Written by Ishan Srivastava, Abhishek Kumar, and B V Surya Vardhan
Residential energy demand in India has been steadily increasing, with the primary causes being a combination of factors including population growth and rapid economic and technological development. The control and monitoring of the power distribution network is an increasingly complicated task as the complexity of the power grid rises. Smart energy management therefore, becomes one of the key components of transforming a traditional city into a smart city. This entails incorporating more renewable energy sources into the system, installing smart lighting fixtures, placing sensors in the grid, putting Advanced Metering Infrastructure (AMI) in place, and deploying smart meters at load centers. The two-way communication system needed for AMI poses a number of difficulties, including the development of an upgrade plan that is financially viable and management of cyber-security issues. Additionally, AMI implementation yields a number of advantages that make grid monitoring and control processes simpler. AMI is needed in the deployment of the majority of contemporary methodologies for detecting and preventing electricity theft. This article presents a variety of challenges and advantages of using AMI in the smart grid.
The efficient use of energy and on-time bill payment are core aspects of energy management in smart cities. The development of the Advanced Metering Infrastructure (AMI) is one of the key factors for smart energy management in a smart city. In modern electricity distribution, smart meters are an essential part of the Advanced Metering Infrastructure (AMI). The procedure necessary to convert the current metering infrastructure to AMI can be very expensive. However, the overall benefits of AMI make it a cost-effective strategy. AMI is fully automated and enables real-time monitoring of energy use. AMI tracks not only how much power is utilized, but also when it is consumed. An AMI system typically consists of smart meters, a customer gateway, a communication network, and head-end. Smart meters primarily convey and record data pertaining to energy. The customer gateway facilitates communication between customer systems and appliances and the AMI network. An AMI head-end handles information exchanges between external systems like the Meter Data Management (MDM) system and the AMI network . The data collected from smart meters not only helps in billing processes but also acts as a crucial component of demand-side management by providing the load profile to the operator, which can be further used in load forecasting. This data can also be used in implementing the dynamic tariff strategy.
Challenges and Benefits of AMI Systems
For an end user, different types of information such as load profile, voltage profile, outage logs, energy usage, peak KW demand, power factor, etc., can be provided by AMI systems.
Challenges: Despite its numerous advantages, implementing AMI poses three significant challenges. AMI systems necessitate the installation of all necessary hardware and software, comprising of smart meters, two-way communication infrastructure, and network management software. Another difficult challenge is to integrate the AMI system with other utility-level systems such as outage management systems, information technology systems, distribution automation systems, metering and billing systems, and so on.
Benefits: AMI can facilitate strategies like outage management, billing, load forecasting, deployment of distribution automation, and remote service connection and disconnection. AMI technology enables electricity distributors to recognize and automatically respond to electric demand, which in turn reduces power outages and hence, the reliability of the system. As a result of this improved reliability and proper billing operation through AMI, the generated revenue of the utility increases, which results in lower electricity rates for the customer. Besides that, the AMI system is useful in combating the issue of power theft, which is rampant in developing nations like India. Greater operational transparency will result from the use of AMI systems that track energy consumption in almost real-time. AMI assists with meter-related concerns such as tampering, swapping, and theft. It also detects billing irregularities and other forms of power theft .
Methodologies of Electricity Theft Detection and Prevention
Power utilities suffer financial losses as a result of electricity theft across the energy sector. It occurs when electricity is used without being paid. The majority of these methods rely exclusively on records of electricity consumption. However, due to the variety of theft strategies (line tapping, meter tampering, etc.), it is difficult to detect fraudulent consumers by simply observing electricity consumption records.
The energy theft detection methodologies comprise3: anomaly or outlier detection - the identification of any unusual activity at the customer can be a case of energy theft; network analysis method, which is governed by detection of non-technical losses in the system; and implementing cyber-security or intrusion detection systems .
Researchers have applied machine learning techniques to theft detection applications. Additionally, machine learning techniques are used in anomaly detection, tracking non-technical losses, and act as a foundation for an intrusion detection system in some cases.
By virtue of constant monitoring capability, the risk of electricity theft can be reduced using the AMI system. Customers' consumption follows a statistical pattern under normal circumstances; deviations from this pattern may indicate malicious activity. In case of electricity theft, the operator can quickly spot the anomaly using the end user's power consumption trend and hence, the spot of power theft can be identified. But the AMI deployment itself opens up certain new avenues for power fraud, such as the two-way communication system's susceptibility to cyber-attacks .
