Power Consumption Behavior Research Articles

Anomalous load profile detection in power systems using wavelet transform and robust regression

In modern power systems, a clean database of historical electric loads is crucial for system analysis, planning studies, load forecasting, and decision making. To this end, the clean database should be free of anomalies while preserving the power consumption behavior of special events like holidays. In recent years, model-based anomaly detection methods have been successfully used for anomaly detection and correction in electrical load. In these methods, the collected load data are first compared with the predicted load obtained from a forecasting model; if the difference is greater than a specific limit, the collected load data are replaced with the predicted value. One of the disadvantages of these methods is the consideration of load profiles of the days with higher levels of difficulty in load forecasting, such as holidays, as anomalous load profiles and replacing them with predicted values, which results in fading some important characteristics of the load series. Therefore in this paper, a two-stage anomaly detection method is proposed to alleviate this problem. In the first stage, the wavelet transform is used to identify suspected load profiles. In the second stage, a robust regression method is used to further inspect the suspected load profiles and decide whether they are anomalies or not. By performing experiments on three datasets (a utility of Iran, New York City, and a North American utility) and comparing the results with four other methods, it’s shown that the proposed method decreases false detection of anomalies while accurately correcting anomalous load profiles. Results of load forecasting with the Iranian dataset corrected by the proposed method showed an average of 2.28% and 1.58% error on the holidays and total days, respectively, which is less than the other four methods.

Shared energy storage system for prosumers in a community: Investment decision, economic operation, and benefits allocation under a cost-effective way

With the rapid development of distributed renewable energy, energy storage system plays an increasingly prominent role in ensuring efficient operation of power system in local communities. However, high investment cost and long payback period make it impossible for prosumers to own the storage system. In this context, considering the complementarity of power generation and consumption behavior among different prosumers, this paper proposes an energy storage sharing framework towards a community, to analyze the investment behavior for shared storage system at the design phase and energy interaction among participants at the operation phase. Therefore, under the sharing mode that the operator invests and virtualizes physical storage system to provide virtual storage services for prosumers, this paper first establishes a two-stage optimization model. Based on a new storage sharing contribution rate index and asymmetric Nash bargaining model, a profit-allocation mechanism is designed. Then, combined with alternating direction method of multipliers, to achieve a mutual benefit and win-win situation. Finally, the above model is simulated by three buildings' realistic data, and the effectiveness of shared energy storage system is analyzed by using different technical and economic performance indexes. Additionally, the influence of battery cost, carbon tax, electricity price, and the uncertainties of renewable energy output power and load on shared storage system planning are also discussed. Numerical results show that, compared with personal energy storage scenario, the proposed storage sharing mechanism can achieve 6.09% cost savings, the self-consumption rate and self-sufficiency rate of renewable energy respectively increase by 5.01% and 5.21%, and all financial evaluation indexes have improved. In short, this paper can give practical guidelines for investors and prosumers to reasonably plan and share energy storage system, and provide realistic references for the government to effectively implement the shared energy storage.

Research on monitoring technology of power stealing behavior in bitcoin mining based on analyzing electric energy data

As we all know, the behavior of stealing electric energy governance is always the difficulty and key point in the management of electric power enterprises. In recent years, as bitcoin’s value continued to climb, the theft of electricity by bitcoin mining user began to appear. In order to solve the power theft problem of bitcoin mining users, we conducted an in-depth study on the power consumption behavior of such users based on the power data analysis technology. This paper analyzes the power consumption characteristics of bitcoin mining users and uses electric data acquire system to monitor power consumption behavior. The paper makes comparative analysis on the massive data such as voltage and power of the electric energy acquisition system, analyzes and calculates the Pearson correlation coefficient between the electricity consumption of each customer and the line loss statistics of the power station by using Pearson correlation algorithm combined with the power loss of the power station, and analyzes the suspected users of stealing electric energy by taking an actual example. Through our research, we found a total of 16 bitcoin miners suspected of stealing electricity. After on-site investigation and evidence collection, we found that 10 of the users did have abnormal power consumption, and the accuracy rate reached 62.50%. Therefore, the economic benefits of this research are very significant.

