Power Consumption Time Series Research Articles

Comparative analysis of power consumption time series in deprived and developed regions of Iran

This paper presents a comparative analysis of power consumption time series at 12 o'clock every day between 2020 and 2022 for one distribution network in Sistan and one in Tehran. The aim of this study is to compare the development and climate differences between these regions, as well as the impact of social, industrial, and environmental factors. By comparing a deprived area with an area in the capital, we aim to identify potential disparities in power consumption and identify potential areas for improvement. We employed the CRP tool software and toolkit for time series analysis and used various methods to compare and predict the predictability of each time series. Our findings suggest significant differences in power consumption between the two regions, which could be attributed to socio-economic and environmental factors. Overall, this study sheds light on the potential impact of regional differences on power consumption and highlights the need for further research in this area.

Optimized hybrid ensemble learning approaches applied to very short-term load forecasting

The significance of accurate short-term load forecasting (STLF) for modern power systems’ efficient and secure operation is paramount. This task is intricate due to cyclicity, non-stationarity, seasonality, and nonlinear power consumption time series data characteristics. The rise of data accessibility in the power industry has paved the way for machine learning (ML) models, which show the potential to enhance STLF accuracy. This paper presents a novel hybrid ML model combining Gradient Boosting Regressor (GBR), Extreme Gradient Boosting (XGBoost), k-Nearest Neighbors (kNN), and Support Vector Regression (SVR), examining both standalone and integrated, coupled with signal decomposition techniques like STL, EMD, EEMD, CEEMDAN, and EWT. Through Automated Machine Learning (AutoML), these models are integrated and their hyperparameters optimized, predicting each load signal component using data from two sources: The National Operator of Electric System (ONS) and the Independent System Operators New England (ISO-NE), boosting prediction capacity. For the 2019 ONS dataset, combining EWT and XGBoost yielded the best results for very short-term load forecasting (VSTLF) with an RMSE of 1,931.8 MW, MAE of 1,564.9 MW, and MAPE of 2.54%. These findings highlight the necessity for diverse approaches to each VSTLF problem, emphasizing the adaptability and strength of ML models combined with signal decomposition techniques.

Разработка гибридной модели прогнозирования потребления электрической энергии для горно-металлургического предприятия

Introduction. Reducing the energy intensity of production is one of the directions of development of enterprises, regions, states, as it affects the level of economic development and environmental well-being. One of the methods that contribute to the rationalization of the use of electric energy is forecasting. The purpose of the study is to develop a model for forecasting electricity consumption, which allows obtaining a reliable forecast of electricity consumption. To achieve this goal, an analysis of literary sources reflecting the current trends of predictive analytics in solving the problem of forecasting electricity consumption was carried out. Materials and methods of research. Singular spectral analysis and deep learning algorithms (multilayer perceptron) were used in the work. At the initial stage of the study, a model of a fully connected neural network (multilayer perceptron) was built, the prediction accuracy of which was more than 94%. Then the algorithm of singular spectral analysis was implemented, and the initial time series of power consumption was decomposed into additive components. The resulting decompositions were used to create a hybrid forecasting model. Research results. A number of experimental studies have been conducted that meet the goal of developing models that most accurately approximate experimental data. The result of the study are two forecasting models: an artificial neural network model and a hybrid model that includes a singular spectral analysis to decompose a number of power consumption into trend and harmonic components and an artificial neural network to predict them. Discussion. With the help of the hybrid model, an accurate result was obtained – the average absolute percentage error for the year is 4.04%. Forecasting was carried out for a year (by months) using data from the previous two years as a training subset. The neural network model provides less accuracy in predicting power consumption over 12 months compared to the hybrid model. Conclusion. Thus, the use of singular spectral analysis made it possible to reduce the magnitude of the forecast error by almost 1%, which confirms the effectiveness of the method and the relevance of the development of hybrid forecasting models. Resume. As a result of the conducted research, it was revealed that one of the most popular and effective forecasting methods is singular spectral analysis and the use of various deep learning algorithms. The developed hybrid model of electric energy consumption for a mining and metallurgical enterprise allows you to build a forecast of electricity consumption, the accuracy of which is more than 95%. Suggestions for practical application and directions for future research. The research results can be useful in the process of production management, since an accurate forecast of electricity consumption is important when planning production, scheduling equipment repairs and maintenance of the electrical network. The prospect of further research is to increase the forecasting horizon, the volume of the initial sample and taking into account additional factors characterizing the technological process.

