This article presents an improved rotor position estimation method for a brushless electrically excited synchronous starter/generator (BESSG). Based on the multistage characteristics of BESSG, a high-frequency voltage is injected into the stator side of the main machine (MM), and the estimated rotor position is obtained through the estimation of excitation current. To suppress the estimated rotor position harmonic error caused by inverter nonlinearities, a novel phase-locked loop based on the gradient descent algorithm is proposed. In this method, the weight of harmonics is adjusted iteratively to improve the accuracy and reliability of the rotor position estimation. Experimental results have verified the feasibility of the proposed strategy.

The stator parameters of the main exciter (ME) are essential for emerging advanced control algorithms of the brushless synchronous starter/generator (BSSG). Due to the special structure of BSSG, the traditional parameter estimation methods of the induction machine are inapplicable to ME. In this paper, taking advantage of the operating characteristics of the rotating rectifier, a stator parameter estimation method of ME with square voltage injection is proposed. According to the harmonic features of the stator currents of ME, an amplitude regulation approach of the injection signal is presented to terminate the commutation process of the rotating rectifier as soon as possible, resulting in prolonged conduction intervals. Since the rotor currents of ME are dc during conduction intervals and thus have no influence on the stator currents of ME, stator parameters of ME are estimated with only stator voltages and currents by the recursive least square algorithm. Finally, the effectiveness of this method is verified by simulation.

The main generator (MG) of the aviation brushless synchronous starter/generator (BSSG) has damping windings in the rotor and its stator inductance is very small, leading to the difficulty of directly applying the traditional high-frequency injection method to sensorless starting control of BSSG at zero- and low-speed range. The I/F starting control method is an open-loop scheme with the advantages of a simple algorithm, easy implementation, and applicability to various machines, so it is a very competitive zero- and low-speed sensorless starting scheme for BSSG. However, due to the heavy load and significant torque fluctuations of BSSG, traditional I/F starting control methods could cause difficulty in regulation, out-of-step fault, and poor reliability. To solve this issue, this paper proposes an enhanced I/F starting control method based on the special structural characteristics of BSSG. This method preserves the advantages of the traditional I/F algorithm while effectively resisting external disturbances, reducing the regulation difficulty and the risk of out-of-step, thus achieving highly reliable zero- and low-speed sensorless starting control of BSSG. The effectiveness of this method is verified by simulation.

The traditional rotor position estimation method of permanent magnet synchronous motor at zero and low speed based on high-frequency pulse voltage signal injection inevitably adopts band-pass and low-pass filters, which reduces the dynamic response speed of the system. In this paper, a new rotor position estimation strategy is proposed. First, in the static coordinate system, the pulse vibration high-frequency signal with the same frequency and amplitude is injected into the α and β axis respectively. At the same time, the DC offset for rotor position estimation is extracted. Then the high-frequency pulse vibration signal is injected into the α axis, the obtained DC offset is used to compensate the current response signal, and the cascade filter is used to expand the bandwidth of the position estimation link and improve the dynamic response; Finally, at zero speed, the positive and negative square wave signals are injected into the α axis or β axis, and the initial position is determined by comparing the current amplitude of the corresponding shaft. The effectiveness of this method is verified by simulation. The results show that this method can quickly and accurately obtain the rotor position information at zero and low speed, and has good stability and dynamic characteristics.