Aiming at the problems of intricate model and low accuracy in online torque estimation of switched reluctance motor (SRM), a novel online torque estimation method based on piecewise analytical modelling is proposed. According to the symmetry of the flux linkage characteristics and the structural characteristics of the stator and rotor pole arc of the switched reluctance motor in one electrical cycle, a method of dividing the half electrical cycle into five intervals is proposed, and the flux linkage analytical model and torque analytical model of each interval are established respectively. The estimation accuracy and operation time of the above models are verified by finite element analysis and constructing an experimental system based on DSP28335. The results show that compared with the traditional single analytical model, the proposed partition analytical model not only effectively improves accuracy of torque estimation, but also significantly reduces the operation time, hence, it is helpful to improve the control accuracy and dynamic performance of switched reluctance motor speed control system, which has good application value.