Consumption Time Series Data Research Articles

Detecting anomalous electricity consumption with transformer and synthesized anomalies

Non-technical losses are consistently a troubling issue for power suppliers. With the application and popularization of smart grid and advanced measurement systems, it has become possible to use data-driven methods to detect anomalous electricity consumption to reduce non-technical losses. A range of machine learning models have been utilized for detecting anomalous electricity consumption and have achieved promising results. However, with the evolution of techniques like electricity theft, coupled with the exponential increase in electricity consumption data, new challenges are constantly being posed for anomalous electricity consumption detection. We propose a Transformer-based method for detecting anomalous electricity consumption. The Transformer is composed of multi-head attention, layer normalization, point-wise feed-forward network, etc., which can effectively handle electricity consumption time-series data. Meanwhile, to alleviate the problem of imbalanced training data between anomalous and normal electricity consumption, we propose a method for synthesizing anomalies. The experimental results demonstrate that our proposed Transformer-based method outperforms the state-of-the-art methods in detecting anomalous electricity consumption, achieving a precision of 93.9%, a recall of 96.3%, an F1-score of 0.951, and an accuracy of 95.6% on a dataset released by the State Grid Corporation of China.

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.

An outlook analysis on China's natural gas consumption forecast by 2035: Applying a seasonal forecasting method

The proposed targets to achieve peak carbon before 2030 and carbon neutrality before 2060 would require China to accelerate the development of natural gas infrastructure. The forecasted future of natural gas consumption in China will not only help learn about the future usage of natural gas but also guide government departments to conduct more scientific planning and management of natural gas production and pipeline channels. Therefore, this paper fully considers the characteristics of natural gas consumption affected by seasonal factors and uses the 2018–2021 quarterly natural gas consumption time-series data to analyze whether natural gas consumption contains trends and seasonality. It is noticeable that seasons impact China's natural gas consumption, so the moving average method is applied to calculate the seasonal index. Based on this, China's natural gas consumption from 2022 to 2035 is forecasted. As results show, China's gas consumption will continue to grow over the next 14 years but at a slower rate. By 2035, it will double from the 2021 level, reaching 765.4 billion cubic meters, but the growth rate will drop to below 4%, entering a period of low growth. Finally, the paper illustrates current situation of natural gas development combined with rationalized suggestions for future natural gas production and energy security in China.

Prophet-EEMD-LSTM based method for predicting energy consumption in the paint workshop

Energy conservation and preventive maintenance of equipment require the ability to accurately predict future trends in shop floor power consumption to keep track of equipment operation and abnormalities. Due to the non-linearity and uncertainty of the workshop electrical energy consumption data, it is very difficult to establish an accurate energy consumption prediction model. For the above problems, an energy consumption prediction model based on Prophet-EEMD-LSTM was proposed. To identify the period features in the energy consumption time series data, the Prophet algorithm was introduced to identify and extract the period and trend features of the energy consumption data. Then for the feature of the uncertainty of the remaining data, this paper used the ensemble empirical mode decomposition (EEMD) method to decompose the data to form the components with specific modalities. Finally, all the extracted feature data were put into the input layer of the Long short-term memory (LSTM), and the final energy consumption values were predicted using the excellent prediction performance of the LSTM on time-series data. The experimental results showed that the model proposed in this paper has high prediction performance for energy consumption data.

Data-Driven Building Energy Consumption Prediction Model Based on VMD-SA-DBN

Prediction of building energy consumption using mathematical modeling is crucial for improving the efficiency of building energy utilization, assisting in building energy consumption planning and scheduling, and further achieving the goal of energy conservation and emission reduction. In consideration of the non-linear and non-smooth characteristics of building energy consumption time series data, a short-term, hybrid building energy consumption prediction model combining variational mode decomposition (VMD), a simulated annealing (SA) algorithm, and a deep belief network (DBN) is proposed in this study. In the proposed VMD-SA-DBN model, the VMD algorithm decomposes the time series into different modes to reduce the fluctuation of the data. The SA-DBN prediction model is built for each mode separately, and the DBN network structure parameters are optimized by the SA algorithm. The prediction results of each model are aggregated and reconstructed to obtain the final prediction output. The validity and prediction performance of the proposed model is evaluated on a publicly available dataset, and the results show that the proposed new model significantly improves the accuracy and stability of building energy consumption prediction compared with several typical machine learning methods. The mean absolute percent error (MAPE) of the VMD-SA-DBN model is 63.7%, 65.5%, 46.83%, 64.82%, 44.1%, 36.3%, and 28.3% lower than that of the long short-term memory (LSTM), gated recurrent unit (GRU), VMD-LSTM, VMD-GRU, DBN, SA-DBN, and VMD-DBN models, respectively. The results will help managers formulate more-favorable low-energy emission reduction plans and improve building energy efficiency.