Online Fault Detection in ReRAM-Based Computing Systems for Inferencing

A resistive switching random access memory (ReRAM)-based computing system (RCS) provides an energy-efficient hardware implementation of vector–matrix multiplication for machine-learning hardware. However, it is susceptible to faults due to the immature resistive switching random access memory (ReRAM) fabrication process. We propose an efficient online fault-detection method for RCS. This method monitors the dynamic power consumption of each ReRAM crossbar and determines the occurrence of faults when a changepoint is detected in the monitored power-consumption time series. To estimate the percentage of faulty cells in a faulty ReRAM crossbar, we compute statistical features immediately before and after the changepoint and use them as independent variables; we use the percentage of faulty cells as dependent variables to train a predictive model using machine learning. In this way, the computationally expensive fault localization and error-recovery steps are carried out only when a high fault rate is estimated. Simulation results show that, with the fault-detection method and the predictive model, the test time is significantly reduced, the hardware overhead is negligible, and high classification accuracy for the MNIST and CIFAR-10 datasets using RCS can still be ensured.

A Power Consumption Model for Cloud Servers Based on Elman Neural Network

Leveraging power consumption models in software systems can achieve easy deployment of low-cost, high-availability power monitoring in cloud datacenters that are usually large-scale, heterogeneous and frequently scaling up. However, traditional regression-based power consumption models generally have two drawbacks. First, their mathematical forms are usually fixed and determined a priori. This may cause unacceptable increase of error or over-fitting as the power signatures of cloud servers are usually uncertain. Second, the characteristic of workload dispatched to cloud servers is constantly changing while regression-based models can hardly generalize to a wide range of servers and workload types. As a novel solution, we in this paper propose a server power consumption model based on Elman Neural Network (PCM-ENN), aiming to allow accurate and flexible power estimation. PCM-ENN is an end-to-end black box model capable of learning the temporal relation between samples in a time series of power consumption. We trained and evaluated PCM-ENN on two power sequence datasets collected from heterogeneous hardware and operating systems running quasi-production benchmarks like CloudSuite. Experimental result shows that PCM-ENN generated accurate estimates on server power consumption with only small errors, outperforming widely-used linear regression model and NARX model in terms of accuracy.

NEURAL NETWORK FORECASTING OF ENERGY CONSUMPTION OF A METALLURGICAL ENTERPRISE

The subject of the research is the methods of constructing and training neural networks as a nonlinear modeling apparatus for solving the problem of predicting the energy consumption of metallurgical enterprises. The purpose of this work is to develop a model for forecasting the consumption of the power system of a metallurgical enterprise and its experimental testing on the data available for research of PJSC "Dneprospetsstal". The following tasks have been solved: analysis of the time series of power consumption; building a model with the help of which data on electricity consumption for a historical period is processed; building the most accurate forecast of the actual amount of electricity for the day ahead; assessment of the forecast quality. Methods used: time series analysis, neural network modeling, short-term forecasting of energy consumption in the metallurgical industry. The results obtained: to develop a model for predicting the energy consumption of a metallurgical enterprise based on artificial neural networks, the MATLAB complex with the Neural Network Toolbox was chosen. When conducting experiments, based on the available statistical data of a metallurgical enterprise, a selection of architectures and algorithms for learning neural networks was carried out. The best results were shown by the feedforward and backpropagation network, architecture with nonlinear autoregressive and learning algorithms: Levenberg-Marquard nonlinear optimization, Bayesian Regularization method and conjugate gradient method. Another approach, deep learning, is also considered, namely the neural network with long short-term memory LSTM and the adam learning algorithm. Such a deep neural network allows you to process large amounts of input information in a short time and build dependencies with uninformative input information. The LSTM network turned out to be the most effective among the considered neural networks, for which the indicator of the maximum prediction error had the minimum value. Conclusions: analysis of forecasting results using the developed models showed that the chosen approach with experimentally selected architectures and learning algorithms meets the necessary requirements for forecast accuracy when developing a forecasting model based on artificial neural networks. The use of models will allow automating high-precision operational hourly forecasting of energy consumption in market conditions.
 Keywords: energy consumption; forecasting; artificial neural network; time series.

