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

In order to save unnecessary energy losses, power companies need to accurately predict electricity consumption. Various classification methods are used for electricity consumption forecasting: traditional time series models, deep neural network models based on Artificial Intelligence, and hybrid models. Most scholars use a single view to forecast electricity consumption. This article predicts electricity consumption from three views: traditional models, deep neural network models, and hybrid models. First, the electricity consumption time series is preprocessed. Then, from three views, models that have achieved good performance in some time series are selected, and then the forecast analysis of the model is given. Finally, the performance of different models in electricity consumption forecasting is evaluated and compared. The results show that the expected performance of the hybrid model is not as good as the traditional model and the deep neural network model in electricity consumption forecasting.

The forecast of electricity consumption is a prediction of the use of electricity for the future by referring to the use of electricity in past data. Estimates of electricity consumption needs are intended to estimate how much electricity consumption is used by customers and must be provided by PT PLN (PERSERO) ULP SIBUHUAN as a provider of electrical energy services. Population growth occurs along with the development of an area, so it affects the demand for electricity and the need for electricity consumption. Load mapping must be done to maintain the continuity and distribution of electrical energy to customers. One way to preserve the continuity of service is to estimate electric power consumption using several forecasting methods. One of the methods that can be used is to combine several load forecasting methods called the combined method. The combined method is a model that incorporates various estimation methods, including econometrics, analytics, and trends, using historical customer data originating from BPS (Central Statistics Agency) Padang Lawas Regency and PT. PLN (PERSERO) ULP SIBUHUAN. The estimated load between 2021 and 2025 is 625,070,452 kWh, with an increase in electricity consumption of 20.7%, household expenses increasing by 1.8%, and business sector expenses by 1.35%. The method used is to calculate manually using Microsoft Excel to obtain forecasting values for population growth; increasing electricity consumption is very important for planning electric power generation, development, and planning of electric power distribution and mapping of loads on customers.

The large airport terminal is the city's transportation hub and logistics center, and its building electricity consumption is twice or even higher than that of general large public buildings. Energy-saving and consumption reduction of the terminal building is not only the need to improve the operational efficiency and reduce the operation cost of the airport but also the need to realize the goal of "double carbon" in China. In this study, a large airport terminal in Beijing is taken as the research object. According to the characteristics of its main electrical equipment system, and the measured data of passenger flow from January 2020 to February 2023, combined with Pearson correlation analysis, K-means cluster analysis and multiple regression analysis, the influence and correlation characteristics of building scale, passenger flow and outdoor meteorological parameters on its electricity consumption are analyzed. The results show that: 1) The daily electricity consumption of the terminal can be analyzed by five levels according to the daily passenger flow, namely, 10,000 < N ≤ 30,000,30,000 < N ≤ 50,000, 50,000 < N ≤ 70,000, 70,000 < N ≤ 100,000, and 100,000 < N ≤ 130,000; 2) The electricity consumption of the terminal can be divided into three categories: basic electricity consumption related to the building scale of the terminal, variable electricity consumption I related to passenger flow, and variable electricity consumption II related to outdoor air temperature and passenger flow. Based on this, the terminal's daily electricity consumption prediction model is constructed, and the measured data verify the model's effectiveness. 3) Put forward the evaluation reference value of the daily electricity consumption of different passenger flow in this terminal, When the year passenger flow reaches the design value of 45 million, the normal daily electricity consumption level ranges from 0.44 to 0.48 kW·h/(m2·day). The research results can provide methods and evaluation reference for electricity consumption prediction, and accurate control of electricity consumption of main equipment systems in airport terminal.

This paper proposes an adaptive fuzzy expert system to concurrently estimate and forecast both long-term electricity and natural gas (NG) consumptions with hike in prices. Using a novel procedure, the impact of price hike is incorporated into energy demand modeling. Furthermore, adaptive network-based FIS (ANFIS) is used to model NG consumption in power generation (NGPG). To cope with random uncertainty in small historical data sets, Monte Carlo simulation is used to generate training data for ANFIS. The proposed ANFIS uses electricity consumption data to improve the estimation of total NG consumption. The unique contribution of this paper is three fold. First, it proposes a novel expert system for electricity consumption and NG consumption in end-use sector with hike in prices. Second, it uses ANFIS-Monte Carlo approach for NGPG. Third, electricity consumption is used in ANFIS for improvement of NGPG consumption estimation. A real case study is presented that illustrates the applicability and usefulness of the proposed model where it is applied for joint forecasting of annual electricity and NG consumption with hike in prices.