Daily Water Demand Prediction Driven by Multi-source Data

The best scheduling of water distribution systems may be supported by accurate and trustworthy water demand forecasts, which is a positive assurance for the development of smart cities and smart water services. This paper studies a new multivariate time-series prediction model that is based on the Convolution Neural Network (CNN) and Gate Recurrent Unit(GRU), taking into account the limitations of the prediction of multivariate time series using a single model. According to the regularity of water use during the time period, the characteristics of water users are clustered and classified to form two types of users: tidal type and irregular type. The factors that affect the daily water consumption such as user type, week, water consumption of the day of the previous three weeks, major events, etc. are used as input vectors, and the 2018-2021 resident daily water consumption time series data of a domestic water company is used as the training sample, respectively establish CNN-GRU models.CNN-GRU approach is checked using the root mean square error(RMSE),mean absolute error(MAE),and mean percentage absolute error(MAPE).The results are compared with Long-Short Term Memory (LSTM),CNN and GRU.Results show that CNN-GRU improves water demand prediction. The CNN-GRU model's RMSE dropped by around 0.648, 0.82, 0.82 when compared to the LSTM, CNN, and GRU models.the MAE decreased by about 0.418,0.47,0.462;and the MAPE decreased by about 0.722,0.649,0.712.

Water Demand Management Based on Water Consumption Data Analysis in the Emirate of Abu Dhabi

In Abu Dhabi, in the United Arab Emirates (UAE), the population and water demand have nearly tripled in the last two decades. Thus, it has become critical to curtail the growing water demand. This study aimed to evaluate the efficacy of Abu Dhabi’s residential water demand management through the installation of water-saving fittings. The analysis of water consumption time-series data revealed that water consumption of the three water user categories was significantly different. Briefly: water-tariff exempt users consumed 95.19 m3/d/connection in 2019, followed by nationals with 5.14 m3/d/connection, and expats at 0.7 m3/d/connection. Nationals began saving water in 2011, which was earlier than the water tariff revision of 2015. In a newly developed area of Khalifa City, the water consumption of new residents was 46.0% less than that of old residents, indicating the effectiveness of water-saving fittings installed in new homes following the Housing Rules’ requirements. Then, based on the estimated number of new connections and the fittings’ saving efficiencies, we estimated that water-saving fittings contributed to 73.1% of the water savings since 2011. These results strongly recommend the introduction of an incentive or subsidy for owners of old houses to replace their outdated fittings with water-saving ones.

Forecast Energy Consumption Time-Series Dataset using Multistep LSTM Models

Smart grid and smart metering technologies allow residential consumers to monitor and control electricity consumption easily. The real-time energy monitoring system is an application of the smart grid technology used to provide users with updates on their home electricity consumption information. This paper aims to forecast a month ahead of daily electricity consumption time-series data for one of the real-time energy monitoring system named beLyfe. Four separate multistep Long Short Term Memory LSTM neural network sequence prediction models such as vanilla LSTM, Bidirectional LSTM, Stacked LSTM, and Convolutional LSTM ConvLSTM has been evaluated to determine the optimal model to achieve this objective. A comparison experiment is performed to evaluate each multistep LSTM model performance in terms of accuracy and robustness. Experiment results show that the ConvLSTM model achieves overall high predictive accuracy and is less computationally expensive during model training than remaining models.