Real-time day ahead energy management for smart home using machine learning algorithm

Smart grid is a sophisticated and smart electrical power transmission and distribution network, and it uses advanced information, interaction and control technologies to build up the economy, effectiveness, efficiency and grid security. The accuracy of day-to-day power consumption forecasting models has an important impact on several decisions making, such as fuel purchase scheduling, system security assessment, economic capacity generation scheduling and energy transaction planning. The techniques used for improving the load forecasting accuracy differ in the mathematical formulation as well as the features used in each formulation. Power utilization of the housing sector is an essential component of the overall electricity demand. An accurate forecast of energy consumption in the housing sector is quite relevant in this context. The recent adoption of smart meters makes it easier to access electricity readings at very precise resolutions; this source of available data can, therefore, be used to build predictive models., In this study, the authors have proposed Prophet Forecasting Model (PFM) for the application of forecasting day-ahead power consumption in association with the real-time power consumption time series dataset of a single house connected with smart grid near Paris, France. PFM is a special type of Generalized Additive Model. In this method, the time series power consumption dataset has three components, such as Trend, Seasonal and Holidays. Trend component was modelled by a saturating growth model and a piecewise linear model. Multi seasonal periods and Holidays were modelled with Fourier series. The Power consumption forecasting was done with Autoregressive Integrated Moving Average (ARIMA), Long Short Term Neural Memory Network (LSTM) and PFM. As per the comparison, the improved RMSE, MSE, MAE and RMSLE values of PFM were 0.2395, 0.0574, 0.1848 and 0.2395 respectively. From the comparison results of this study, the proposed method claims that the PFM is better than the other two models in prediction, and the LSTM is in the next position with less error.

Forecasting of daily power consumption dynamics

The purpose of the study is to simulate the time series of power consumption in a power grid segment. It gives us be able to forecast the future state from values of the series according to current and past estimates. We want to test two different approaches: Box-Jenkins method and recurrent neural networks, for detect which of them give more accuracy and speed of prediction calculation. This article discusses the using Box-Jenkins method.

Machine Learning-Enabled Distribution Network Phase Identification

The distribution network has typically been the least observable and most dynamic and locally controlled element in the power grid. Complete information about the network topology is continuously changing and is not always readily available when needed. This makes the phase identification and network rebalancing a hard, costly, and time-consuming task for electric utilities, however, it is of great importance to future grid planning and advanced distribution management system (ADMS) type operation. Phase identification traditionally is executed manually, although there are existing voltage measurement based methods that are not always reliable. This paper develops a machine learning based data mining method for an accurate and efficient phase identification of residential customers in a distribution network by leveraging power consumption data collected through the advanced metering infrastructure (AMI). The proposed method uses a high-pass filter to remove the redundant and irrelevant parts of the power consumption time series, then identifies the residential customers' phase connectivity by proposing a modified clustering algorithm. Simulation results show the effectiveness of the proposed method in phase identification in both small and large networks and under complete and incomplete data scenarios.

AN IMPROVED PRE-FORECASTING ANALYSIS OF ELECTRICAL LOADS OF PUMPING STATION

Relevance of research. In order to reduce energy losses, an accurate and timely forecast of the amount of consumed electricity is necessary. Accurate forecasting of electrical loads of industrial enterprises and their divisions (productions, workshops, departments etc.) allows planning of normal operating conditions, concluding contracts for the electricity supply with the electricity supply company under more favorable conditions, and improving the electricity quality, which ultimately affects the final cost of the products produced by an enterprise. So far, more than 150 forecasting methods of electrical loads have been developed. Usually, the most convenient one is selected based on the forecaster experience by creating and analyzing several forecasting models in order to identify the best. Therefore, in order to simplify the forecasting procedure, it is necessary to develop the methodology for forecasting analysis. This methodology should enable canceling forecasting algorithms that will create lower quality forecasts. The main objective is to develop the methodology for making a forecasting analysis of power consumption on the example of a pumping station of an enterprise with a continuous cycle of work to increase the efficiency of energy consumption and implementation of energy-saving measures. Objects of research: the process of forecasting electrical loads of a pumping station of the enterprise with a continuous cycle of work. Methods of research: fundamental principles of the theory of electrical engineering, statistical methods for power consumption forecasting, the method for detecting the trend of radio signals, and fractal analysis of time series. Research results. The methodology for forecasting analysis of power consumption, which makes it possible to apply the most appropriate methods to forecast the operational power consumption, is developed. For the first time, the radio signal trend detection method is applied to identify the trend of electrical loads. The variation ranges of the fractal parameters of time series of electrical loads are established depending on the variation coefficient of the time series for different periods of time. The Brown method of exponential smoothing that is used to forecast the electrical loads, in the case of identifying the smoothing constant α is in the beyond set ( ), is further improved. The regularity of changes in the fractal parameters of time series of power consumption of a pumping station with an increase in the time series duration and their field of application are explained.

