Intelligent vibration analysis of industrial cooling fans

Abstract

Industrial cooling fans are responsible for maintaining stable temperatures for delicate components. Therefore, a cooling system failure can certainly lead to machine downtime. Fault Condition Monitoring (FCM) is a predictive maintenance method that can be applied to cooling fans for fault prediction. As the components of a cooling fan wear off, its vibration tends to vary. Thus, this paper uniquely elaborates on three intelligent vibration analysis techniques that are applicable in the FCM of cooling fans. In this research, 1) image encoding with convolutional neural network (CNN), 2) moving average, and 3) fuzzy logic techniques are designed, employed, and their potentials as FCM tools are compared. The vibration data is collected from an experimental test bench that consists of a fan, an accelerometer, and a microcontroller, among others. Once enough data is obtained, the three vibration analysis techniques are applied using Python and MATLAB. The results reported in this paper demonstrate the effectiveness of these intelligent vibration analysis techniques in the FCM of cooling fans and possibly other industrial equipment. The novelty of the research revolves around the fan fault classification techniques that are being compared. The image-encoding technique described in this paper has yet to be applied for fault classification. Additionally, while fuzzy logic and moving average are popular methods, this is the first time that they are being used for vibration analysis of cooling fans. Furthermore, this is also a novel comparative study of different vibration analysis techniques.