How social learning affects customer behavior under the implementation of TOU in the electricity retailing market

As a representative motivative scheme in demand response implementation, time-of-use (TOU) pricing has significant potentials in achieving load-shifting goals. Meanwhile, social learning among customers, becoming popular with the development of information technologies, also profoundly affects customer consumption decisions. Still, its impacts on power customer consumption behaviors under implementing the TOU scheme remain unexplored. We propose an evolutionary game-theoretical model that incorporated social learning to illustrate how customer power consumption behaviors evolve if they learn from each other, where consuming relations between power consumers and retailers are formulated on account of the bipartite graphs. In every period, the retailers establish the TOU scheme based on the historical demand to optimize its current profits and load-shifting effects, and the customers, in turn, dispatch their consumptions for utility maximizations subjected to the proposed prices. We simulate the different scenarios of consumer behavior decisions verify our model by empirical data. Our results demonstrate how social learning changes consumer behavior and maximizes all consumers' utility. Noteworthy, Social learning brings volatility to the market, which incurs fluctuations for both firms' payoffs and customer utilities. Besides, we derive that the rigid price caps can guarantee customers positive utilities, and more communications among customers will lead to the decay of the entire retailing profits. This research provides decision support for consumer behavior and pricing of power retailing companies in the context of social learning.

Evaluation of Online Machine Learning Algorithms for Electricity Theft Detection in Smart Grids

Electricity theft-induced power loss is a pressing issue in both traditional and smart grid environments. In smart grids, smart meters can be used to track power consumption behaviour and detect any suspicious activity. However, smart meter readings can be compromised by deploying intrusion tactics or launching cyber attacks. In this regard, machine learning models can be used to assess the daily consumption patterns of customers and detect potential electricity theft incidents. Whilst existing research efforts have extensively focused on batch learning algorithms, this paper investigates the use of online machine learning algorithms for electricity theft detection in smart grid environments, based on a recently proposed dataset. Several algorithms including Naive Bayes, K-nearest Neighbours, K-nearest Neighbours with self-adjusting memory, Hoeffding Tree, Extremely Fast Decision Tree, Adaptive Random Forest and Leveraging Bagging are considered. These algorithms are evaluated using an online machine learning platform considering both binary and multi-class theft detection scenarios. Evaluation metrics include prediction accuracy, precision, recall, F-1 score and kappa statistic. Evaluation results demonstrate the ability of the Leveraging Bagging algorithm with an Adaptive Random Forest base classifier to surpass all other algorithms in terms of all the considered metrics, for both binary and multi-class theft detection. Hence, it can be considered as a viable option for electricity theft detection in smart grid environments.

Power Monitoring and Passenger Classification on Logistics Elevator

The elevator has an important role in assisting transportation and logistics activities in a building. However, if the elevator is not used wisely, then the power consumption will be inefficient. A policy of elevator usage is necessary to ensure the effectiveness of elevator power consumption. Therefore, in this study, elevator power consumption monitoring is proposed. The power consumption behavior can be learned so a suitable policy can be made accordingly. Two elevators in Telkom Campus Surabaya are monitored to understand the daily electrical energy usage. Internet of Things (IoT) based real-time power monitoring system is used to monitor the electrical energy usage of the elevator. A raspberry pi is used to collect the data of electrical usage via a current and power sensor. The data is sent to the cloud, which later is displayed through a dashboard website. The result shows that the elevator usage on weekdays and weekends is different. The peak power on weekdays is obtained from 15.00 to 16.00, meanwhile, on weekends, the peak occurs from 9:00 to 10:00. On weekdays, the total power consumed by the elevator is 51.74kW, while on weekends, it is 11.94kW. Restrictions on the use of lifts are applied to periods when the lift has few passengers and has a short distance. From the results obtained, the total power consumed can decrease by an average of 37%. It is expected that the suggested policies can reduce elevator power consumption and the monthly cost of electrical energy.

