In order to avoid the money loss and injuries caused by the structural damage, early detection of the small cracks in a structure is very important. Conventional damage detection techniques, such as ultrasonic methods, strain energy methods, magnetic field methods, etc., are usually lack of sensitivity or hard to be applied to different surfaces requiring complex sensing systems. In this thesis, a new piezoelectric coating sensor is developed to detect the crack initiation, and it can apply to any surfaces of interest. Two sensitive crack measurement methods, wavelet Entropy and Frequency Comparison Function (FCF), are introduced to evaluate the crack for the structure based on the vibration signals from the new sensor. During operating, the piezoelectric composite coating sensor is applied at the welding joint of a vibrating structure to send warning and dynamic signals for damage detection and evaluation, when the crack occurs. Entropy and FCF methods are introduced to quantify the weak dynamic perturbations, which are caused by the strain concentration and/or crack breathing at the crack tip. A finite element model (FEM) of a welded beam subjected to the dynamic base motions is established for case studies to show the efficiency of the proposed smart coating and measurement methods. The effects of strain/stress concentration and crack breathing on the structural dynamic response are simulated by creating the nonlinear material property around the crack area and the contact pair of the crack walls, respectively. From simulations, both methods are found to be sensitive to the initiated closed crack. The Entropy method can detect a crack of 5% thickness of the beam thickness. Meanwhile, it is feasible and sensitive for both open and closed cracks detection. The FCF method can detect a closed crack with a size of 3% of the beam thickness. In addition, FCF is fast and efficient with no required data pre-progressing, like the filtering and smoothing functions, and can hence be used for real-time crack detection. Experimental validations are conducted for both methods, and the results prove high sensitivity and feasibility of both proposed crack detection methods.