شماره ركورد كنفرانس :
583
عنوان مقاله :
Acoustic Condition Monitoring of Centrifugal Pump by Neural Network
عنوان به زبان ديگر :
Acoustic Condition Monitoring of Centrifugal Pump by Neural Network
پديدآورندگان :
Ahmadi Hojjat نويسنده University of Tehran, Karaj, Iran - Department of Mechanical Engineering of Agricultural Machinery - Faculty of Agricultural Engineering and Technology , Farokhzad Saeid نويسنده University of Tehran, Karaj, Iran - Department of Mechanical Engineering of Agricultural Machinery - Faculty of Agricultural Engineering and Technology
كليدواژه :
wavelet transform , Artificial neural network , Fault diagnosis , Acoustic signal , Centrifugal Pump
عنوان كنفرانس :
نهمين كنفرانس ملي نگهداري و تعميرات
چكيده لاتين :
Pumps play a significant role in industrial plants and need continuous monitoring to minimize loss of production. This paper focuses on the use of artificial neural network for fault diagnosis of centrifugal pumps. The centrifugal pump conditions were considered to be healthy pump and impeller defect, bearing defect, seal defect and cavitation which were five neurons of output layer with the aim of fault detection and identification . Features vector which is one of the most significant parameters to design an appropriate neural network were extracted from analysis of acoustic signals in time-frequency domain by means of wavelet transform method. The characteristic features of acoustic signals such as mean, standard deviation, variance, skewness, kourtosis etc, introduced were used as input to ANN. Different neural network structures are analyzed to find the optimal neural network with regards to the number of hidden layers. The results show that the designed system is capable of classifying records with ٩١.٦٢% accuracy with one hidden layers of neurons in the neural network.
شماره مدرك كنفرانس :
4490309
