Piezoelectric-based energy harvesting is an efficient way to convert ambient vibration energy into usable electric energy. The piezoelectric harvester can work as a sustainable and green power source for different electric devices such as sensors and implanted medical devices. However, its application on civil infrastructures has not been fully studied yet. This dissertation aimed to study and improve the piezoelectric-based energy harvesting on civil infrastructures, especially on bridge structures. To reach the objective, a more accurate model for piezoelectric composite beams was built first, which can be adopted for the modeling of different kinds of energy harvesters. The apparatus consists of a cantilevered beam harvester with a piezoelectric patch. A coupled electromechanical modal model based on Euler-Bernoulli theory was used. The model includes both direct and inverse piezoelectric effects and can provide a better prediction for the dynamic response and energy output of a harvester. In order to validate the model, the piezoelectric energy harvester was fabricated and some modal tests were performed. In the next step, the formulation regarding to an Euler-Bernoulli beam under moving mass by considering effects of inertial, coriolis and centripetal forces was derived by using Newton method. Next, the equations were developed for an elastic moving beam. For verifying, the results of the numerical solution of the moving mass problem were compared to the experimental tests data of the litterature. Then, the piezoelectric energy harvesting from a bridge structure under moving point mass and distributed mass with different speeds was studied, theoretically and experimentally. Furthermore, analytical and numerical investigation of a cantilevered beam type harvester with a piezoelectric patch under band-limited ambient random acceleration as the input was validated by experiment. The harvested electric energy can be used to power the energy needs for devices such as wireless sensors. The focus was to investigate the effect of low-amplitude and low-frequency bandpass random excitations that is usual in ambient vibrations and it could be more realistic spectra for excitations as compared to Gaussian white noise on a harvester. Keywords : Smart structures, Energy harvesting, Piezoelectric patch, Model validation, Moving mass, Random vibrations