本論文主要是針對馬達在改變轉速時的旋轉故障偵測,研製一個智慧型故障診斷系統。此系統透過無線感測器節點量測馬達的振動訊號,經由無感測器之轉速估測找出馬達變速時的旋轉頻率,並使用主成份分析法截取故障特徵,最後利用終端引點改良式動態結構類神經網路進行診斷鑑別。 由於傳統的馬達旋轉故障診斷系統只能針對固定頻率進行診斷,故不適用於馬達在改變轉速的情況下。因此本論文將透過越零偵測法估測出馬達轉速後,進而找出馬達變速時相對應的旋轉頻率,作為故障診斷的依據。除此之外,馬達運轉在低頻時,故障特徵在頻譜上的振幅會變小,雜訊則會變大,容易使得類神經網路產生誤判。針對此現象,本論文引入了主成份分析法,利用主成份的概念,將故障特徵參數精簡化。除了可讓類神經網路能更快速收斂,還可以去除不必要的雜訊,以提高故障診斷的鑑別度。 本論文使用MATLAB軟體撰寫訊號處理、越零偵測和類神經網路功能模組,以Visual Basic實現人機介面。最後利用測試機台產生故障訊號,驗證所提之方法的有效性。由實驗結果得知,所研製的智慧型馬達變速旋轉故障診斷系統的確能夠有效的鑑別出馬達變速時的旋轉故障。
This thesis is aimed at detecting adjustable-speed motor rotary faults and implements an intelligent diagnosis system. The system measures the vibration signals by using the wireless sensor node mounted on the motor. A sensorless speed estimation algorithm is developed to find out the mechanical rotary frequency of adjustable-speed motor. Moreover, the principal component analysis is used to intercept the fault characteristics. Finally, the system adopts the dynamic structure neural network to establish the diagnosis functionality. Since the traditional rotary motor fault diagnosis system is only capable of diagnosing a fixed-frequency fault, it is not suitable for adjustable-speed motors. Therefore, this thesis estimates the rotary speed of adjustable-speed motor via zero crossing detection method. Furthermore, the rotary frequency of the adjustable-speed motor is calculated for fault diagnosis. In addition, as the motor operates at low-frequency, the amplitude of fault characteristic frequency is decreased and the influence of noise is increased. As a result, the diagnosis module is easy to generate a false alarm. To overcome this difficulty, this thesis uses the concept of principal component analysis to extract the fault characteristic parameters. Not only the convergence speed in the neural network training is increased, but also the noise effect is eliminated. Hence, the identification accuracy in the fault diagnosis is increased. This thesis uses the MATLAB software to develop the modules of signal process, zero crossing detection, and neural network and uses the Visual Basic software to implement the Human-machine interface. Finally, a test platform is used to generate the faulty motor vibration signals and perform vibration experiments. From the experimental results, the implemented intelligent adjustable-speed motor rotary fault diagnosis system is found capable of identifying adjustable-speed motor rotary faults.