Characteristics Of Appliances Research Articles

Research on Privacy Issues in Smart Metering System: An Improved TCN-Based NILM Attack Method and Practical DRL-Based Rechargeable Battery Assisted Privacy Preserving Method

Smart meters, as a key component of Advanced Metering Infrastructure (AMI), collect fine-grained electricity consumption data for demand response in smart grids. While this data improves the grid’s accuracy, it also poses significant threats to users’ privacy. In this paper, we study the privacy issue in smart metering systems from both attacker and defender perspectives. First, we propose an improved Temporal Convolutional Network (TCN) based Non-Intrusive Load Monitoring (NILM) attack method, which infers electrical appliance usage from public load curves, addressing the gradient vanishing, gradient exploding, and other problems while improving attack accuracy. Second, we develop a rechargeable battery-assisted energy management system to hide load characteristics of electrical appliances by adding physical noise, thus resisting NILM attacks. To address the privacy-cost trade-off optimization problem, we propose a Practical Deep Reinforcement Learning-based Rechargeable Battery assist Privacy Preserving Method (PRoP) that learns optimal battery charging/discharging policies. We design a novel privacy measurement method and constraints to ensure the feasibility of system deployment and prove PRoP’s effectiveness in resisting NILM attacks. Comprehensive evaluations demonstrate that our improved TCN-based NILM method achieves an attack success rate of over 80% on various electrical appliances, improving attack performance (MAE, RMSE) by 20% compared to existing methods while reducing model training time. Moreover, our proposed PRoP achieves a better trade-off between privacy protection and electricity cost than existing battery-assisted methods, reducing costs by 5% and the attack success ratio to 36%, while increasing the MAE and RMAE obtained by NILM by 3 times. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —Smart grid, which can support bidirectional information transmission, has a series of advantages, such as high efficiency and high stability. However, it also brings a significant threat to users’ electricity privacy. Although encryption-based privacy protection methods have been deployed on terminal devices of smart grids to prevent privacy leaks, this method can often only defend against intrusive attacks and has little effect on non-intrusive attacks. To this end, this paper studies the privacy issues caused by non-intrusive attacks. Specifically, to better study the protection method, we first investigate the attack mechanism and design an improved TCN-based Non-Intrusive Load Monitoring method. Then, we propose a Practical Reinforcement learning-based rechargeable battery-assisted Privacy preserving method (PRoP) to defend against this attack physically. The most practical contribution of this paper is that, compared with existing battery-assisted privacy protection methods, we do not blindly pursue algorithm performance but fully consider the practical factors of deployment, such as limiting battery capacity and constraining battery charging and discharging behavior. This can guide practitioners to better apply this technology in practice.

Multi-Time-Scale Shapelet-Based Feature Extraction for Non-Intrusive Load Monitoring

The performance of non-intrusive load monitoring (NILM) system is closely related to the uniqueness and comprehensiveness of the load features extracted from the equipment. However, existing feature extraction methods for NILM are difficult to extract interpretable intuitive features representing the uniqueness of load, and lack of comprehensive attention to the features on different time scales. Therefore, this paper proposes a multi-time-scale shapelet based feature extraction framework. First, shapelet, the most unique segment of a sequence, which can distinguish it from other sequences is introduced as a feature and multi-time-scale current shapelets are proposed to extract comprehensive characteristics of electric appliances and offer intuitive interpretability. Meanwhile, a shapelet transform is constructed, so that the essence of shapelet based NILM is transformed from sequence matching to a general classification problem. Next, an ensemble bagging classifier is proposed on multi-time scales to effectively balance the classification errors on each time scale and enhance the performance of NILM system. In addition, this framework can support small sample training to facilitate rapid deployment of NILM applications at the initial practical stage. The results show that this framework can effectively focus on the uniqueness of load on different time scales, provide interpretability for NILM, and achieve good classification effects.

