A rule-based model for electricity theft prevention in advanced metering infrastructure.

The deployment of smart electricity meter (SEM) via the advanced metering infrastructure (AMI) has come under cyber-attacks as adversaries continue to exploit the communication links for possible evasion of electricity bill payments. Various detection models relying on energy consumption data offer a disadvantage of delayed detection and consequent huge financial losses before frauds are detected. Moreover, existing techniques mostly concentrate on detection of electricity thefts and rely on energy consumption data alone as the basis of theft perpetration whereas other potential parameters which could be exploited for electricity theft prevention exist in AMI. In this study, AMI parameters, which are indicative of electricity thefts are preselected and modelled for electricity theft prevention. First, a given AMI network is sectioned into zones with the selected parameters modelled to define security risks by formulated set of rules based on real-time scenarios. Fuzzy inference system is then employed to model the security risks to ascertain the compromised state of the monitored parameters at the defined scenarios. The result of the developed model at 50% weight of each of the modelled parameters with interdependencies show clear indications of the modelled parameters and their interactions in the determination of risks. The decisions on monitored parameters evaluated at every timestep reveal varied dense velocity behaviours for every scenario. The result is suitable for monitoring the AMI in reporting and/or disconnecting any compromised SEM within a considerable timestep before huge losses are incurred. Implementation of this scheme will contribute a significant success in the attempt to prevent electricity theft perpetration via the AMI. [ABSTRACT FROM AUTHOR]