Abstract This research focuses on the prediction and experimental verification of P-bifurcation as well as the effectiveness in reducing vibrations and harvesting energy with the use of an inertially nonlinear energy harvesting device attached to a single-degree-of-freedom structure subjected to Gaussian broadband base excitation, modeled as white noise. Four experimental scenarios were tested, including three with different resistive loads and one with an open circuit. Frequency domain optimization involved an optimization routine that was designed to minimize the mean squared error in the pendulum velocity's frequency content below two cycles per second while constraining the RMS velocity discrepancy between the simulations and actual experiments to be below 3%. This facilitated accurate predictions of power, vibration suppression, and P-bifurcation. Using the fitted model, an analytically derived P-bifurcation boundary in the noise intensity versus electrical damping plane was presented and experimentally verified. Additionally, power spectral densities for electric power and relative suspended mass velocity were determined for the IPVA system and compared with a top-performing linear system. Results indicated that the monomodal system was the least effective in energy harvesting, while the bimodal and rotational systems significantly enhanced mean and resonant peak power by up to four and two times, respectively. Near the resonant frequency, the peak relative velocity power spectral density decreased by a factor of four, and the mean square relative velocity improved by a factor of two.