Computational intelligence methods for rolling bearing fault detection

Rolling bearings are very commonly used in many industrial applications. Therefore, detecting problems in their performance is very essential. This can be done by analyzing vibration signals resulting from their operation, as recorded by accelerometers. The current investigation aims to evaluate the efficiency of various computational intelligence algorithms, in detecting and correctly classifying faults in rolling bearings. A supplementary goal is to determine the optimum location for the accelerometers, in order for the aforementioned algorithms to identify faults on each bearing. Results indicate the most efficient computational intelligence methods for fulfilling the aforementioned goals, and suggest an optimum experimental setup, in order to successfully detect bearing faults.