Demand Response Requirements from the Cultural, Social, and Behavioral Perspectives

Demand-side response programs, commonly known as demand response (DR), are interesting ways to attract consumers’ participation to improve electric consumption patterns. Customers are encouraged to modify their usage patterns in reaction to price increases through DR programs. When wholesale market prices are high or network reliability is at risk, DR can help to establish a balance between electricity generation and consumption by providing incentives or considering penalties. The overall objective of adopting DR programs is to increase network reliability and decrease operational costs. Nevertheless, the successful deployment of DR programs requires a set of conditions without which no success can be guaranteed. Implementing DR programs and achieving customers’ optimal power consumption behavior could be obtained through technical methods, such as using smart home appliances and big data techniques. However, even if each of these approaches is correctly implemented, they are not able to address all aspects of the problem. The findings of several studies demonstrate that, in addition to technical and economic concerns, social, cultural, and behavioral variables play a significant role in DR implementation. Therefore, this paper investigated the social, cultural, and behavioral variables as critical requirements for implementing DR programs. Furthermore, a theoretical framework and an analytical model of the elements impacting the electricity consumption are introduced that should be considered by the planners.

Factors associated with undernutrition and improvement in Indonesia

Stunting among children under five years of age is one of the big problems in Indonesia, damaging adulthood health and work productivity. Studies on the individual and household factors of this issue had been conducted massively. Nonetheless, regional level study on the same issue is still lacking whereas it is important in understanding the problems to formulate better policies. This study aimed to analyze factors associated with stunting prevalence at the provincial level. Data for this study was obtained from the publication of the National Socioeconomic Survey 2018 and Basic Health Research 2018. Multiple linear regression was applied to analyze factors associated with stunting prevalence. There were proportion of expenditure on fish, meat, eggs and milk, tofu and tempe, fruits and vegetables (F&V), cigarettes, and energy and protein intake. Prevalence of stunting at the provincial level was very wide, from 16.2% to 37.9%. Protein adequacy was higher (106.2%) than the Recommended Dietary Allowance. Average proportion of food expenditure at provincial level was 50.1%, in each province ranging from 40.6% to 57.2%; while the highest proportion of expenditure was to purchase F&V (12.8%). The model predicted that the proportion of household food expenditure, proportion of F&V, and protein adequacy were significantly associated with stunting prevalence. Thus, policy on improving household purchasing power and food consumption behavior with balanced nutrition is the key factor in reducing the prevalence of stunting. Social protection programs such as cash transfer and food aid for low-income households should be continued with more accurate recipients.

Electricity Consumption Data Analysis Using Various Outlier Detection Methods

Nowadays, detecting abnormal power consumption behavior of householders has become a big concern in the smart energy field; overcoming this limitation will help in identifying efficient solutions to reduce power consumption. This paper proposes a new methodology for detecting abnormal energy consumption in residential buildings based on hourly readings of energy consumption and peak energy consumption. The proposition is implemented using three unsupervised outlier detection methods (isolation forest, one-class SVM, and k-means). The authors propose this solution to help residents in reducing operating costs by detecting consumption failures that cannot be detected easily. On the other hand, energy providers will have the access to detailed data about anomalies, faulty appliances, and houses with poor power control strategy in general, which will help in pinpointing overconsumption problems, thus enhancing human awareness and reducing energy consumption.

Probabilistic based CMOS Adder for High Speed Communication Systems

Power efficient is an important availability for various mobile devices and communication system applications. The proposed probabilistic adder is to trade a lesser amount of accuracy with reduced power dissipation. In this paper, the probabilistic adder is eliminating the some part of the carry propagation path in least significant bit to reduce the power consumption and transistor count. The power consumption and probabilistic error behaviour of the proposed adder is designed and compared with other adders.