A Systematicity Review on Residential Electricity Load-Shifting at the Appliance Level

Load-shifting is a demand-side management (DSM) strategy to support the efficiency of the electricity grid during hours of peak demand. Load-shifting at the appliance level is an interesting topic to review, since appliance usage is one of the main inputs of the load-profile analysis. More literature reviews on load-shifting at the appliance level are required, as this is a specific issue in the body of literature on load-profile research, though only a limited number of studies are available at this time. It is also essential to focus on appliance usage patterns to improve our understanding of the impacts and characteristics of different appliances. Existing studies on load-shifting have used commonly structured literature reviews; our work addresses the transparency of each stage and substage in the selection of the final list of studies. The findings show that efficiency has been achieved in installed-capacity reductions; costs, including those of emission reductions; and peak consumption reductions. The most frequently used method in load-shifting at the appliance level is to develop load-shifting optimization algorithms. This work contributes by providing a transparent process of drawing up a systematicity literature review as a source of knowledge and grounded theory. It also contributes to specific research on load-shifting at the appliance level by highlighting and discussing the key findings for the reader. In particular, it contributes to improving energy efficiency by describing load-shifting methods at the appliance level and identifying both controllable and uncontrollable appliances. This detailed literature review at the appliance level can make valuable contributions in support of decision- and policymaking by illuminating new dynamic systems specifically in load-shifting and in demand-side management in general for energy efficiency purposes.

The Impact of RMB Exchange Rate Fluctuation on China's Air Conditioning Industry

Based on the background of China's current trade with other countries is becoming increasingly close, the study of exchange rate stability on the impact of China's trade with other countries to promote China's economic stability has a certain theoretical significance. In the study of air conditioning industry import and export trade factors affecting the attention of the national exchange rate is very important, in the two countries trade relations, the exchange rate is a measure of the two countries to buy, sell the price of the product of the vital factors, exchange rate changes for the enterprise will bring a certain degree of impact, stabilization of the exchange rate for the promotion of our country and the trade of other countries in the long term and good cooperation has important practical significance. The air-conditioning manufacturing industry is a very representative research object. The cost-sensitive home appliance industry is a labor-intensive industry, which provides a huge number of jobs for China and creates a huge amount of foreign exchange income. At the same time, the home appliance industry, as a representative of China's manufacturing industry, faces industrial upgrading pressure and is in the maturity of a fully competitive industry, which has been severely impacted by exchange rate changes. It is of great significance to correctly measure and analyze the foreign exchange risk exposure of the import and export of the air-conditioning industry and its influencing factors. From a micro point of view, enterprises can be based on foreign exchange risk exposure situation targeted adjustment of financial strategies, strategies and management methods. From a macro point of view, the country can be based on the overall foreign exchange risk exposure characteristics of home appliances to formulate the correct macroeconomic policy.

Electricity Theft Detection Based on Contrastive Learning and Non-Intrusive Load Monitoring

Electricity theft has caused enormous damage to grids safety and economy globally, bringing plentiful attention to electricity theft detection. However, the inherent problems of data imbalance, data sparsity, and data shift due to the residents will or the seasons remain challenges. Besides, there is a lack of appliance-level information provided by non-intrusive load monitoring to improve the data graininess in electricity theft detection. To address these problems, an electricity theft detection algorithm based on contrastive learning and non-intrusive load monitoring is proposed. Firstly, a semi-supervised learning architecture composed of Gramian angular field encoding for sequence visualization and contrastive learning architecture featuring few-shot learning is established for load monitoring and initial detection considering the inherent operating characteristics of appliances. Furthermore, after the filtration of typical regular-switched appliances by Kendalls coefficient of concordance, in-depth detection of abnormal operation routines of appliances is conducted for suspicious residents, aiming to improve the fault-tolerant ability and resist the emergence of unknown electricity theft methods. Finally, the electricity theft probability is computed to confirm fraudulent users. Both the public and practical datasets are utilized to verify the effectiveness of the proposed study, and the results show an overall better performance compared to other state-of-the-art algorithms.

Non-Intrusive Arc Fault Detection and Localization Method Based on the Mann–Kendall Test and Current Decomposition

In recent years, electrical fires caused by arc faults have been increasing, seriously affecting the safety of people’s lives and property. Considering the complex arc fault characteristics of actual low-voltage users, the non-intrusive arc fault detection and localization method is studied. First, the characteristics of arc current waveforms are analyzed, and event detection based on the Mann–Kendall Test is performed for the difference between the current waveforms of two adjacent cycles, rather than using the current waveforms directly. Then, the current waveforms of the two segments are calculated via subtraction to obtain the current waveform of the electric appliances causing the event. A current feature parameter database of the normal and arc currents is constructed via harmonic analysis, and a multi-appliance current decomposition model considering the sparse operation characteristics of appliances is established; thus, the arc localization problem is transformed into an optimization problem. Finally, a genetic algorithm is used to optimize the differential current decomposition results, and then, locate the arc fault. A household arc fault simulation experiment is carried out for the common electric appliances of actual low-voltage users. The experimental results show that the proposed non-intrusive arc fault detection and localization method is effective.

