Because of immeasurable variables， synchronous generator parameter identification need to solve complex differential equations， which might lead to difficulties in the parameters identification process. This paper presents a new generator model， on which all the variables can be measured. Firstly， the model is a good guarantee of both the precision and the stationarity of the parameter identified. Secondly， because all the variables in the model can be measured， it can be avoided to solve complex differential equations， so that the stability of identification process and the arithmetic speed can be improved and the parameter identification will become simple and convenient. The identifiability of the model parameters is analyzed in detail to conclude that the model parameters are uniquely identifiable when the steady-state equations are used as an additional condition. The model parameters are identified using least squares identified algorithm according to steady-state data and dynamic-state data respectively. The test results with PSASP show that the model presented in this paper is correct， feasible and valid.