A Data-Driven Soft Sensor to Forecast Energy Consumption in Wastewater Treatment Plants: A Case Study

Energy consumption is vital to the global costs of wastewater treatment plants (WWTPs). With the increase of installed WWTPs worldwide, the modeling and forecast of their energy consumption have become a critical factor in WWTP design to meet environmental and economic requirements. The accurate and swift energy consumption forecasting soft-sensors are not only supportive to the daily electric and financial budgeting by WWTP practitioners on the micro-scale, but also beneficial to local municipal operation and fundamental to regional environmental impact estimation on the macro-scale. Energy consumption in WWTPs is influenced by different biological and environmental factors, making it complicated and challenging to build soft-sensors. This article intends to provide short-term forecasting of WWTP energy consumption based on data-driven soft sensors using traditional time-series and deep learning methods. Ten data-driven soft sensors, including the ordinary least square, exponential smoothing state space, local regression, auto-regressive integrated moving average (ARIMA), structural time series model, Bayesian structural time series, non-linear auto-regressive, long short-term memory with and without updates, and gated recurrent units have been investigated and compared for WWTP energy consumption forecasting. Energy consumption time-series data from a membrane bioreactor-based WWTP in the middle east is used to evaluate the performances of the proposed soft-sensors. Results showed that ARIMA achieved slightly improved performances, among others. The employment of adaptive deep learning-based soft sensors is expected to enhance the capabilities of the deep models to quickly and accurately follow the trend of future data.

An Empirical Analysis of Energy Consumption, FDI and High Quality Development Based on Time Series Data of Zhejiang Province.

This paper aims to identify the relationship among energy consumption, FDI, and economic development in China from 1993 to 2017, taking Zhejiang as an example. FDI is the main factor of the rapid development of Zhejiang’s open economy, which promotes the development of the economy, but also leads to the growth in energy consumption. Based on the time series data of energy consumption, FDI inflow, and GDP in Zhejiang from 1993 to 2017, we choose the vector auto-regression (VAR) model and try to identify the relationship among energy consumption, FDI, and economic development. The results indicate that there is a long-run equilibrium relationship among them. The FDI inflow promotes energy consumption, and the energy consumption promotes FDI inflow in turn. FDI promotes economic growth indirectly through energy consumption. Therefore, improving the quality of FDI and energy efficiency has become an inevitable choice to achieve the transition of Zhejiang’s economy from high speed growth to high quality growth.

Dominant Data Set Selection Algorithms for Electricity Consumption Time-Series Data Analysis Based on Affine Transformation

The explosive growth of time-series data, the scale of time-series data (TSD) suggests that the scale and capability of many Internet of Things (IoT)-based applications has already been exceeded. Moreover, redundancy persists in TSD due to the correlation between information acquired via different sources. In this paper, we propose a cohort of dominant dataset selection algorithms for electricity consumption time-series data with a focus on discriminating the dominant dataset that is small dataset but capable of representing the kernel information carried by time-series data with an arbitrarily small error rate less than ". Furthermore, we prove that the selection problem of the minimum dominant dataset is an NP-complete problem. The affine transformation model is introduced to define the linear correlation relationship between time-series data objects. Our proposed framework consists of the scanning selection algorithm with O(n3) time complexity and the greedy selection algorithm with O(n4) time complexity, which are respectively proposed to select the dominant dataset based on the linear correlation distance between time-series data objects. The proposed algorithms are evaluated on the real electricity consumption data Harbin city in China. The experimental results show that the proposed algorithms not only reduce the size of the extracted kernel dataset but also ensure the time-series data integrity in terms of accuracy and efficiency.

Hurst Index Analysis of Social Electricity Consumption Change Trend Based on R/S Analysis

With the rapid development of China’s electricity industry, the electricity consumption is also changing with a certain trend. Therefore, it is necessary to study the trend of the power consumption statistics for the whole society in China. In this research, according to 2016-2017 electricity consumption data of China Energy journal and Zhejiang electric power journal is analyzed by R/S analysis method. Hurst indexes are estimated to study the self-similarity properties in different times. The results show that the R/S analysis method can effectively analyze the time series data of electricity consumption, and the calculated Hurst index can better reflect the self-similarity trend of the electricity consumption of the whole society in the future.