Optimization of the electricity consumption behaviors of users under uncertain electricity prices and consumption patterns

Real-time electricity prices and customer demands are inextricably linked in the smart grid environment. For the purpose of reducing grid load peaks, enhancing grid security, and optimizing customers’ electricity bills, it is critical to model dynamic real-time electricity prices and flexible customer requests, along with providing an optimized plan to guide users’ decisions regarding the use of household appliances and quantify their needs. The solution to this problem, however, is not straightforward due to the following factors. Firstly, the real-time electricity price is constantly changing, so it must be estimated from grid data associated with it. Furthermore, there are often some variations in the actual power consumption of appliances due to the uncertainty of human behavior, the characteristics of appliances, and the influence of the environment. In addition, when optimizing, serious problems such as tripping and circuit overload must be taken into account. Therefore, considering random electricity prices and random electricity consumption, in the paper, we present the models to simulate real-time changes in electricity prices and energy consumption behaviors, as well as a method to help users optimize their use of household appliances and quantify their requirements with lower electricity cost. By analyzing the experimental results, we demonstrate that our work can be used to provide a scheduling scheme for appliances that minimizes electricity costs and reduces grid pressure.

Event-driven non-intrusive load monitoring algorithm based on targeted mining multidimensional load characteristics

Nowadays, the advancement of nonintrusive load monitoring (NILM) has been hastened by the ever-increasing requirements for the reasonable use of electricity by users and demand side management. Although existing researches have tried their best to extract a wide variety of load features based on transient or steady state of electrical appliances, it is still very difficult for their algorithm to model the load decomposition problem of different electrical appliance types in a targeted manner to jointly mine their proposed features. This paper presents a very effective event-driven NILM solution, which aims to separately model different appliance types to mine the unique characteristics of appliances from multi-dimensional features, so that all electrical appliances can achieve the best classification performance. First, we convert the multi-classification problem into a serial multiple binary classification problem through a pre-sort model to simplify the original problem. Then, ConTrastive Loss K-Nearest Neighbour (CTLKNN) model with trainable weights is proposed to targeted mine appliance load characteristics. The simulation results show the effectiveness and stability of the proposed algorithm. Compared with existing algorithms, the proposed algorithm has improved the identification performance of all electrical appliance types.

A novel dynamic load-priority-based scheduling strategy for home energy management system

In order to increase renewable energy consumption and enhance energy efficiency, residents participate in the utility in the form of demand response. To address the challenges of imprecise energy efficiency services and inadequate response of household energy use, a novel dynamic-load-priority-based scheduling strategy is proposed. Firstly, according to the operation law of household appliances, the control models of three kinds of flexible loads are established respectively. Secondly, according to the characteristics of household appliances, the priority evaluation index of household appliances including user comfort, electricity cost and task urgency index is proposed. The Technique for Order Preference by Similarity to Ideal Solution based on entropy weight is used to prioritize the household appliances that control the scheduling queue. Finally, according to the sorting results, the load control strategy is implemented to reasonably arrange the orderly work of household electrical equipment. Compared with the uncontrolled state, the proposed strategy can limit the total power of appliance load operation to the demand range in the peak load condition, and save 3% of electricity cost and reduce 5.4 kW of peak load under the demand limit of 12 kW, and 32% of electricity cost and 10.6 kW of peak load under the demand limit of 7 kW, without affecting the basic life of customers. In the period of new energy saturation, the load can also be dispatched to participate in the new energy consumption, thus realizing the coordinated interaction between residential energy consumption and the utility.

