Data analysis is playing a major role in the transition of the conventional grid to the smart grid. It contributes to almost all the fields of power systems; however, there is much untapped potential in the residential sector, especially demand side management. For planning and managing demand-side clustering, load profile clustering is a very critical and sensitive subject. Therefore, this paper presents a complete framework for clustering residential load profiles at the household level non-intrusively. The framework first preprocesses the datasets and than clusters the similar consumption patterns. In preprocessing, this paper proposes a novel statistic-backed machine learning technique for outlier detection. Finally, compares two major clustering techniques and evaluates the results. The framework is successfully tested with an open-source dataset of 200 houses provided by the US Department (NREL).

This paper introduces a comprehensive investigation into the realm of electric load profile analysis, shedding light on its pivotal aspects and far-reaching implications for effective electricity management. The study places a particular emphasis on essential components such as household consumption segmentation, the identification of peak demand instances, computation of the load factor, and the extraction of consumption trends. These elements collectively contribute to a deeper understanding of power system optimization, underlining their significance in shaping strategies for efficient resource allocation and consumption control. By harnessing the capabilities of Google Colab, a cloud-based notebook environment built upon the Jupyter platform, this research aims to empower both data scientists and machine learning developers. This collaborative digital space facilitates seamless Python code execution while offering a shared platform for real-time cooperation, thus enhancing productivity and streamlining research efforts in load profile analysis.

Non-intrusive load monitoring (NILM) is a technique for disaggregating the total energy consumption of a building into individual appliance-level energy consumption. Event detection is a critical component of NILM systems as it involves the identification and classification of different electrical events from the aggregate power signal. In this article an event detection method for NILM systems has been proposed that is based on the analysis of the statistical properties of the aggregate power signal. Specifically, we use a sliding window approach and K-Means clustering to detect number of devices from the power signal and then apply a threshold-based algorithm to detect electrical events. We evaluate the proposed method on a public dataset and demonstrate its effectiveness in accurately detecting electrical events. The proposed method has the potential to improve the accuracy with recall of 98.84% carried out on Pecan Street Datanort Inc.

Non-Intrusive load monitoring (NILM) is an emerging technology for extracting potent information from a consumer’s electric load profile. The NILM techniques are gaining popularity among researchers as they reduce the requirement for external hardware for load monitoring. These techniques require only the smart meter’s data, thus reducing the need for other measuring and sensing devices. In this paper, the research status of non-intrusive load monitoring has been reviewed. The utilisation of NILM techniques for event detection, load monitoring and energy disaggregation has been discussed in this paper. Furthermore, the research gap and future aspects are discussed. This paper has a great utility factor for upcoming researchers interested in NILM techniques.

Dynamic pricing is a pricing technique in which companies set extremely flexible costs for products or services based on customer demands. Dynamic pricing is famous because of its capacity to boost a company’s revenue. Day-ahead electricity pricing is an important technique for producers of electricity, by which the grid stability can be improved by energy procurement price. In this paper, a novel technique of demand-side management (DSM) has been used to design individual price policies, where each and every end customer receives a separate electricity pricing scheme. This is designed to incentivize demand management in order to optimally manage flexible demands. A general artificial neural network-based stochastic process for consumer’s power demand has been made to minimize the mean electricity price paid by the users. The projected pricing in Manchester’s ISO New England market was calculated using MATLAB Software’s ANN fitting tool. Hourly historical data of temperature, electrical load, and natural gas price from the ISO New England market were utilized in the forecasts. The artificial neural network (ANN) was trained on hourly data from 2004 to 2007 and evaluated on out-of-sample data in 2008. The simulation results revealed extremely accurate day-ahead estimates with very small amount of price forecasting error.

The concept of data analysis exit in the literature theoretically for years. However, rapidly increasing technology has facilitated the practical implementation also. However, the data science field has excelled to its limits in various fields, but still, a lot of scope for research exists in data analysis for a power system. As each field has salient features, the power system also has its own salient features and complexities. One of such complexity is dealt with in this paper, i.e., outliers detection from residential customers' electrical load profile. Due to salient features of residential customers' load profiles, most of the algorithm results in false detection. In this paper, a novel approach to detect outliers dedicated to residential customers load profiles is proposed. After outlier detection, the clustering of different consumer's profiles is done using k-means clustering. Each cluster is represented with a class representative profile to reduce the computation burden of the aggregator and extract better high-grade information from it.

In recent year, growth in electric power demand, electricity production cost and emission of greenhouse gases have increased rapidly. As a result, a special emphasis on distributed generation is needed. The best way to solve the problem is to develop distributed generation, primarily from renewable energy resources. These tools are placed close to the residence in order to produce a few kilowatts. Besides several benefits, there are major challenges in its management and optimal use. These challenges have become one of the main concerns for researchers. In this paper, an innovative model is proposed for strategic energy management to facilitate demand response. Its aim is to improve the efficiency of households that include generation units such as wind turbines, solar panels, storage units and uncontrollable or controllable loads. This optimization management's main goal is to maximize micro-grid profitability for 24 h a day. The cumulative effect of this model shows that the benefit of the micro-grid has greatly improved.

Load forecasting refers to the prediction of load behavior for the future. As the accurate forecasting is the foundation for various planning, programs, etc. Therefore, there is a necessity to increase the accuracy of the forecasted load. In this work, firstly a day ahead, hourly load is forecasted. A polynomial regression method has been proposed and is compared with linear regression. The error associated with the forecasted load dataset is evaluated. Then, the anomaly or the fallacy in the calculated load dataset is evaluated. An anomaly detection equation using statistical parameters has also been proposed.

This paper presents a literature review of scheduling techniques and communication protocols (STCP), which make adaptable smart home (SH) capable of implementing demand response. In addition, it presents advantages and disadvantages of various STCP mentioned in the literature. Also, the paper provides the pathway to future researchers for designing of smart homes, as it briefly mentions various specifications of STCP so that one could select according to their application.