The utilization of data analysis techniques like data mining, machine learning, and deep learning can be cost-effective due to the availability of data through the AMI system. The classifier is then used to search for unusual patterns. In some of the recent work proposed for the solution of electricity theft detection, an ensemble machine learning model is proposed6, which is based on customer consumption patterns. Deep learning based methods are also reported in a journal where a Convolution Neural Network (CNN) based scheme is proposed for theft detection in smart grid. A real data set of 5,000 customers is used , this work proposed a Consumption Pattern Based Electricity Theft Detector (CPBETD) which uses SVM anomaly detector and silhouette plots to identify the different distributions in the dataset for theft detection. This method employs distribution transformer meters to detect non-technical losses at the transformer level. Using big data analytics, researchers devised a different technique to address the problem of electricity theft.
There are certain challenges while using a machine learning classifier for theft detection like data imbalance, i.e. there is difference in range for the normal and abnormal data samples, it is difficult to train the machine learning model since theft data samples are generally not available.
A work is proposed that employs some indicative parameters that can be used by AMI to detect electricity theft . Positive intrusion status, false signature on SEM data, erratic outage notification, de-energized or SEM outage, timestamp status, false pricing, flagged observer meter status, and so on, are examples of these parameters.
There have been some recent patent applications proposing various methods for detecting electricity theft. The significant proportion of these inventions includes an intelligent unit or a smart device for detecting anomalies, which is coupled to a memory and processing unit. Researchers also proposed a larceny prevention system which consists of a microprocessor based intelligent ammeter, different sensors, and intelligent electric meter, which stores historical data of customer power consumption. The researchers have already implemented deep learning methods to solve electricity theft problems. Some researchers also proposed an anti-electricity theft smart meter which has an electric quantity measuring unit for monitoring the input and output of electric power. The user’s encrypted data is also tested using an anomaly detection model based on full homomorphism. Apart from these, several other inventions that use transformer data for theft detection have also been reported. A bibliographic patent trend (based on applicants’ country) is shown in Figure 1, which indicates that many innovations have been reported to solve the electricity theft problem.
Figure 1: Applicants’ country-based patents on electricity theft problem.
Even though several norms and regulations exist in India to prevent electricity theft, incidences of energy theft still occur on a daily basis. AMI implementation can potentially provide a solution for electricity theft prevention. The infrastructure for conventional metering is being upgraded by the Indian government. As part of this campaign, the energy management system in India’s smart cities includes not only the integration of renewables into the grid but also the installation of smart meters at the customers’ end. According to the National Smart Grid Mission by Government of India, as of November 2022, a total of 51,32,368 smart meters are already installed in across and 1,12,75,739 are sanctioned smart meters. In the near future, with the new techniques proposed in the field of theft detection, we can expect a drastic reduction of non-technical losses in the modern power distribution network.
- Otuoze, A.O., Mustafa, M.W., Mohammed, O.O., Saeed, M.S., Surajudeen-Bakinde, N.T. and Salisu, S., “Electricity Theft Detection by Sources of Threats for Smart City Planning,” IET Smart Cities, vol. 1, issue 2, pp. 52-60, 2019.
- Otuoze, A.O., Mustafa, M.W., Abioye, A.E. et al., “A Rule-Based Model for Electricity Theft Prevention in Advanced Metering Infrastructure,” Journal of Electrical Systems and Inf Technology, 9, 2, 2022.
- Gerasopoulos, S.I., Manousakis, N.M. & Psomopoulos, C.S., “Smart Metering in EU and The Energy Theft Problem,” Energy Efficiency, 15, 12, 2022.
- S. McLaughlin, B. Holbert, A. Fawaz, R. Berthier and S. Zonouz, “A Multi-Sensor Energy Theft Detection Framework for Advanced Metering Infrastructures,” IEEE Journal on Selected Areas in Communications, vol. 31, no. 7, pp. 1319-1330, July 2013.
- P. Jokar, N. Arianpoo and V. C. M. Leung, “Electricity Theft Detection in AMI Using Customers’ Consumption Patterns,” IEEE Transactions on Smart Grid, vol. 7, no. 1, pp. 216-226, Jan. 2016.
- Sravan. K. Gunturi, and Dipu Sarkar, “Ensemble Machine Learning Models for The Detection of Energy Theft,” Electric Power Systems Research, vol. 192, 106904, March 2021.
This article was edited by Sidharth Sabyasachi.