Electricity consumption prediction is an important part of power planning and the basis of power dispatch planning. The SVM and GBDT algorithms that were optimized by the PSO algorithm are used to build the machine learning-based electricity consumption prediction model in this paper. The decomposed and reconstructed high-frequency signals and low-frequency signals are optimized by the particle swarm algorithms SVM and GBDT for power consumption prediction, respectively. Improve the clustering performance of the traditional fuzzy C-mean algorithm for unbalanced data. The performance of the algorithms is analyzed in two different application scenarios, namely, artificial dataset and power users' real measurement dataset. Power consumption prediction in the manufacturing industry is performed. It is found that the ARI, FMI, and AMI index values of the improved algorithms proposed in this paper are 0.9543, 0.9347, and 0.9344, respectively, on the grid user-measured dataset, while the traditional DPC and K-means clustering algorithms are less effective. The machine learning algorithm optimized after wavelet decomposition increased R² by 8.98%, MAPE decreased by 19.78%, and RMSE decreased by 11.53% compared to PSO-GBDT, and the predictive evaluation indexes were all improved, and the two machine learning algorithms in this paper optimized by wavelet decomposition combined with PSO have good predictive effect R² increased from 0.801 to 0.842, and the two machine learning algorithms designed in this paper based on wavelet decomposition have good predictive effect R² increased from 0.801 to 0.842. The designed machine learning model for electricity consumption prediction based on wavelet decomposition and PSO has excellent performance, and the design expectation has been fulfilled. This paper makes a useful exploration and proposes an effective method for accurate prediction of electricity consumption in the manufacturing industry.

At present, the electricity market reform has entered a deep area, electricity consumption forecasting has become increasingly important, accurate electricity consumption forecasting provides a reference basis and decision-making support for power dispatching and market transactions, and residential power consumption prediction can help users choose appropriate power suppliers and power supply programs according to the market situation, and provide the reliability and economy of power consumption. Residential electricity consumption is complex, affected by many factors and prone to significant noise disturbances, which results in electricity consumption data that is characterized by non-stationary, intermittent and erratic fluctuations. Therefore, using a single model is challenging to accurately predict household electricity consumption. To meet this challenge, this paper designs the structure of this three-step residential electricity consumption forecasting. At the first stage, we first analyzed cumulative effects of temperature on residential electricity consumption, and an hourly temperature correction model combining meteorological factors is constructed, the adjusted hourly temperature data are then entered into the predictive model. We have developed a Variable Modal Decomposition (VMD) data decomposition technique optimized for non-governmental organizational models, which improves the problem of subjectivity in parameter setting in traditional VMD, thus enhancing the performance and accuracy in data decomposition. In the second stage, developed a BiLSTM-AM based integrated deep learning model and dynamically adjusted the weights of the influencing factors by introducing an Attention Mechanism (AM) to enhance the stability of the model, also predict multiple IMF components obtained after the NGO-VMD decomposition, respectively. In the third stage, the training residuals of the BiLSTM-AM model are used as target variables to correct the prediction error in BiLSTM-AM using the XGBoost regression model. A variety of model configurations were constructed using actual data from a coastal province in southern China, and the computational results show that the integrated prediction model exhibits excellent stability and accuracy.

With the development of intelligence and data construction of electric power system in China, a large amount of data accumulated by electric power enterprises provide a data base for the fine prediction of electric power consumption. High-precision power forecasting model has far-reaching influence on urban planning and construction, smart grid development and so on. The problem of power big data privacy leads to the phenomenon of Power Data Island, which leads to the shortage of the accuracy of the current power forecasting methods. Federal learning breaks the long-standing Data Island phenomenon in power industry, and satisfies the privacy and security requirements of power data in application. This paper presents a high-precision federal learning method for enterprise electricity consumption forecasting, which takes into account weather conditions and enterprise tax information. Based on the FATE platform, this method combines the enterprise electricity consumption data with the tax data, and uses the third party coordinator to conduct encryption training, then constructs the enterprise electricity consumption forecast model based on the SecureBoost algorithm. The simulation results show that the federated learning method can effectively improve the accuracy of enterprise electricity consumption forecasting model.KeywordsFederated learningEnterprise electricity consumption forecastingPower big dataFATE

