Distribution Transformer Secondary Bushing Temperature Detection Device using Feed Forward Neural Network

Sari

ABSTRAK

Transformator distribusi mengubah tegangan listrik tinggi menjadi rendah. Pada sekunder transformator, tegangan dan arus listrik cukup besar sehingga terjadi disipasi panas berlebihan karena timbulnya tahanan listrik pada titik koneksi sekunder transformator dengan kabel keluaran. Hal ini menyebabkan unbalance current dan overheat sehingga terjadi lost contact yang mengganggu pasokan listrik serta drop tegangan. Sayangnya pemeriksaan di lapangan dilakukan tiap enam bulan sekali padahal lost contact dapat terjadi sewaktu-waktu. Sehingga kami mengusulkan pengembangan alat deteksi overheat real-time pada bushing sekunder menggunakan metode klasifikasi suhu berbasis Feed Forward Neural Network (FFNN) yang dilengkapi dengan Internet of Things. FFNN berhasil mengklasifikasikan suhu dengan nilai 0 untuk suhu 30˚C-50˚C, nilai 0 untuk suhu 51˚C-90˚C yang memerlukan perbaikan, dan nilai satu untuk suhu di atas 90˚C dengan relay memutus, kemudian sistem mengirimkan notifikasi lost contact realtime. Sehingga alat ini meningkatkan keefektifan pemeriksaan dan dapat diterapkan guna mengurangi tindakan pemeriksaan secara langsung.

Kata kunci: Transformator Distribusi, Lost Contact, Internet of Things, Feed Forward Neural Network

 

ABSTRACT

The distribution transformer turns high voltage into low voltage. On the secondary transformator, the voltage and current are sufficiently large that excessive heat dissipation occurs due to the appearance of electric retention at the point of secondary connection of the transformator to the output cable. This causes current imbalance and overheating, resulting in lost contact that disrupts power supply and voltage drop. Unfortunately, field inspections are carried out every six months and lost contact can occur at any time. So we suggested developing a real-time overheat detection tool on secondary bushing using a temperature classification method based on the Feed Forward Neural Network (FFNN) equipped with the Internet of Things. With FFNN, the system successfully classifies the temperature with a value of 0 for a temperature of 30 ̊ C-50 ̊ C, a value 0 for the temperature of 51°C-90 ̊ C that requires repair, and a value 1 for a temperatur above 90 ̊ C with a relay disconnect, then the system sends a real-time lost contact notification. Thus this tool increases the effectiveness of inspection and can be applied to reduce inspection actions directly.

Keywords: Distribution Transformer, Lost Contact, Internet of Things, Feed Forward Neural Network