查詢結果分析
來源資料
相關文獻
- 基於累積能量譜之水下目標自動辨識演算法
- 快速時頻階次譜於機械故障診斷
- Neural Network Procedures for Taguchi's Dynamic Problems
- A Fast and Efficient Competitive Learning Design Algorithm Based on Weight Vector Training in Transform Domain
- 專家系統振動訊號圖型判別之研究
- 反傳遞模糊類神經網路於流量推估之應用
- 類神經網路(Neural Networks)的種類及其在影像處理上的應用
- C++Fuzzy類神經網路物件導向發展系統之建立
- 臺灣汽保費率之估計--對數線性費率模式與類神經網路之比較
- 運用類神經網路於股價指數之套利--以日經225指數為例
頁籤選單縮合
題 名 | 基於累積能量譜之水下目標自動辨識演算法=Automatic Classification of Underwater Objects Based on Accumulated Energy Spectrum |
---|---|
作 者 | 鄭又銘; | 書刊名 | 新新科技年刊 |
卷 期 | 11 2015.01[民104.01] |
頁 次 | 頁204-214 |
分類號 | 595.2 |
關鍵詞 | 累積能量譜; 聲信號特徵; 短時傅立葉轉換; 類神經網路; Accumulated energy spectrum; Signal features; Short-Time fourier transform; Artificial neural network; |
語 文 | 中文(Chinese) |
中文摘要 | 本文提出由水下目標聲信號累積能量譜擷取特徵資訊之方法,並以此為基礎發展水下目標自動辨識演算法。本文方法可一併處理穩暫態的聲信號,以降低發展辨識系統時的複雜度。時域聲信號經短時傅立葉轉換後得到離散功率譜密度函數,經離散積分後可得到累積能量譜,再藉多項式迴歸分析得到連續譜趨勢,將之分離並提取譜值大者作為聲信號特徵資訊。本研究採用倒傳遞類神經網路架構發展水下目標自動辨識演算法,則採用此特徵資訊作為輸入值。演算法經訓練後,於驗證過程和白色噪音環境下之測試分析均有相當好的辨識表現,運算效率亦極高。目前研究結果顯示本文提出之累積能量譜法所截取之水下目標聲信號特徵具有代表性,並可應用於自動辨識演算。 |
英文摘要 | A general method is proposed in the present study to draw characteristic features from the acoustical signals of underwater objects in time domain, upon which automatic classification algorithms are developed. With the advantage of being capable of handling both periodically steady and unsteady acoustical signals, the present method is based on the analysis of accumulated energy spectrum and effectively reduces complexities in designing classification system. Applying the Short-Time Fourier Transform, the discrete power spectral density is obtained of an acoustical signal, and upon integration of that with respect to time an accumulated energy spectrum is immediately established. Characteristics of the acoustical signal could be decided by picking the largest components of the spectrum with approximated trend being removed. With the so drawn signal features as the inputs, two nack-propagation artificial neural network models are developed for automatic classification of underwater objects. In terms of accuracy and operation speed, both models exhibit excellent capabilities in verifications, as well as in tests concerning white noise environments. This study suggests the present method of accumulated energy spectrum is able to draw significant features from underwater acoustical signals, which could be effectively used for the purpose of objects classification. |
本系統中英文摘要資訊取自各篇刊載內容。