國立虎尾科技大學 |

Language: English

Back

人工智慧聲音識別深度學習應用於脈衝渦電流非接觸式檢測系統 = = Pul...

林棋桀

人工智慧聲音識別深度學習應用於脈衝渦電流非接觸式檢測系統 = = Pulsed Eddy Current Non-Contact Detection System by Using AI Deep Learning Sound Recognition /

Record Type:	Language materials, printed : Monograph/item
Title/Author:	人工智慧聲音識別深度學習應用於脈衝渦電流非接觸式檢測系統 =/ 林棋桀.
Reminder of title:	Pulsed Eddy Current Non-Contact Detection System by Using AI Deep Learning Sound Recognition /
remainder title:	Pulsed Eddy Current Non-Contact Detection System by Using AI Deep Learning Sound Recognition.
Author:	林棋桀
Published:	雲林縣 :國立虎尾科技大學 , : 民113.07.,
Description:	[10], 79面 :圖, 表 ; : 30公分.;
Notes:	指導教授: 陳志維.
Subject:	Object Detection. -
Online resource:	電子資源

人工智慧聲音識別深度學習應用於脈衝渦電流非接觸式檢測系統 = = Pulsed Eddy Current Non-Contact Detection System by Using AI Deep Learning Sound Recognition /
林棋桀

人工智慧聲音識別深度學習應用於脈衝渦電流非接觸式檢測系統 =Pulsed Eddy Current Non-Contact Detection System by Using AI Deep Learning Sound Recognition /Pulsed Eddy Current Non-Contact Detection System by Using AI Deep Learning Sound Recognition.林棋桀. - 初版. - 雲林縣 :國立虎尾科技大學 ,民113.07. - [10], 79面 :圖, 表 ;30公分.

指導教授: 陳志維.

碩士論文--國立虎尾科技大學車輛工程系碩士班.

含參考書目.

本論文探討利用人工智慧（Artificial Intelligence, AI）深度學習技術進行聲音辨識，結合渦電流技術和有限元素分析，處理和分析待測物感應出的聲音數據。研究利用線圈脈衝電流產生脈衝電磁場，使待測物產生一個渦電流，線圈磁場與待測物所形成之磁場交互作用產生一個電磁作用力，進而讓待測物產生震動形成聲音。運用麥克風擷取該聲音並進行資料收集和預處理如特徵提取、聲音拆分等等。隨後將資料輸入至卷積神經網路(Convolutional Neural Network, CNN)之中訓練得到理想模型。訓練過程中運用反向傳播算法來不斷優化模型，以提高其準確性和泛化能力，並利用學習率調整和正則化等技術，以避免過擬合和提高模型的穩定性。實驗將兩者不同技術實現並利用人機介面進行整合，使待測物進行渦電流檢測後的辨識結果呈現於介面上，達成聲音辨識結合脈衝渦電進行非接觸式及時檢測的目標。.

(平裝)Subjects--Topical Terms:

1151221
Object Detection.

人工智慧聲音識別深度學習應用於脈衝渦電流非接觸式檢測系統 = = Pulsed Eddy Current Non-Contact Detection System by Using AI Deep Learning Sound Recognition /
LDR:03544cam a2200241 i 4500 001 1129991
008 241015s2024 ch ak erm 000 0 chi d
035 $a (THES)112NYPI0162004
040 $a NFU $b chi $c NFU $e CCR
041 0 # $a chi $b chi $b eng
084 $a 008.158M $b 4442:2 113 $2 ncsclt
100 1 $a 林棋桀 $3 1449027
245 1 0 $a 人工智慧聲音識別深度學習應用於脈衝渦電流非接觸式檢測系統 = $b Pulsed Eddy Current Non-Contact Detection System by Using AI Deep Learning Sound Recognition / $c 林棋桀.
246 1 1 $a Pulsed Eddy Current Non-Contact Detection System by Using AI Deep Learning Sound Recognition.
250 $a 初版.
260 # $a 雲林縣 : $b 國立虎尾科技大學 , $c 民113.07.
300 $a [10], 79面 : $b 圖, 表 ; $c 30公分.
500 $a 指導教授: 陳志維.
500 $a 學年度: 112.
502 $a 碩士論文--國立虎尾科技大學車輛工程系碩士班.
504 $a 含參考書目.
520 3 $a 本論文探討利用人工智慧（Artificial Intelligence, AI）深度學習技術進行聲音辨識，結合渦電流技術和有限元素分析，處理和分析待測物感應出的聲音數據。研究利用線圈脈衝電流產生脈衝電磁場，使待測物產生一個渦電流，線圈磁場與待測物所形成之磁場交互作用產生一個電磁作用力，進而讓待測物產生震動形成聲音。運用麥克風擷取該聲音並進行資料收集和預處理如特徵提取、聲音拆分等等。隨後將資料輸入至卷積神經網路(Convolutional Neural Network, CNN)之中訓練得到理想模型。訓練過程中運用反向傳播算法來不斷優化模型，以提高其準確性和泛化能力，並利用學習率調整和正則化等技術，以避免過擬合和提高模型的穩定性。實驗將兩者不同技術實現並利用人機介面進行整合，使待測物進行渦電流檢測後的辨識結果呈現於介面上，達成聲音辨識結合脈衝渦電進行非接觸式及時檢測的目標。.
520 3 $a Artificial Intelligence (AI) deep learning techniques for speech recognition, combined with eddy current technology and finite element analysis to process and analyze sound data induced by test objects is proposed in this thesis. The research aims to improve the accuracy and application range of sound recognition, and to study the potential of eddy current technology in non-destructive testing. The goal of the study is to develop a sound recognition system based on deep learning and to use eddy current technology for material inspection, achieving efficient and accurate detection of material defects. Pulse currents is used to generate pulsed electromagnetic fields, causing the test object to produce an eddy current. The interaction between the coil's magnetic field and the magnetic field generated by the test object creates an electromagnetic force, which in turn causes the test object to vibrate and produce sound. Sound is captured by a microphone for data collection to undergoes preprocessing, such as feature extraction and sound segmentation, to prepare for training, which is then for Convolutional Neural Network (CNN) to train and obtain the optimized model. During the training process, backpropagation algorithm is used to continuously optimize the model to increase its accuracy and generalization capabilities. Additionally, various optimization techniques, such as learning rate adjustment and regularization, are employed to prevent overfitting and enhance the model's robustness..
563 $a (平裝)
650 # 4 $a Object Detection. $3 1151221
650 # 4 $a Sound Recognition. $3 1383716
650 # 4 $a Surface Defects. $3 1247438
650 # 4 $a Deep learning. $3 1127428
650 # 4 $a 深度學習. $3 1127425
650 # 4 $a 表面瑕疵. $3 1383719
650 # 4 $a 聲音辨識. $3 1383713
650 # 4 $a 非接觸式檢測. $3 1450721
650 # 4 $a 渦電流. $3 1450720
856 7 # $u https://handle.ncl.edu.tw/11296/g6crnb $z 電子資源 $2 http