An ensemble-policy non-intrusive load monitoring technique based entirely on deep feature-guided attention mechanism

Non-intrusive load monitoring (NILM), as an important part of intelligent electricity consumption, improves the cognitive level of the load by analyzing the bus power in a specific range to obtain refined electricity consumption data at appliance level in that range, such as the number of loads, load categories, load operating status and corresponding energy consumption. For a long time, the research on NILM algorithm has been quite in depth, especially with the deep learning method being used in NILM, breakthrough progress has been made. In this paper, an integrated policy non-intrusive load monitoring technique based entirely on deep feature-guided attention mechanism is constructed based on sequence-to-point deep neural networks. Firstly, according to the operating cycle characteristics of different appliances, the data are segmented with different sliding windows and the integerized power values are mapped to high-dimensional feature vectors by word embedding, which transforms the value regression problem of decomposed power into a multiclassification problem to find the probability of decomposed power values at each discrete integer power value. Then the encoder composed of deep residual network is used for multi-layer feature extraction, which is used to obtain richer feature information, and the extracted feature quantities will be sent to the decoder through two paths respectively, one is mapped to the power feature space through word embedding as the input to the decoder; the other is sent to the decoder based entirely on the attention mechanism through convolution as the checked vector K and the content vector V. The decoder obtains the information with the highest relevance to the current moment by assigning weights to the before and after information, and thus decodes the final load decomposition result. Finally, this paper introduces a snapshot technique to obtain an inexpensive multi-model integration strategy, which improves the generalization ability of the model and the accuracy of the load decomposition. The effectiveness of the proposed method is verified by experiments on the REDD dataset and UK-DALE dataset.

Non‐intrusive load monitoring by voltage–current trajectory enabled asymmetric deep supervised hashing

With the development of smart grids, appliance-level data information plays a vital role in smart power consumption. Nowadays, appliance signatures detected by non-intrusive load monitoring (NILM) can be used for anomaly detection, demand response, and electricity management. These applications increase the requirements for the accuracy of appliance identification in NILM. And it has been proved that the voltage–current (V–I) trajectory can be applied as an effective load signature to represent the electrical characteristics of appliances with different statuses in previous researches. In this paper, a V–I trajectory enabled asymmetric deep supervised hashing (ADSH) method has been proposed for NILM. ADSH method converts load identification into the large-scale approximate nearest neighbour search. Different from the existing methods, ADSH treats the query images and database images in an asymmetric way in order to improve accuracy. More specifically, ADSH learns a deep hash function only for query images, while the hash codes for database images are directly learned. The experimental results on REDD and PLAID datasets show that the proposed method significantly improves the accuracy of load identification compared with state-of-art methods.

Energy disaggregation considering least square error and temporal sparsity: A multi-objective evolutionary approach

In energy disaggregation (ED), an accurate estimation of each appliance power consumption over time is crucial for practical applications. Most optimization-based ED approaches are single-objective and the disaggregated appliance power consumption profiles provided by them do not match the actual operational characteristics of the devices. In this paper, we propose a Multi-objective ED (MOED) framework considering two conflicting objectives - 1) minimizing the least square error between the measured and estimated aggregated signals; and 2) minimizing the temporal sparsity of appliance switching events. The multi-objective consideration results in a set of trade-off solutions where each solution represents a prospective disaggregation corresponding to the operation of different devices present. From the trade-off set, a disaggregated solution where the operational characteristics of appliances are close to practical operation is identified by evaluating statistical similarity. For each prospective solution, statistical similarity is evaluated with respect to a set of reference signals corresponding to the practical device operation provided by the device manufacturer. In addition, to effectively solve the MOED problem, an initialization based on Integer Programming (IP), problem-specific mutation and crossover operators are proposed. The performance of the proposed MOED framework is tested on real-world dataset from University of Victoria (UVic) and compared with other optimization approaches. On various indicators MOED shows better performance than conventional and improved Integer Programming based single-objective ED algorithms. In addition, the disaggregated signals provided by MOED are close to the practical operational characteristics of the respective appliances compared to disaggregated signals provided by the state-of-the-art ED approaches.

Resident Behavior Detection Model for Environment Responsive Demand Response

Due to emerging environmental problems from fossil fuel usage and increasing renewable energy resources in microgrids, the need of demand response (DR) is intensified to stabilize irregular balance and to mitigate carbon dioxide emissions simultaneously. However, in order to estimate practical and factual DR potential, an analysis of load characteristics including user behavior is required based on information data. Thus, in this study, a novel analytic approach is proposed to detect resident behavior for environment responsive DR potential estimation. A new framework to determine optimal DR potential is proposed with related data analysis models. In each analysis model, resident occupancy behavior and forecasted renewable energy generation are estimated. During the process of resident behavior detection, sub-metering data of appliances are analyzed and profiled for multiple purposes by a new method based on hidden Markov model and time varying Markov chain. The optimal DR potential is estimated considering guaranteed residents' comfort constraints and the dynamic characteristics of appliances. The proposed framework is tested on the single household environment of residential region and the results present that the proposed model is environmentally and economically effective, while also suggesting meaningful implications through resident behavior data analysis.

