CCHP Systems Considering the Correlation between Heating, Gas and Electrical Loads Based on Deep Learning

Combined cooling, heating, and power (CCHP) systems is a distributed energy system that uses the power station or heat engine to generate electricity and useful heat simultaneously. Due to its wide range of advantages including efficiency, ecological, and financial, the CCHP will be the main direction of the integrated system. The accurate prediction of heating, gas, and electrical loads plays an essential role in energy management in CCHP systems. This paper combined long short-term memory (LSTM) network and convolutional neural network (CNN) to design a novel hybrid neural network for short-term loads forecasting considering their correlation. Pearson correlation coefficient will be utilized to measure the temporal correlation between current load and historical loads, and analyze the coupling between heating, gas and electrical loads. The dropout technique is proposed to solve the over-fitting of the network due to the lack of data diversity and network parameter redundancy. The case study shows that considering the coupling between heating, gas and electrical loads can effectively improve the forecasting accuracy, the performance of the proposed approach is better than that of the traditional methods.

Research on Short-Term Load Prediction Based on Seq2seq Model

Electricity load prediction is the primary basis on which power-related departments to make logical and effective generation plans and scientific scheduling plans for the most effective power utilization. The perpetual evolution of deep learning has recommended advanced and innovative concepts for short-term load prediction. Taking into consideration the time and nonlinear characteristics of power system load data and further considering the impact of historical and future information on the current state, this paper proposes a Seq2seq short-term load prediction model based on a long short-term memory network (LSTM). Firstly, the periodic fluctuation characteristics of users’ load data are analyzed, establishing a correlation of the load data so as to determine the model’s order in the time series. Secondly, the specifications of the Seq2seq model are given preference and a coalescence of the Residual mechanism (Residual) and the two Attention mechanisms (Attention) is developed. Then, comparing the predictive performance of the model under different types of Attention mechanism, this paper finally adopts the Seq2seq short-term load prediction model of Residual LSTM and the Bahdanau Attention mechanism. Eventually, the prediction model obtains better results when merging the actual power system load data of a certain place. In order to validate the developed model, the Seq2seq was compared with recurrent neural network (RNN), LSTM, and gated recurrent unit (GRU) algorithms. Last but not least, the performance indices were calculated. when training and testing the model with power system load data, it was noted that the root mean square error (RMSE) of Seq2seq was decreased by 6.61%, 16.95%, and 7.80% compared with RNN, LSTM, and GRU, respectively. In addition, a supplementary case study was carried out using data for a small power system considering different weather conditions and user behaviors in order to confirm the applicability and stability of the proposed model. The Seq2seq model for short-term load prediction can be reported to demonstrate superiority in all areas, exhibiting better prediction and stable performance.

APPLICATION OF THE MAIN COMPONENT METHOD FOR IDENTIFICATION OF MAJOR FACTORS IN INFLUENCE ON EQUITY OF ELECTRICITY

The purpose and objectives of the study are to reduce the information that affects the formation of the regime of electricity consumption of energy market objects by using the method of the main components and identifying common general features. The article describes the application of the method of the main components to determine the factors that affect the most on the magnitude of the electric load. The results of the article can be used to determine the main factors for forming a model for analysis and prediction of electrical loads of energy market objects.

Modelling and forecasting PV production in the absence of behind‐the‐meter measurements

AbstractThis paper deals with hourly day‐ahead prediction of the net electricity load at nine photovoltaic installations in a Swedish regional electricity network. The objective of the study was to develop, test, and evaluate a set of methods to predict the contribution of PV power to the grid without knowing the production and consumption “behind‐the‐meter.” An indirect and a direct approach for prediction of the net load were evaluated. For the indirect approach, a model of the gross production was first estimated based on the open‐source software PVLIB. The model was then used to predict the net load given a forecast of the gross consumption. Since we lacked a model of the latter, we used a “perfect forecast,” in terms of measured gross consumption, to estimate the performance of this approach. In the direct approach, a model of the net load was estimated using either linear regression or an artificial neural network. Here, the model was used for prediction of the net load without any information about the gross consumption. Both approaches rely on information from a numerical weather prediction model together with net load measurements from the previous day. Forecasts using the indirect approach with perfect information about the gross consumption resulted in a normalized (with installed nominal power) RMSEn of 11%. The direct approach with the artificial neural network also resulted in an RMSEn of 11%, even though it did not have any information from behind the meter. Linear regression had an RMSEn of 12%.

GCC electrical long-term peak load forecasting modeling using ANFIS and MLR methods

It is important to develop a suitable model to calculate electricity demand forecasting requested by decision makers. The present study deals with the electrical long-term peak load demand forecasting using a developed Adaptive Neuro-Fuzzy Inference System (ANFIS) and Multiple Linear Regression (MLR) methods. The MLR model is formulated as a function of population and Gross Domestic Product (GDP) for the Gulf Cooperation Council (GCC) region. The Neuro-Fuzzy is thereafter trained using previous sets of data. This training gives a future annual electricity load prediction. The results obtained from the developed models have an acceptable level of mean errors. In general, the GCC region has high-energy consumption influenced by a number of factors, such as population and GDP. The annual variation of both population and GDP growth scenarios is based on development in the country. The obtained results will encourage the GCC through the energy field development and setting the future plans for it. The novelty of the present study is to avoid an increase in generation capacity in mid-term and long-term plans, which will help the GCC countries to avoid load shedding and meet the energy demands in different sectors. The developed models will help the economic development of the GCC countries. It also helps with finding the optimum time for electrical energy trading. The obtained results for the GCC illustrate the average percentage error calculated which was found to be close to 2% and 0.53% in multilinear regression and Neuro-Fuzzy, respectively. These results reduce capital investment, limiting the equipment installed and the expected load needed for better load distribution in the region. In conclusion, the Neuro-Fuzzy is the most accurate technique compared with MLR to estimate future electricity demand and, at the same time, it can be used in power system planning and development.

A new hybrid model to predict the electrical load in five states of Australia

Short-term electrical load forecasting is an important part in the management of electrical power because electrical load is an extreme, complex non-linear system. To obtain parameter values that provide better performances with high precision, this paper proposes a new hybrid electrical load forecasting model, which combines ensemble empirical mode decomposition, extreme learning machine, and grasshopper optimization algorithm for short-term load forecasting. The most important difference that distinguishes this electrical load forecasting model from other models is that grasshopper optimization can search suitable parameters (weight values and threshold values) of extreme learning machine, while traditional parameters are selected randomly. It is applied in Australia electrical load prediction to show its superiority and applicability. The simulation studies are carried out using a data set collected from five main states (New South Wales, Queensland, Tasmania, South Australia and Victoria) in Australia from February 1 to February 27, 2018. Compared with all considered basic models, the proposed hybrid model has the best performance in predicting electrical load.