KẾT HỢP THUẬT TOÁN MUSIC VÀ LMS TRÊN MẢNG MICROPHONE DẠNG CHỮ NHẬT PHỤC VỤ TRONG PHÒNG HỌP=COMBINING MUSIC AND LMS ALGORITHMS ON THE RECTANGULAR MICROPHONE ARRAY FOR MEETING ROOMS

0 Người đánh giá. Xếp hạng trung bình 0

Tác giả: Văn Sáng Đoàn, Mạnh Kha Hoàng, Thị Hòa Nguyễn, Văn Tùng Nguyễn, Xuân Thành Phạm

Ngôn ngữ: vie

Ký hiệu phân loại:

Thông tin xuất bản: Tạp chí Khoa học & Công nghệ - Trường Đại học Công nghiệp Hà Nội, 2024

Mô tả vật lý: tr.184

Bộ sưu tập: Metadata

ID: 484204

Thêm vào giỏ

Nghiên cứu này đề xuất mô hình microphone định hướng có cấu trúc mảng chữ nhật đồng nhất (Uniform Rectangular Array - URA) phục vụ trong phòng họp. Hoạt động của microphone dựa trên việc kết hợp thuật toán phân loại tín hiệu đa đường (Multiple Signal Classification - MUSIC) và thuật toán quân phương trung bình tối thiểu (Least Mean Squares - LMS), có thể theo dõi và điều khiển búp sóng theo hướng tín hiệu mong muốn, từ đó giảm công suất tiêu thụ, tiết kiệm năng lượng cho microphone. Qua mô phỏng đối với ba tín hiệu trong dải tần từ 1,3 kHz tới 1,4kHz, kết quả nhận được phù hợp với yêu cầu đề xuất. Khi so sánh hiệu suất đối với các thuật toán điều khiển búp sóng truyền thống (Conventional Beamforming - CB), đáp ứng không méo phương sai cực tiểu (Minimum Variance Distortionless Response - MVDR), MUSIC đạt tỷ số đỉnh trên trung bình (Peak-to-Average Power Ratio - PAPR) 50dB tại tỷ số tín hiệu trên nhiễu (Signal-to-Noise Ratio - SNR) 36dB trong khi MVDR và CB lần lượt là 36dB và 10dB. Trước bão hòa, lỗi trung bình căn quân phương (Root Mean Square Error - RMSE) của MUSIC là 1,3 tại SNR = 0dB thấp nhất so với MVDR 1,9 và CB 5,1. Sau bão hòa, MVDR chạm tới mức lỗi của MUSIC trong khi CB vẫn trên 5. Thuật toán LMS tuy có thời gian hội tụ chậm hơn thuật toán quân phương tối thiểu đệ quy (Recurisve Least Squares - RLS) khoảng 0,06s, nhưng sau đó không có sự chênh lệch về biên độ tín hiệu.This study presents a beamforming microphone that utilizes a uniform rectangular array (URA) structure specifically designed for meeting rooms. The microphone operates by combining the multiple signal classification (MUSIC) algorithm and the least mean squares (LMS) algorithm to effectively track and direct the beam towards the desired signal direction. This results in reduced power consumption and energy savings for the microphone. The results achieved through simulation for three signals within the frequency range of 1.3kHz to 1.4kHz satisfy the specified criteria. When comparing the performance of MUSIC with the conventional beamforming (CB) algorithm and minimum variance distortionless response (MVDR) algorithm, the MUSIC algorithm achieved a peak-to-average power ratio (PAPR) of 50dB at a signal-to-noise ratio (SNR) of 36dB. In contrast, the MVDR and the CB achieved PAPRs of 36dB and 10dB, respectively. Prior to reaching saturation, the root mean square error (RMSE) of the MUSIC algorithm was 1.3° at the SNR of 0dB, which was the smallest compared to the MVDR algorithm at 1.9° and the CB algorithm at 5.1°. Following the point of saturation, the MVDR algorithm achieved the same level of error as the MUSIC algorithm. However, the CB algorithm remained higher than 5 degrees. Despite the LMS algorithm's slower convergence time of approximately 0.06s in comparison to the Recursive Least Squares (RLS) algorithm, there is no difference in error thereafter.

1. Đọc trực tuyến

1. Đọc trực tuyến

Tạo bộ sưu tập với mã QR