A bottom-up short-term residential load forecasting approach based on appliance characteristic analysis and multi-task learning

Residential load forecasting has been an intractable problem due to the small load scale and strong randomness of residents' power consumption behaviors. The traditional methods only use load data at household-level and cannot adequately consider the power consumption habits of users, so the forecasting effect is limited. To address these challenges, this paper proposes a bottom-up approach considering the load characteristic of the appliance-level. The novelty lies in the following three aspects: 1) considering the working principles and load characteristic of appliances, household appliances are classified into continuous and intermittent load appliances, to achieve more refined load forecasting by aggregating loads of appliances; 2) considering the fluctuation of load curves, the seasonal-trend decomposition procedure based on Loess (STL) is applied to continuous load appliances to obtain more fine-grained and regular load data, which can give full play to the advantage of gated recurrent unit (GRU) for time series prediction; 3) in order to take power consumption behavior of users and correlations of appliances into account, a multi-task learning (MTL) network is designed for intermittent load appliances, and the model can enhance computing efficiency by predicting multiple appliances simultaneously. Case studies show that the proposed method can effectively reduce error of residential load forecasting.

Empirical power consumption characterization and energy aware scheduling in data centers

Energy-efficient management is key to reduce operational cost and environmental contamination in modern data centers. Energy management and renewable energy utilization are strategies to optimize energy consumption in high-performance computing. In any case, understanding the power consumption behavior of physical servers in datacenter is fundamental to implement energy-aware policies effectively. These policies should deal with possible performance degradation of applications to ensure quality of service. This manuscript presents an empirical evaluation of power consumption for scientific computing applications in multicore systems. Three types of applications are studied, in single and combined executions on Intel and AMD servers, for evaluating the overall power consumption of each application. The main results indicate that power consumption behavior has a strong dependency with the type of application. Additional performance analysis shows that the best load of the server regarding energy efficiency depends on the type of the applications, with efficiency decreasing in heavily loaded situations. These results allow formulating models to characterize applications according to power consumption, efficiency, and resource sharing, which provide useful information for resource management and scheduling policies. Several scheduling strategies are evaluated using the proposed energy model over realistic scientific computing workloads. Results confirm that strategies that maximize host utilization provide the best energy efficiency.

Optimal control and learning for cyber‐physical systems

Optimal control and learning for <scp>cyber‐physical</scp> systems

Experimental and theoretical studies on energy characteristics of hydraulic hybrids for thermal management

The development of fuel-efficient vehicles has increased along with the needs of consumers and the rise of environmental issues. Hybrid vehicles have become a popular alternative to conventional powertrains in the passenger vehicle market. Hydraulic hybrid vehicles have been applied to certain applications with clear benefits, such as high power density. To increase the base of academic knowledge, the energy characteristics of hydraulic hybrids were experimentally and theoretically studied for thermal management. First, measurement data for a series hydraulic hybrid system were collected by hardware in the loop test rig for different temperature conditions. The experimental results were compared for different thermal management conditions, thus showing the power consumption behaviors at different system temperatures. Also, a theoretical approach with thermodynamic properties was introduced for analyzing the energy characteristics of hydraulic hybrid systems. The theoretical results revealed possible explanations for the relationship between energy saving and thermal management for hydraulic hybrid systems. This study was worthy not only in that the experimental data itself worth as reference data for energy characteristics of hydraulic hybrid systems for thermal management, but also in that a novel theoretical approach for energy characteristics analysis was introduced, using energy and exergy analysis for the first time.

Abnormal behavior analysis of electricity consumption based on improved random forest with grey relation projection

s: With the construction of ubiquitous power Internet of things and the deepening application of big data technology, it is possible to mine the abnormal behavior of electricity consumption from the massive power consumption information. In view of the current application of anti stealing power in power system, an improved random forest algorithm based on grey relation projection is proposed to analyze the abnormal behavior of power consumption. The improved stochastic Sen algorithm is used to build the diagnosis model of abnormal electricity consumption of users, and analyze the power data of the power consumption information acquisition system, so as to improve the accuracy of the intelligent diagnosis algorithm of anti stealing electricity. The effectiveness of the proposed algorithm can be obtained by verification test.

