Pump is one of most crucial components for transferring fluids in any industry; further, bearing is the most important element in any self-priming centrifugal pump. The performance of the pumps highly depends on the conditions of various parts, especially bearing. If the parts of the pumps are in good conditions, then it leads to smooth operations which ultimately leads to high performance. But with number of cycles, the defects get introduced various parts especially in the bearings and impeller. The diagnosis and identification of these faults is a major challenge as it is costly and time-consuming. In past decade, an advancement is made in fault diagnosis where the faults are identified with the help of machine learning techniques. By diagnosing the faults through these methods, results are lesser downtime during maintenance. This study focuses in the usage of two prominent signal processing techniques, namely wavelet packet transform (WPT) and maximal overlap discrete wavelet packet transform (MODWPT). After processing the signals, features are extracted which are then selected using the reliefF method. These selected features are used as the input to various machine learning models like support vector machine model (SVMM) which are then compared to make the approach for diagnosing the faults and make it more intelligent. The signal processing technique namely MODWPT and with SVMM as the machine leaning model shows the best results.