This paper presents a novel approach to forecast day-ahead electricity consumption for residential households where highly irregular human behaviour plays a significant role. The methodology requires data from fiscal smart meters, which makes it applicable to real scenarios where personal data gathering is not feasible. These data are rarely complete; therefore, a robust combination of machine-learning techniques is used to handle missing data and outliers. The novelty of this method relies on identifying and predicting user electricity consumption behaviour as a procedure to improve the forecasting of the overall electricity consumption of each individual customer. The methodology uses Gaussian mixture clustering to identify behaviour clusters and an eXtreme Gradient Boosting classification (XGBoost) model to predict the day-ahead behaviour pattern. This predicted user behaviour cluster is fed into an Artificial Neural Network (ANN) to enable an improved capturing of the highly unpredictable user conduct for the forecast of electricity consumption. A novel metric, namely the Euclidean Distance-based Accuracy (EDA), is finally proposed to enable a more thorough evaluation of time series classification algorithms. The whole development is tested over 500 residential users placed in a southeastern region of Spain. The results showed that, when the novel approach was used, the MAPEd and NRMSEd were reduced by 7% and 9% respectively, increasing to a 20% and 17% respective reduction for the best cases according to EDA. This methodology sets the basis for massive user-centred analyses, very profitable to any electricity company.

The forecast of electricity consumption has a siginificant impact on both economics and society.Various models have been developed to make this forecast accurate.Among them is the highly successful ARIMA model which is based on time series analysis.However, when applied local in China, ARIMA forecast is not as accurate as expected, due to the moving holiday effect of Chinese New Year feastive period.This paper proposes a revised ARIMA model which adjusts its forecast by the use of a China New Year (CNY) coefficient.Preliminary experimental results on real data show that this revised model outperforms the traditional one.

The forecasting technology of electricity consumption is essential both in power system scheduling and in the open power market. However, in the current electricity sales company's electricity consumption forecasting problems, the historical electricity data and the error distribution of electricity consumption forecasting are usually nonlinear and non-Gaussian. Due to the small user scale, some random events and data collection errors will cause outliers. These outliers has a greater impact on the prediction results, which makes traditional prediction methods no longer applicable. In order to solve this problem, this paper uses generalized maximum correntropy criterion as a new loss function to replace the mean squared error loss function in the traditional ELM, and constructs the GMCCELM electricity prediction model. Finally, the historical electricity consumption data of a commercial user in Guangdong province and the corresponding temperature data are applied to the electricity consumption prediction model established in this paper. An actual case is given to illustrate the superiority and effectiveness of the electricity consumption prediction model compared with other existing methods. The prediction results show that, compared with the traditional ELM and other algorithms, the robust prediction model established in this paper can effectively improve the accuracy of electricity consumption forecasting.

Relevance. The need for accurate forecasting of electricity consumption to improve efficiency and reduce costs at industrial enterprises, which leads to increased competitiveness of goods manufactured by the enterprise. Traditional forecasting methods often do not take into account complex interactions between various factors affecting energy consumption and do not provide the necessary forecast accuracy. The principal component analysis method offers a promising prospect – reducing the volume of processed data (dimensionality) without significant loss of information, which simplifies forecast models while maintaining their accuracy. Aim. To develop an accurate and efficient model for forecasting electricity consumption at industrial enterprises using the principal component analysis method. This model is aimed at eliminating the limitations of traditional forecasting approaches by reducing data dimensionality and increasing the accuracy of predictions, which ultimately improves the efficiency of electricity consumption and reduces financial costs, including those due to forecasting errors. Methods. The principal component analysis method, which allows us to reduce the volume of processed data (dimensionality) by transforming a large set of correlated variables into a smaller set of uncorrelated principal components. The study included the following stages: data import and factor analysis, correlation matrix construction, analysis of selected and accumulated variances for each factor, factor loading matrix construction, dimension reduction, development of a mathematical model using linear regression, and forecast installation and validation. Results. The application of the principal component analysis method allowed us to create a model for forecasting electricity consumption. Its application showed that the first principal component explains 69.65% of the total variance, the second component – ​​17.28%, i. e. together they explain almost 87% of the variance. The developed model provides good agreement between the actual and forecast values ​​of electricity consumption in several time intervals with an average error level within the range of +3 to –5%. This indicates the suitability of the model for forecasting electricity consumption, although some discrepancies indicate the need for its further improvement.

