ABSTRACT
Losses or electricity loss is one of the results of the application of the historical electric power reading system, AMR (Automatic Meter Reading). To detect the losses itself is still checking directly the data of each incoming customer. So, we need a system to make it easier to analyze and evaluate the data. The K-Means method was used in this study to cluster data based on historical use of electrical power and to determine the optimal number of groups used by the Davies-Bouldin Index (DBI) method. The representation of each process in the application will be designed using a flowchart. The Black-box method is used to test application functionality. Based on the results of testing, the K-Means method is able to cluster the historical electricity usage with the most optimal number of clusters, four groups based on the calculation of the DBI method. Losses or loss of electrical power in the system electric power distribution, which is usually used at certain times, is one measure that is efficient or not an electric power system operation. This AMR system can be used optimally for account issuance, customer load analysis, calculation of losses or distribution losses, and electricity network development planning. IoT with ad-hoc network capabilities can solve several problems. However, because they both rely on identity nodes to communicate with each other, they are both vulnerable to Sybil attack. Sybil attackers illegally change into several different identities to carry out various malicious activities such as damaging data aggregation, voting, and disrupting routing. Several defense machineries have been proposed for Sybil attacks on WANET, which are mostly based on cryptography, location/position, network behavior, resource testing, and trust. However, the drawbacks are that not all machinery are suitable for use in networks with limited resources. The research method used in this study is a qualitative approach using the case study method, which analyzes and calculates directly the causes of energy losses through AMR data analysis, such as meter reading errors, measurement abnormalities, and errors during wiring, which is a mistake based on non-technical losses. Besides that, K-Means also has a fairly high accuracy based on the size of objects, so this algorithm is relatively more measurable and efficient for processing large amounts of objects. The weaknesses in the K-Means algorithm are analyzing and determining the best number of k in clustering data in a dataset. To get the optimal k value, the author uses the Davies–Bouldin Index (DBI) method. The minimum DB Index value is the most optimum clustering scheme. In this chapter, we present a survey, classification, and comparison of various defense machinery that have been proposed for non-IoT WANETs. We emphasize the issue of how the advantages and disadvantages of this defense mechanism when applied to the IoT infrastructure can effectively recognize properties of Sybil attack.