Short-term Time Series Data Prediction of Power Consumption Based on Deep Neural Network

Under the background of the rapid development of Internet technology and the popularity of smart grids, the analysis and prediction of short-term time series data of users’ power consumption has important guiding significance for grid planning, management decision of economic sector and optimization and allocation of power resources. Considering that the traditional statistical-based time series analysis method is weak in generality and can not handle the complex linear problem in prediction, the long-term dependence of the ordinary cyclic neural network model is insufficient, and the time series data has multidimensional problems, a deep neural network is proposed. The PCA-LSTM model is used for time series data prediction. The model firstly uses the PCA (principal component analysis) method to reduce the dimensionality of the electricity consumption time series data, optimizes the number of input variables, and inputs the data into the long- and short-term memory network LSTM for training prediction. The experimental results show that the LSTM network prediction based on PCA improves the accuracy of short-term time series data prediction, and also improves the convergence speed of LSTM network. It proves that the method has better prediction performance and versatility.

Achieving Food Security in a Climate Change Environment: Considerations for Environmental Kuznets Curve Use in the South African Agricultural Sector

This study relates agricultural income and agricultural carbon dioxide (CO2) emissions in the context of environmental Kuznets curves for South Africa. We posit likely relationships between UN Sustainable Development Goals (SDG) 1, 2 and 13, relating food production to climate change action. CO2 emissions, income, coal energy consumption and electricity energy consumption time series data from 1990 to 2012 within the South African agricultural sector were used. The autoregressive distributive lag bounds-test and the error correction model were used to analyse the data. The results show long-run relationships. However, agricultural income was only significant in the linear and squared models. Changes in agricultural CO2 emissions from the short run towards the long run are estimated at 71.9%, 124.7% and 125.3% every year by the linear, squared and cubic models, respectively. Exponentially increasing agricultural income did not result in a decrease in agricultural CO2 emissions, which is at odds with the Kuznets hypothesis. The study concludes that it will be difficult for South Africa to simultaneously achieve SDGs 1, 2 and 13, especially given that agriculture is reliant upon livestock production, the largest CO2 emitter in the sector. The sector needs to shift to renewable energy consumption with fewer CO2 emissions.

Load and Household Profiles Analysis for Air-Conditioning and Total Electricity in Malaysia

Load profile of household air-conditioning (AC) and total electricity consumption is essential to increase the stability of the energy demand on the grid. Therefore, field measurements on time series data of total and AC electricity consumption from 20 households were conducted from March 2016 to August 2017. The questionnaire survey was carried out simultaneously to grasp the profile of each family. The average total daily and AC consumption were 14.5 kWh/day and 3.9 kWh/day, respectively. The average hourly electricity consumption for total was 0.6 kWh/hour, meanwhile for AC was 0.2 kWh/hour. About 20% of the total peak demand was contributed by the consumption of AC. The indoor air temperature was measured in the bedroom (BR) when AC was switched ON and OFF with an average of 27 ∘C and 29 ∘C, respectively. However, the indoor air temperature in the living room (LR) was 2∘C and 1∘C higher if compared to BR for both conditions. Based on the questionnaire survey, 92% of the occupants preferred a temperature setting below than the level recommended by the Malaysian standard i.e., 24 ∘C. These results might be beneficial to understand the occupant behavior of electricity demand in Malaysia for designing smart grid energy systems in the future.

The relationship between energy consumption and gross domestic product in Hong Kong (1992–2015): Evidence from sectoral analysis and implications on future energy policy

The causal relationship between energy consumption and gross domestic product in Hong Kong from 1992 to 2015 is investigated in this study. Different from the previous studies focusing on the causal relationship between total energy consumption and total gross domestic product per capita, this study further investigates the causal relationship from sectoral perspective, including residential, commercial, industrial and transportation sectors. For each sector, the time series data of sectoral energy consumption and sectoral per capita value added are collected. To conduct the Granger causality test, the unit root test is first applied to analyse the stationarity of time series. The cointegration test is then employed to examine whether causal relationship exists in long-term. Finally, based on the aforementioned tests, both vector error correction model and vector autoregression model can be selected to determine the Granger causality between time series. It is interesting to find that the sectoral energy consumption and corresponding sectoral per capita value-added exhibit quite different causal relationships. For both residential sector and commercial sectors, a unidirectional causal relationship is found running from the sectoral per capita value added to sectoral energy consumption. Oppositely, for industrial sector and transportation sector, a unidirectional causal relationship is found running from sectoral energy consumption to sectoral per capita value added. Regarding the Granger causality test results, the indicative suggestions on energy conservation policies, energy efficiency policies and greenhouse gas emission reduction policies are discussed based on the background of Hong Kong’s economic structure and fuel types.

