Machine Condition Monitoring through Frequency Analysis of Maintenance Records: A Neuro-Fuzzy Approach

Machine condition monitoring and fault diagnosis fields are charged with developing new technologies to detect and diagnose the machinery problems. In fact, several methods of fault diagnostics have been developed and applied effectively to identify the machine faults at an early stage using different quantities (Measures or Readings) such as current, voltage, speed, temperature, and vibrations. The major problem of diagnosis techniques is that they require constant human interpretation of the results; therefore, the research has been underway for a longtime to automate the diagnosis process. Recently, artificial intelligence-based techniques (e.g., neural networks, fuzzy logic, genetic algorithms, hybrid methods, and intelligent agents) have been recently utilized widely with the monitoring system to support the diagnosis system. In this paper, a machine condition monitoring and diagnostic system is introduced with experimental verification. An adaptive neuro-fuzzy inference system (ANFIS) is used to monitor and predict the fault types. The system uses a piezoelectric accelerometer to generate a signal related to machine condition and fault type. The power spectral density (PSD) of this signal is used as an input to ANFIS, which in turn outputs a value for predicted fault type. Experimental validation runs were conducted to compare the actual fault types with the predicted ones. The comparison shows that the adoption of trapezoidal membership function in ANFIS achieved a very satisfactory fault prediction accuracy of 99.9%.