AN IMPROVED PRE-FORECASTING ANALYSIS OF ELECTRICAL LOADS OF PUMPING STATION

Relevance of research. In order to reduce energy losses, an accurate and timely forecast of the amount of consumed electricity is necessary. Accurate forecasting of electrical loads of industrial enterprises and their divisions (productions, workshops, departments etc.) allows planning of normal operating conditions, concluding contracts for the electricity supply with the electricity supply company under more favorable conditions, and improving the electricity quality, which ultimately affects the final cost of the products produced by an enterprise. So far, more than 150 forecasting methods of electrical loads have been developed. Usually, the most convenient one is selected based on the forecaster experience by creating and analyzing several forecasting models in order to identify the best. Therefore, in order to simplify the forecasting procedure, it is necessary to develop the methodology for forecasting analysis. This methodology should enable canceling forecasting algorithms that will create lower quality forecasts. The main objective is to develop the methodology for making a forecasting analysis of power consumption on the example of a pumping station of an enterprise with a continuous cycle of work to increase the efficiency of energy consumption and implementation of energy-saving measures. Objects of research: the process of forecasting electrical loads of a pumping station of the enterprise with a continuous cycle of work. Methods of research: fundamental principles of the theory of electrical engineering, statistical methods for power consumption forecasting, the method for detecting the trend of radio signals, and fractal analysis of time series. Research results. The methodology for forecasting analysis of power consumption, which makes it possible to apply the most appropriate methods to forecast the operational power consumption, is developed. For the first time, the radio signal trend detection method is applied to identify the trend of electrical loads. The variation ranges of the fractal parameters of time series of electrical loads are established depending on the variation coefficient of the time series for different periods of time. The Brown method of exponential smoothing that is used to forecast the electrical loads, in the case of identifying the smoothing constant α is in the beyond set ( ), is further improved. The regularity of changes in the fractal parameters of time series of power consumption of a pumping station with an increase in the time series duration and their field of application are explained.

Monthly Electricity Consumption Forecasting Method Based on X12 and STL Decomposition Model in an Integrated Energy System

With the rapid development and wide application of distributed generation technology and new energy trading methods, the integrated energy system has developed rapidly in Europe in recent years and has become the focus of new strategic competition and cooperation among countries. As a key technology and decision-making approach for operation, optimization, and control of integrated energy systems, power consumption prediction faces new challenges. The user-side power demand and load characteristics change due to the influence of distributed energy. At the same time, in the open retail market of electricity sales, the forecast of electricity consumption faces the power demand of small-scale users, which is more easily disturbed by random factors than by a traditional load forecast. Therefore, this study proposes a model based on X12 and Seasonal and Trend decomposition using Loess (STL) decomposition of monthly electricity consumption forecasting methods. The first use of the STL model according to the properties of electricity each month is its power consumption time series decomposition individuation. It influences the factorization of monthly electricity consumption into season, trend, and random components. Then, the change in the characteristics of the three components over time is considered. Finally, the appropriate model is selected to predict the components in the reconfiguration of the monthly electricity consumption forecast. A forecasting program is developed based on R language and MATLAB, and a case study is conducted on the power consumption data of a university campus containing distributed energy. Results show that the proposed method is reasonable and effective.

Poster abstract: state of operation recognition for heat pumps from smart grid monitoring data

Coordinated (de)activation of heat pumps in residential heating systems can support the balance between energy production and consumption in a smart grid. Stopping the heat pump in an inappropriate operation state may result in users’ discomfort or damages to the system. In this contribution, a procedure to split domestic hot water provision (DHW) cycles from space heating cycles in the power consumption time series of heat pumps in residential heating systems is presented. The procedure bases on a support vector machine (SVM) classifier applied on statistical properties of the cycles after principal component transformation. The procedure is tested on real-world power consumption time series with 1 s resolution from three different buildings monitored over 1 year. Due to the absence of ground truth validation data, training data for the SVM algorithm are distilled from the measurement data by outlier removal and subsequent K-means classification. The dependence of energy demand for space heating (heating curve) and DHW provision on the ambient temperature are investigated for validation and agrees with Swiss building standards.

