Determining a set of suspicious electricity customers using statistical ACL Tukey’s control charts method

• Application of ACL Tukey’s control chart to estimate the number of suspects is given.
• Theoretical foundation is balance of total, invoiced and non-technical losses energy.
• Point of paper is indicator’s estimation of suspicious customers in boundary region.
• Series of all customers with the same limits are observed simultaneously.
• Observation’s distribution of suspicious customers is mostly right.

Estimation a set of suspicious electricity customers is the first stage for detecting fraud of electricity. This paper deals with this stage. Based on historical data about monthly measurements of electricity, customers’ time series are formed and then analysis of these series is performed using Tukey’s control charts as one of the statistical method. The asymmetric control limit (ACL) Tukey’s control charts are chosen due to a rather expressed asymmetry and dominantly presented right distribution of time series data. As usual, the choice of upper (UCL) and lower (LCL) control limits is not based on allowed number of observations outside of the control limits. The choice of these limits is based on the balance of total energy, registered energy and energy losses. The essence of this approach is the simultaneous observation all customers’ time series of the controlled set with the same control limits and the same percentage of total and normalized energy losses in the observed distribution network. The criterion for finding the number of suspicious customers and their addresses is allowed error between the number of registered customers with one or more data (observations) outside the control limits for given losses and the number of suspicious customers’ indicators based on certain balance of energy and the estimated total percentage of losses in distribution network.