Data based Reliability Assessment of 11 kV distribution Feeder using A- LSTM and FMEA

Reliability assessment is critical for planning and operating power distribution systems, as frequent outages cause major economic and technological losses, especially for industrial and commercial consumers. To evaluate the distribution system reliability, this study employs an 11 kV/415V Srinivasamangalam feeder from the 132/33 kV Tirupati Substation, Andhra Pradesh, India. An Attention-based Long Short-Term Memory (A-LSTM) neural network and Failure Mode and Effects Analysis (FMEA) are applied for 11 KV Srinivasamangalam feeder for enhancement of feeder performance. Under various scenarios of operations like feeder reconfiguration, distributed energy resource (DER) allocation, including photovoltaic (PV) systems and electric vehicles (EVs) and fault isolation were carried out for reliability assessment. Based on failure data available at 11kV Srinivasamangalam substation for the years 2021 to 2023, FMEA technique has identified critical nodes for DER placement, voltage stability and supply continuity. The implementation of A-LSTM-based feeder reconfiguration reliability indices were significantly improved by reduced SAIDI, CAIDI, and SAIFI while raising ASAI and voltage stability. A-LSTM enables to handle various fault events, restoration sequences, and reconfiguration actions over a period of one year and provides predictive reliability assessment that extends beyond the static, event-specific scope of FMEA. Thus, the suggested A-LSTM and FMEA based reliability assessment provided a data-based approach for resilient and sustainable power distribution system operation.