Eksplorasi sistem penghapus bising lalu lintas secara adaptif untuk ruang kerja

PRASETYOWATI, Sri Arttini Dwi, Promotor Prof. Adhi Susanto, M.Sc., Ph.D

2010 | Disertasi | S3 Ilmu Teknik

Permasalahan penghapusan isyarat bising didasarkan pada watak khas sumbernya. Maka representasi watak bising harus ditemukan sebelum diupayakan peredamannya. Kesesuaian skema untuk permasalahan yang dihadapi pada kebisingan dari kendaraan di jalan raya mutlak diperlukan. Karena berbagai macam kendaraan silih berganti menjadi sumber bising, proses peghapusan harus secara adaptif selalu mengikuti watak bising yang berubahubah. Secara mendasar algoritma LMS (Least Mean Square) menjadi pendekatan atau pemecahan penghapusan bising yang dilandasi galat kuadrat rerata minimum, disamping yang lain, yaitu NLMS (Normalized LMS) and RLS (RecursiveLeast Square). Hasil penelitian menunjukkan nilai-nilai langkah (μ ), panjang filter (L), dan tundaan waktu terbaik untuk masing-masing macam bising. Nilai-nilai optimal yang didapatkan adalah μ =0,001, L = 230, tundaan 100 sampel (0,002 detik).

The problems of noise cancelling systems are based on the specific characteristics of their sources. Therefore, the knowledge of each noise source should be represented prior to the excecution of the designed noise cancelling scheme. Since the noises emitted vary with the types of the passing vehicles, the scheme should be well adjusted automatically. Five representing types of noise producing vehicles were looked into throughly to design the needed adaptive scheme, which will follow the needed adaptive scheme, which will follow the everchanging traffic noise. The robust LMS (Least Mean Square) algorithm was applied first, then followed by the NMLS (Normalized LMS) and RLS (Recursive Least Square) ones. The results show the optimum step size (μ ), filter lengths (L), and time delays for each noise type. Their best values are μ =0,001, L = 230, the time delays = 100 sample (0,002 second).

Kata Kunci : LMS (Least mean square),Delay,Nilai langkah, Panjang filter,delay, step size, filter length.