||Distribution State Estimation (DSE) is one of the important elements in Distribution Management System (DMS). DSE can provide a near real time system model that can be used to enable DMS functions e.g., volt/var control, distribution automation, and service restoration. Based on the data obtained from Distribution State Estimation with Bad Data Detection (DSE-BDD) result, distribution system operators can monitor the system states and make informed decisions on load adjustment, voltage control, and feeder reconfiguration. |
However, failures in measuring devices or telemetry equipment, delays in the data transmission, and energy theft could result in bad data and lead to inaccurate DSE solutions. Nowadays, distribution system states have become more dynamic due to the integration of intermittent distributed generation (DG). Therefore, bad data issue becomes more complicated. Distribution system may contain single or multiple bad data. In some cases, multiple bad data are interacting with each other which makes it difficult to detect.
In order to solve the bad data problem in DSE, this study proposes a bad data detection procedure. It uses the information in a residual sensitivity matrix to detect possible bad data. Single and multiple non-interacting bad data are identified by using the largest normalized residual (LNR) information. Multiple interacting bad data are identified by using the combination of LNR and residual sensitivity information to build an intracting measurement table and to detect the suspected bad data. A genetic algorithm is used to identify the bad data in the suspected set.