In this paper, a new nonlinear observer is presented for DC-DC boost converter using immersion and invariance (I&I) technique. The proposed nonlinear observer is easy to implement, simple to tune and realizes global exponential convergence of the estimation error to zero. It is assumed that the input voltage and output load of converter are unavailable and the I&I observer is designed to estimate unavailable parameters using output voltage and inductor current information. Considering optimal performance of observer, an improved particle swarm optimization algorithm is employed to determine observer gains. The search space of parameters in the proposed optimization algorithm is modified and according to this modification, the weak parameters is eliminated and the new parameters are defined for optimization process instead of them. In order to validate the effectiveness of the proposed observer, the practical test is implemented and the experimental results confirm the efficiency of the proposed I&I observer.