Detect the electricity theft event using text CNN

The big data technology has been widely used in power consumption behavior analysis and power user portrait. In this paper, the electricity data is constructed as two-dimensional time-series. Based on the designed data structure, a special kind of Artificial Neural Networks (ANNs) named as text convolutional neural networks (TCNN) is proposed for electricity theft detection. Moreover, considering the imbalance of electricity theft data in realistic datasets, a data augmentation method is proposed. Numerical results obtained on realistic datasets validate the proposed model.

DGX-A100 Face to Face DGX-2—Performance, Power and Thermal Behavior Evaluation

Nvidia is a leading producer of GPUs for high-performance computing and artificial intelligence, bringing top performance and energy-efficiency. We present performance, power consumption, and thermal behavior analysis of the new Nvidia DGX-A100 server equipped with eight A100 Ampere microarchitecture GPUs. The results are compared against the previous generation of the server, Nvidia DGX-2, based on Tesla V100 GPUs. We developed a synthetic benchmark to measure the raw performance of floating-point computing units including Tensor Cores. Furthermore, thermal stability was investigated. In addition, Dynamic Frequency and Voltage Scaling (DVFS) analysis was performed to determine the best energy-efficient configuration of the GPUs executing workloads of various arithmetical intensities. Under the energy-optimal configuration the A100 GPU reaches efficiency of 51 GFLOPS/W for double-precision workload and 91 GFLOPS/W for tensor core double precision workload, which makes the A100 the most energy-efficient server accelerator for scientific simulations in the market.

Construction of on-line monitoring system for self provided power plants in Shandong Power Grid

Based on the analysis of the problems existing in the operation of self provided power plants in Shandong power grid, it is proposed to develop and build an on-line monitoring system for self-contained power plants, which can realize on-line monitoring of power generation and consumption operation and on-line analysis of power generation and consumption behavior characteristics, realize automatic collection of electricity data, provide scientific and quantitative monitoring means, and promote standardized operation and orderly development of self provided power plants.

Abnormal Power Consumption Detection Based on Data-Driven

Based on high dimensional random matrix theory and machine learning algorithm, a method to detect abnormal power consumption behaviour of users is proposed. Firstly, the K-means clustering algorithm is used to divide the power loads into load types that obey specific distribution law or with random fluctuation. Then the linear eigenvalue statistics (LES) index can be used to detect the abnormal power consumption behaviour for the former such as unimodal load or bimodal load. And the difference between the actual and predicted value of regression model based on XGBoost algorithm can be used as the basis for judging abnormal power consumption behaviour of the latter. The method proposed in this paper is applicable to different types of loads and can implement a good discriminant effect.

Study on intelligent anti–electricity stealing early-warning technology based on convolutional neural networks

In recent years, electricity stealing has been repeatedly prohibited, and as the methods of stealing electricity have become more intelligent and concealed, it is growing increasingly difficult to extract high-dimensional data features of power consumption. In order to solve this problem, a correlation model of power-consumption data based on convolutional neural networks (CNN) is established. First, the original user signal is preprocessed to remove the noise. The user signal with a fixed signal length is then intercepted and the parallel class labelled. The segmented user signals and corresponding labels are input into the convolutional neural network for training, and the trained convolutional neural network is then used to detect and classify the test user signals. Finally, the actual steal leak dataset is used to verify the effectiveness of this algorithm, which proves that the algorithm can effectively carry out anti–-electricity stealing by warning of abnormal power consumption behavior. There are lots of line traces on the surface of the broken ends which left in the cable cutting case crime scene along the high-speed railway in China. The line traces usually present nonlinear morphological features and has strong randomness. It is not very effective when using existing image-processing and three-dimensional scanning methods to do the trace comparison, therefore, a fast algorithm based on wavelet domain feature aiming at the nonlinear line traces is put forward to make fast trace analysis and infer the criminal tools. The proposed algorithm first applies wavelet decomposition to the 1-D signals which picked up by single point laser displacement sensor to partially reduce noises. After that, the dynamic time warping is employed to do trace feature similarity matching. Finally, using linear regression machine learning algorithm based on gradient descent method to do constant iteration. The experiment results of cutting line traces sample data comparison demonstrate the accuracy and reliability of the proposed algorithm.