Currently many commercial buildings are not continuously monitored for energy consumption, especially small buildings which constitute 90% of all such buildings. However, readily available data from the electric meters can be used for monitoring and analyzing energy consumption. Efficient utilization of available historical data (from these meters) can potentially improve energy efficiency, help to identify common energy wasting problems, and, in the future, enable various Smart Grid programs, such as demand response, real-time pricing etc. This paper describes application of computational intelligence techniques for prediction of electric power consumption. The proposed approach combines regression and clustering methods, in order to improve the prediction accuracy of power consumption, as a function of time (of the day) and temperature, using real-life data from several commercial and government buildings. Empirical comparisons show that the proposed approach provides an improvement over the currently used bin-based method for modeling power consumption.

Improved-performance day-ahead electricity demand forecast is important to deliver necessary information for right decision of energy management of microgrids. It supports microgrid operators and stakeholders to have better decisions on microgrid flexibility, stability and control. The available conventional forecasting methods for electricity demand at national or regional level are not effective for electricity demand forecasting in microgrids. This is due to the fact that the electricity consumption in microgrids is many times less than the regional or national demands and it is highly volatile. In this paper, an integrated Artificial Intelligence (AI) based approach consisting of Wavelet Transform (WT), Simulated Annealing (SA) and Feedforward Artificial Neural Network (FFANN) is devised for day-ahead prediction of electric power consumption in microgrids. The FFANN is the basic forecasting engine of the proposed model. The WT is utilized to extract relevant features of the target variable (electric load data series) to obtain a cluster of enhanced-feature subseries. The extracted subseries of the past values of the electric load demand data are employed as the target variables to model the FFANN. The SA optimization technique is employed to obtain the optimal values of the FFANN weight parameters during the training process. Historical information of actual electricity consumption, meteorological variables, daily variations, weekly variations, and working/non-working day indicators have been employed to develop the forecasting tool of the devised integrated AI based approach. The approach is validated using electricity demand data of an operational microgrid in Beijing, China. The prediction results are presented for future testing days with one-hour time interval. The validation results demonstrated that the devised approach is capable to forecast the microgrid electricity demand with acceptably small error and reasonably short computation time. Moreover, the prediction performance of the devised approach has been evaluated relative to other four approaches and resulted in better prediction accuracy.

In recent years, with the deepening of China’s electricity sales side reform and electricity market opening up gradually, the forecasting of electricity consumption (FoEC) becomes an extremely important technique for the electricity market. At present, how to forecast the electricity accurately and make an evaluation of results scientifically are still key research topics. In this paper, we propose a novel prediction scheme based on the least-square support vector machine (LSSVM) model with a maximum correntropy criterion (MCC) to forecast the electricity consumption (EC). Firstly, the electricity characteristics of various industries are analyzed to determine the factors that mainly affect the changes in electricity, such as the gross domestic product (GDP), temperature, and so on. Secondly, according to the statistics of the status quo of the small sample data, the LSSVM model is employed as the prediction model. In order to optimize the parameters of the LSSVM model, we further use the local similarity function MCC as the evaluation criterion. Thirdly, we employ the K-fold cross-validation and grid searching methods to improve the learning ability. In the experiments, we have used the EC data of Shaanxi Province in China to evaluate the proposed prediction scheme, and the results show that the proposed prediction scheme outperforms the method based on the traditional LSSVM model.

Accurate monthly electricity consumption forecasting (ECF) is very important for the profitability of electricity retailers in the competitive electricity market. High resolution historical data is rarely used in monthly ECF since it often requires to be combined with multi-step iterative forecasting methods, which will lead to the accumulation of errors and the decrease in forecasting accuracy. This paper proposes a decomposition-accumulation principle-based monthly ECF approach using EMD-XGBoost hybrid model to address the above issues. First, the historical electricity consumption data is divided into seven parts with week label (the day of the week). Then, each part is forecasted separately using the EMD-XGBoost hybrid model. Finally, the monthly ECF result is the accumulation of the seven partial forecasting results. The proposed forecasting approach can reduce the steps of iteration and consider the calendar effects meanwhile the EMD-XGBoost hybrid model can mine data information of various parts in depth without additional influence factors. The case study using real data from PJM demonstrates the effectiveness and superiority of the proposed approach.