隨著科技不斷的進步，日常生活中各種產品逐漸朝向小體積、高精度以及節能等方向發展，由於零件與結構的微型化可以提昇產品效能及縮小配置空間，因此如何使零件與結構微型化慢慢成為新的技術發展方向，而微細加工技術在微型技術發展過程中佔有相當重要的地位。在切削加工過程中，刀具的狀態對於整體切削的品質與成本的影響很大，所以開發刀具狀態偵測系統有其必要性。由於在切削加工過程中，經驗豐富的技師常能藉由切削的聲音得知刀具的狀態，然而應用聲音建構自動化刀具狀態偵測系統容易因背景噪音的干擾而造成系統之誤動作，因此為了提升偵測系統的穩定性，抑制噪音技術扮演著重要的角色。 本研究探討環狀排列麥克風陣列在降低干擾噪音之效能，以及對於刀具磨耗偵測系統的影響。實驗過程採用直徑700μm之微細銑刀，工件為SK2 高碳鋼，並在切削過程以喇叭提供人工噪音源，並利用麥克風陣列擷取切削時的聲音訊號。擷取之聲音訊號接續以單支麥克風偉納濾波系統、麥克風陣列濾波系統，以及整合陣列與偉納濾波系統分別進行濾波處理，分析不同系統之噪音濾除效能。在分析各濾波系統對刀具磨耗辨識之性能影響方面，結果顯示整合環列麥克風陣列與偉納濾波系統較其他濾波系統能更有效的去除雜訊，達到100%的辨識成功率。此外，使用高頻麥克風所接收之聲音訊號，因其於高頻部份的訊號不易受到外在干擾噪音的影響，在高能量噪音之干擾之下，雖無導入噪音去除系統，系統還能有效的偵測刀具磨秏之狀態。 Micro machining draws a much attention lately due to the increasing need in miniaturized devices. At the same time, the tool condition monitoring plays an important role in improving the cutting quality and efficiency in micro-mechanical machining. In tradition, an experienced technician can detect the tool condition easily by the sound generated during cutting. However, the sound signal contaminated by the background noise will make the sound based monitoring system not reliable as expected. Therefore, how to reduce the noise effect play a crucial role in developing the sound based tool condition monitoring system for practical applications. This research focus on the study of the capability of microphone array on the noise reduction and the improvement of sound based tool wear monitoring. Two arrangements for the installation of microphone arrays were discussed in this thesis. In experimental setup, 700um micro end mill was used in cutting SK2 workpiece. To simulate the noise occurring during factory, the speaker was installed inside the cutting chamber to generate the broad band noise during cutting. After sound signals were collected by microphone during cutting with various tool wear conditions, they were processed by system with the microphone array filter, the Wiener filter, and the combination of the microphone array filter and the Wiener filter, respectively. The results show that 100% classification rate can be obtained by processing the sound signal with the combination of microphone array and the Wiener filter. In the study of adopting the high frequency microphone with bandwidth up to 80 kHz, the results show that the system without the implementation of noise reduction algorithm still can detect the tool wear effectively due to the features higher than 20 kHz was not deteriorated by the generated noise.