Continuous Wavelet and Hilbert-Huang Transforms Applied for Analysis of Active and Reactive Power Consumption

Abstract Analysis of power consumption presents a very important issue for power distribution system operators. Some power system processes such as planning, demand forecasting, development, etc.., require a complete understanding of behaviour of power consumption for observed area, which requires appropriate techniques for analysis of available data. In this paper, two different time-frequency techniques are applied for analysis of hourly values of active and reactive power consumption from one real power distribution transformer substation in urban part of Sarajevo city. Using the continuous wavelet transform (CWT) with wavelet power spectrum and global wavelet spectrum some properties of analysed time series are determined. Then, empirical mode decomposition (EMD) and Hilbert-Huang Transform (HHT) are applied for the analyses of the same time series and the results showed that both applied approaches can provide very useful information about the behaviour of power consumption for observed time interval and different period (frequency) bands. Also it can be noticed that the results obtained by global wavelet spectrum and marginal Hilbert spectrum are very similar, thus confirming that both approaches could be used for identification of main properties of active and reactive power consumption time series.

Anomaly detection for visual analytics of power consumption data

Commercial buildings are significant consumers of electrical power. Also, energy expenses are an increasing cost factor. Many companies therefore want to save money and reduce their power usage. Building administrators have to first understand the power consumption behavior, before they can devise strategies to save energy. Second, sudden unexpected changes in power consumption may hint at device failures of critical technical infrastructure. The goal of our research is to enable the analyst to understand the power consumption behavior and to be aware of unexpected power consumption values. In this paper, we introduce a novel unsupervised anomaly detection algorithm and visualize the resulting anomaly scores to guide the analyst to important time points. Different possibilities for visualizing the power usage time series are presented, combined with a discussion of the design choices to encode the anomaly values. Our methods are applied to real-world time series of power consumption, logged in a hierarchical sensor network.

A Visual Analytics Approach for Peak‐Preserving Prediction of Large Seasonal Time Series

AbstractTime series prediction methods are used on a daily basis by analysts for making important decisions. Most of these methods use some variant of moving averages to reduce the number of data points before prediction. However, to reach a good prediction in certain applications (e.g., power consumption time series in data centers) it is important to preserve peaks and their patterns. In this paper, we introduce automated peak‐preserving smoothing and prediction algorithms, enabling a reliable long term prediction for seasonal data, and combine them with an advanced visual interface: (1) using high resolution cell‐based time series to explore seasonal patterns, (2) adding new visual interaction techniques (multi‐scaling, slider, and brushing & linking) to incorporate human expert knowledge, and (3) providing both new visual accuracy color indicators for validating the predicted results and certainty bands communicating the uncertainty of the prediction. We have integrated these techniques into a well‐fitted solution to support the prediction process, and applied and evaluated the approach to predict both power consumption and server utilization in data centers with 70–80% accuracy.

Fractal Impeller for Stirred Tank Reactors

Stirred tank reactors are used for variety of applications at different scales of operation. The conventional impellers tend to develop regions having nonuniform energy dissipation rates in the stirred reactor. In this work, we propose a novel fractal impeller, which helps in reducing such nonuniformities and help develop a uniform randomness throughout the reactor. The impeller geometry is discussed in detail. Experimental measurements of the power consumption, mixing time, suspension quality, and the ability for gas dispersion were carried out, and the performance is compared with the conventional impellers. The impeller is seen to have a low power number, and it can generate a uniform suspension of particles even at relatively lower impeller speeds and can efficiently disperse gas into liquid to yield relatively higher gas hold-up values. The Fourier analysis of the power consumption time series data indicates that no specific prominent frequency events exist in the reactor, and the spectrum showed several frequency events to exist in the reactor with almost identical prominence.

Clustering time series, subspace identification and cepstral distances

In this paper a methodology to cluster time series based on measurement data is described. In particular, we propose a distance for stochastic models based on the concept of sub- space angles within a model and between two models. This distance is used to obtain a clustering over the set of time series. We show how it is related to the mutual information of the past and the future output processes, and to a previously defined cepstral distance. Finally, the methodol- ogy is applied to the clustering of time series of power consumption within the Belgian electricity grid.

Tutorial and Survey

ABSTRACTThis paper illustrates the use of the Box‐Jenkins methodology by analyzing the Ohio Electrical Power Consumption time series. The three basic analytical steps discussed are the following:1) Model identification ‐ by matching sample autocorrelation functions against theoretical autocorrelation functions.2) Non‐linear estimation of parameters ‐ by minimizing the sum of the squared residuals.3) Diagnostic checking ‐ by analyzing the pattern of the autocorrelation function of the residuals.Regular and adaptive forecasts are then developed using the appropriate model that emerges from the time series analysis.