The polyphase motor has good fault tolerant operation due to its inherent multi-phase structure. In order to improve its control performance in transient operation conditions, accurate motor models and parameters are essential. In this paper, a mathematical modeling and parameter identification method for the whole link of multiphase drive system is proposed. Taking a six-phase motor as an example, a unified state-space model of the six-phase motor in the stationary αβ coordinate system is derived. Then, based on the hardware architecture of the control system and the software execution timing, the control delay from the instruction voltage to the actual voltage and the sampling delay characteristics from the actual current to the feedback current are analyzed. On this basis, a full link response model from command voltage to feedback current in the control system is established. Finally, an off-line parameter identification method for data-driven state-space model based on improved particle swarm optimization algorithm is proposed. The experimental data analysis shows that the overall model error is within 5% and the maximum error is not more than 10%.