Capacity Expansion Pathways for a Wind and Solar Based Power Supply and the Impact of Advanced Technology—A Case Study for Germany

Wind and solar PV have become the lowest-cost renewable alternatives and are expected to dominate the power supply matrix in many countries worldwide. However, wind and solar are inherently variable renewable energy sources (vRES) and their characteristics pose new challenges for power systems and for the transition to a renewable energy-based power supply. Using new options for the integration of high shares of vRES is therefore crucial. In order to assess these options, we model the expansion pathways of wind power and solar photovoltaics (solar PV) capacities and their impact on the renewable share in a case study for Germany. Therefore, a numerical optimization approach is applied on temporally resolved generation and consumption time series data to identify the most efficient and fastest capacity expansion pathways. In addition to conventional layouts of wind and solar PV, our model includes advanced, system-friendly technology layouts in combination with electric energy storage from existing pumped hydro storage as promising integration options. The results provide policy makers with useful insights for technology-specific capacity expansion as we identified potentials to reduce costs and infrastructural requirements in the form of power grids and electric energy storage, and to accelerate the transition to a fully renewable power sector.

In-depth data on the network structure and hourly activity of the Central Chilean power grid

Network science enables us to improve the performance of complex systems such as traffic, communication, and power grids. To do so, it is necessary to use a well-constructed flawless network dataset associated with the system of interest. In this study, we present the dataset of the Chilean power grid. We harmonized data from three diverse sources to generate a unified dataset. Through an intensive review on the raw data, we filter out inconsistent errors and unrealistic faults, making the data more trustworthy. In contrast to other network dataset for power grids, we especially focus on preserving the physical structure of nodes’ connection incorporating the ‘tap’ structure. As a result, we provide three different versions of the dataset: ‘with-tap’, ‘without-tap’, and ‘reduced versions’. Along with structure, we incorporate various attributes of the nodes and edges such as the geo-coordinates, voltage of transmission lines, and the time series data of generation or consumption. These data are useful for network scientists to analyze the performance and dynamic stability of power grids.

Development and Validation of Algorithm to Distinguish ON/OFF State of Household Air Conditioners on The Basis of Time-Series Data of Electricity Consumption

Development and Validation of Algorithm to Distinguish ON/OFF State of Household Air Conditioners on The Basis of Time-Series Data of Electricity Consumption

Electricity Consumption Prediction Using XGBoost Based on Discrete Wavelet Transform

The purpose of this paper is to predict the daily electricity consumption of the next month. It is considerably important for people to cope with the problem well. Although few articles mentions the topic of electricity consumption prediction, numerous papers include some topic similar to the topic in this paper, such as rainfall forecasting, wind speed prediction and water flow forecasting. Moreover, a number of techniques and algorithms are employed to cope with those issues and achieve outstanding performance. Those techniques and algorithms are considerably remarkable, but the accuracy of them is not excellent enough on the long-term prediction of time series. In this paper, we propose a hybrid model which integrate discrete wavelet transform and XGBoost to forecast the electricity consumption time series data on the long-term prediction, namely DWT-XGBoost. The original time series data can decompose into approximate time series data and detail time series data by the discrete wavelet transform. And those time series data by decomposition are as features input into the prediction model that is XGBoost. Furthermore, the parameters of XGBoost are obtained by a grid search method. The performance of the proposed model in this paper is measured against with other hybrid models such as integrating discrete wavelet transform and support vector regression, integrating discrete wavelet transform and artificial neural networks, and unitary XGBoost. The comparison results show that the DWT-XGBoost outperforms other models and is a novel method on the long-term prediction of time series.