<p>Active health monitoring and fault diagnosis methods are essential to improve the safety and reliability of rotating machineries and to prevent from catastrophic failure. The conventional fault diagnosis methods require battery-support sensors. This paper presents a novel piezoelectric smart bearing to fulfill fault detection without using battery-support sensors. An electromechanical coupling model of the unbalanced flexible rotor with piezoelectric smart bearings is established using the prominent principle of piezoelectric transducers and Lagrange equation. This model also takes in to account the nonlinearity due to the breathing transverse crack. Numerical exploration for the voltage response when the crack grows deeper is performed using the frequency response, orbit diagram, power spectrum and bifurcation diagram. Then, a test rig has been designed and built for experimental validation. The obtained results show that the voltage responses of the system contain the fault characteristic frequencies. So, the proposed smart bearing is capable of detecting the unbalance and crack faults and can be used for self-powered condition monitoring of rotating machines.</p>