Tao Meng

To enhance robustness and achieve more precise rotor position estimation in switched reluctance motors (SRMs), this article proposes an improved sensorless control method based on a particle filter observer. First, the unscented Kalman filter was chosen as the observer to replace the traditional particle filter, and the latest observation information was used to calculate the mean and variance. Second, to solve the particle degradation problem of the particle filtering algorithm and prevent falling into local optima, random resampling is used for weight calculation. The proposed method will have a more accurate parameter basis to improve the accuracy of position estimation, thereby optimizing the poor tracking performance when dealing with continuous changes. The performance of the proposed method was evaluated through experiments under different working conditions. The results show that the position estimation error based on the improved particle filter observer is within 1.25 ${{}^{\circ}}$ , demonstrating the effectiveness and good dynamic performance of the proposed method.

This article proposes a rotor position estimation method for brushless electrically excited synchronous starter/generator (BEESSG) at low speed region. Based on the multistage characteristics of BEESSG, two high frequency pulsating voltages with different frequencies are injected into phase $\alpha$ and phase $\beta$ of stator of the main machine simultaneously. The rotor position is estimated by the high frequency induced signals extracted from the stator windings of the main exciter through a position estimator. Low-pass filters are replaced by an auxiliary square wave sampling method to demodulate the rotor position from high frequency induced signals. Due to the amplitude unbalance of demodulated position related signals. A novel first quadrant definite integral method is proposed in this article for position compensation. Furthermore, the effectiveness and feasibility of the proposed method are verified by experimental results.

Cable-driven parallel robots (CDPRs) have been applied very well because of its advantages. Like other parallel robots, the forward kinematic solution of the parallel robot driven by cable is usually faced with complex nonlinear equations, and its solution is always a difficult problem. The traditional numerical iteration method is heavily dependent on the selection of initial values, which is easy to fall into local convergence, leading to the result of calculation is not necessarily the ideal spatial pose. The neural network has good nonlinear fitting ability and can give correct spatial pose, but the solution accuracy needs to be improved. The convergence and precision of the forward kinematic solution of the CDPRs is solved by the improved algorithm combining neural network and Levenberg- Marquardt algorithm.

Recently, most wireless power transmission (WPT) systems have used a "one-to-one" charging method of one receiving coil to one transmitting coil. However, when the receiving coil is offset in this way, the mutual inductance between the transmitting and receiving coils will be weakened, and the flexibility of the WPT system application will also be reduced. Therefore, how to realize the "one-to-many" transmission mode of one transmitting coil to multiple receiving coils and ensure the power stability of multiple loads is an urgent problem to be solved. In order to solve the above problems, the LCC-S topology is combined with the three-phase circuit, and a three-phase omnidirectional WPT system coil scheme is proposed based on the power characteristics of the three-phase LCC-S topology in this paper. Firstly, the influence of the resonance compensation topology on power characteristics is analyzed through circuit theory and a more suitable LCC-S topology is chosen. Then, the optimal coil scheme of the three-phase omnidirectional WPT system is proposed based on the law of mutual inductance. Compared with the original scheme, the theoretical calculation results show that the power fluctuation of the system is reduced by 75.13%. Finally, a set of three-phase WPT system model and test systems similar to the theory are established, which effectively verify the improvement of the proposed scheme, and verify the correctness and feasibility of the scheme.

The changeable magnetic saliency of main machine (MM) in aviation three-stage stator/generator (ATSG) makes it difficult to implement the sensorless starting in the low speed region by the machine saliency based methods. To estimate the rotor position of ATSG (which is also the rotor position of MM), a novel rotor position estimation method based on the harmonic currents in the main exciter (ME) stator windings produced by the rotating rectifier is proposed in this paper. The sine and cosine functions on the rotor position without time delay are extracted from the ME stator currents by a constant frequency signal sampling method. Then, the estimated ME rotor position can be obtained by processing the sine and cosine functions through a phase-locked loop. As the ME and MM are mounted on the same shaft, the rotor position of MM can be obtained by the pole ratio between ME and MM. High-frequency (HF) signal injection is not required in the proposed method, so the HF torque ripple and HF loss caused by the injected HF signal can be avoided. The good performance of the proposed method is verified by the experimental results.

This paper proposes a novel on-line rotor position estimation method for an aircraft wound-field synchronous starter/generator (WFSSG) in the low-speed region. The method is based on the fact that mutual inductance between stator and rotor windings of the main machine (MM) varies with the rotor position, so it is independent on magnetic saliency. High-frequency (HF) rotating voltage is imposed in the stator windings of the MM, and the rotor position information is extracted from the HF response current signals in the stator windings of the main exciter (ME) in WFSSG. Theoretical expressions of the HF response current signals containing rotor position information in the rotor and stator windings of the ME are derived. Then, harmonics in the ME stator windings are analyzed, and optimum frequencies of the injected HF voltage for MM and excited voltage for the ME are obtained. Also, a novel signal processing method without time delay is adopted to improve the signal to noise ratio of the extracted rotor position information. The performance of the proposed method is verified by the experimental results, and accurate rotor position can be estimated even the magnetic saliency of the MM changes greatly during the start-up process.

Wound-rotor synchronous starter/generator (WRSSG) has been widely adopted in the modern aircrafts. The uncertainty of the inductive saliency in WRSSG (The saliency may change, disappear or reverse during starting process) makes it difficult to implement the sensorless starting in the low speed region. This paper proposes a dual-machine injection method for the sensorless control of WRSSG, which is able to eliminate the dependency on the machine saliency in rotor position estimation. Using the multi-stage structure of WRSSG, in the proposed method, a high frequency (HF) voltage is injected in the main machine (MM) of WRSSG, and HF current response in the main exciter (ME) side of WRSSG is detected and used to estimate the rotor position since the ME is connected with the MM through a rotating rectifier. Experiments about the proposed dual-machine injection method on a WRSSG prototype in low speed region were carried out, and the experimental results verified the effectiveness and advantages of the proposed method.