MATHEMATICAL MODEL OF UNSTEADY HEAT TRANSFER OF PASSENGER CAR WITH HEATING SYSTEM

Purpose. The existing mathematical models of unsteady heat processes in a passenger car do not fully reflect the thermal processes, occurring in the car wits a heating system. In addition, unsteady heat processes are often studied in steady regime, when the heat fluxes and the parameters of the thermal circuit are constant and do not depend on time. In connection with the emergence of more effective technical solutions to the life support system there is a need for creating a new mathematical apparatus, which would allow taking into account these features and their influence on the course of unsteady heat processes throughout the travel time. The purpose of this work is to create a mathematical model of the heat regime of a passenger car with a heating system that takes into account the unsteady heat processes. Methodology. To achieve this task the author composed a system of differential equations, describing unsteady heat processes during the heating of a passenger car. For the solution of the composed system of equations, the author used the method of elementary balances. Findings. The paper presents the developed numerical algorithm and computer program for simulation of transitional heat processes in a locomotive traction passenger car, which allows taking into account the various constructive solutions of the life support system of passenger cars and to simulate unsteady heat processes at any stage of the trip. Originality. For the first time the author developed a mathematical model of heat processes in a car with a heating system, that unlike existing models, allows to investigate the unsteady heat engineering performance in the cabin of the car under different operating conditions and compare the work of various life support systems from the point of view their constructive solutions. Practical value. The work presented the developed mathematical model of the unsteady heat regime of the passenger car with a heating system to estimate the efficiency of unsteady, transitional temperature states in passenger cars, taking into account the design features of the heating system and the regulatory requirements. This allows the development and implementation of optimal technical characteristics of heating appliances and the construction of an algorithm for controlling their operation in accordance with operating conditions, taking into account the thermal inertia of the car in the transitional modes of heating, on the basis of mathematical modeling.

Measurement of Static Frequency Characteristics of Home Appliances in Smart Grid Systems

The current transformation of power systems is aiming towards distributed source integration and general decentralization. Renewable energy sources and support of local energy supply create conditions for widespread use of new technologies and smart grids. As the electrical grids become more electrically independent, the importance of frequency control will rise. Stability of the system in such cases is no longer only relying on rotating inertia of generators as in the centralized grid. This known scenario has already been analyzed by many with computational models for optimal safety precautions of the grid. This paper aims to update the common home appliance frequency characteristics through measurements and compare them to those currently used. These devices were divided into two groups: general categorization and light sources. Subsequently, the frequency sensitivity coefficients were evaluated and analyzed home appliances were sorted into three categories according to the size of their frequency sensitivity coefficient values: positive, negative, and no effect. The results were compared with studies aimed at evaluating the static load characteristics. A simplified simulation of the frequency control, presented in the discussion section, was carried out to determine the consequences of the newly measured characteristics and concludes the paper.

Non-Intrusive Load Monitoring via Deep Learning Based User Model and Appliance Group Model

Non-Intrusive Load Monitoring (NILM) increases awareness on user energy usage patterns. In this paper, an efficient and highly accurate NILM method is proposed featuring condensed representation, super-state and fusion of two deep learning based models. Condensed representation helps the two models perform more efficiently and preserve longer-term information, while super-state helps the model to learn correlations between appliances. The first model is a deep user model that learns user appliances usage patterns to predict the next appliance usage behavior based on past behaviors by capturing the dynamics of user behaviors history and appliances usage habits. The second model is a deep appliance group model that learns the characteristics of appliances with temporal and electrical information. These two models are then fused to perform NILM. The case study based on REFIT datasets demonstrates that the proposed NILM method outperforms two state-of-the-art benchmark methods.

Home appliances recommendation system based on weather information using combined modified k-means and elbow algorithms

&lt;p&gt;The recommendation system is an intelligent system gives recommendations to users to discover the best interesting items. The purpose of this proposed recommendation system is to develop a system to find the best electrical devices according to weather conditions and user preferences. The proposed solution relies on the characteristics of electrical appliances and their suitability to weather conditions in any city. The proposed solution is the first recommendation system combines devices properties, weather conditions, and user preferences using a new combination of algorithms. The clustering algorithms are the most applicable in the field of recommendation system. The proposed solution relies on a combination of Elbow method, pro­­posed modified K-means and Silhouette algorithm to find the best number of clusters before starting the clustering process. Then calculate the weights for each cluster and compare them with the weather weights to find the required clusters sorted from the near to far according to a computed threshold. The empirical results showed that the proposed solution demonstrated a 94% accuracy to match the characteristics of the recommended devices with the climatic characteristics of the region and user preferences. The accuracy is measured using Silhouette algorithm.&lt;/p&gt;

Residential Household Non-Intrusive Load Monitoring via Smart Event-based Optimization

Home energy management requires accurate information about the appliances’ consumption pattern. This information can help consumers save energy, control their usage by shifting their usage to off-peak hours and reduce their electricity costs. Non-intrusive load monitoring (NILM) in which the power consumption profile of appliances are extracted from the aggregated signal of a household, provides this information. For the NILM problem, machine learning approaches as the training-based solutions require large training datasets for an accurate disaggregation and the optimization-based approaches employs prior information about the characteristics of appliances. This paper proposes a novel event-based optimization algorithm. In its first stage, the prior information about appliances is extracted from the events of the consumption profiles of appliances by means of clustering. Then, a new event-based down-sampling method and transition filtering are designed for decreasing the computation time of optimization. At the last stage of the proposed algorithm, post-processing considering ON duration of appliances and varying states are proposed to increase the accuracy of the power profile reconstruction. The proposed approach was successfully tested for the low-frequency dataset of a house from the REDD. Numerical results show the advantages of the proposed algorithm, marked improvement over classification-based NILM considering small training dataset and its applicability in disaggregating the power consumption measured by the smart meter.

A Piecewise Probabilistic Harmonic Model of Aggregate Residential Loads

Harmonics produced by residential loads have become an increasing concern in recent years due to the extensive adoption of power-electronic-based home appliances. Harmonic modeling of aggregate residential loads (ARLs) is of great significance since it is the prerequisite for harmonic evaluation and necessary mitigation measures in the residential areas. In this paper, a piecewise probabilistic harmonic model is proposed to analyze the aggregate harmonics of residential loads. Based on the field measurement data, a unified dominant harmonic coupled matrix model (UD-HCMM) is established. This model directly exhibits the coupled relationship between the harmonic voltages and currents of residential loads. In view of the random and fluctuating harmonic characteristics of home appliances, the multi-Gaussian curve fitting method is used to fit the UD-HCMM parameters. Thereafter, the probabilistic UD-HCMMs are reconstructed using the Newton-Raphson method. Finally, a piecewise UD-HCMM is derived to characterize the time-varying harmonics of ARLs. The accuracy and validity of the proposed method are verified through simulations and field measurements. In addition, the proposed method is applied to evaluate the harmonics of a medium voltage distribution system with massive residential loads.

Non-Intrusive Appliance Identification with Appliance-Specific Networks

The problem of noninstrusive load monitoring (NILM) is usually formulated as a single-channel blind source separation task, whose successful solution enable fast and convenient load identification and energy disaggregation. When applied at test time, NILM algorithms aim to identify the operating characteristics of individual appliances from an aggregate power measurement of the entire house. Recent advances in deep learning gave rise to many methods that mostly focus on learning a direct mapping from aggregate measurement to individual appliance power. However, these methods are not only computationally expensive, but they often suffer from overfitting and do not generalize very well. In this article, we propose a novel NILM method that leverages advances in statistical learning that have not been properly applied in this domain before. The proposed method consists of three stages: first, a Bayesian nonparametric learning-based approach for appliance state extraction; second, synthetic minority oversampling technique for data augmentation and mitigating the heavy imbalance in switching events; and third, appliance-specific lightweight long short-term memory networks for status classification for each appliance. We adopt a “differential” input (the difference before and after the switching event) to reduce the complexity of network training and make the proposed method robust to multiappliance switching events. Experiments are conducted to demonstrate the effectiveness of the proposed method, achieving superior performance when compared to recent methods. An ablation study is conducted to demonstrate the effectiveness of each module of our method. Finally, we investigate the quality of generated synthetic samples.