Search

To achieve improved multinode temperature estimation with limited training data in permanent magnet synchronous motors, a novel approach of a lumped-parameter thermal network (LPTN) informed neural network is proposed in this article. First, the parameter and model uncertainties of third or higher order LPTNs with global parameter identification for temperature estimation are systematically stated based on numerical analysis. Then, a two-step parameter identification strategy for a third-order LPTN with simplified thermal transfer paths is proposed to resolve parameter uncertainty. This strategy uses only air-gap structure information to make all parameters converge to their unique solutions without the need for additional geometrical parameters or material features. In response to model uncertainty, an LPTN-informed long short-term memory (LSTM) framework is designed to compensate for model unaccounted errors and extend temperature estimation nodes that the highly abstract low-order LPTN does not consider. Experimental temperature estimation results validate the effectiveness of the proposed LPTN-informed LSTM framework under a limited 23.8 h of training data.

Accurate position observation in interior permanent magnet synchronous motors (IPMSMs) model-based sensorless drive relies on the precision of model parameters. This article first derives an analytical function that quantitatively relates position observation error to mismatches in multiple model parameters and the current lead angle in one unified equation. The position error mechanism highlights the potential limitations of relying solely on steady-state-based parameter estimation. It also emphasizes the importance of introducing current transient states for online multiparameter estimation. To reduce the position observation error caused by parameter mismatch, a multiparameter estimation method with extra small signal injection based on the dual extended Kalman filter with a position error prior prediction structure is proposed for the IPMSM sensorless drive. The method is built upon the misaligned synchronous frame model where the asymmetry in dq-axis magnetic path, flux coupling terms, and position observation error are well considered. This makes the proposed method effective for both steady and transient states within the rated speed. Additional experiments corroborate both the validity of the analysis and the effectiveness of the proposed technique.

5G is a promising technology for improving the Quality of Service (QoS) in Internet of Vehicles (IoV) applications, including Vehicular Edge Computing (VEC). However, 5G networks have a limited communication range due to their radio-frequency properties, which can be a challenge in dynamic IoV environments. To address this issue, we propose a VEC architecture based on heterogeneous cellular networks, in which vehicles can select the appropriate communication network by classifying tasks according to their maximum tolerable latency. In order to further enhance the overall performance of the VEC system, we developed an efficient scheme that optimizes task offloading decisions and computation resource allocation in the proposed architecture. We analyze and formulate the optimization problem and use the linear relaxation improved branch-and-bound algorithm to solve it. Through extensive simulations, we demonstrate that the proposed scheme is superior to other solutions in computing latency, energy consumption, and failure rate.

Short-circuit faults in permanent magnet (PM) machines have consistently been a hindrance for safety-critical applications. However, short-circuit currents, especially turn-to-turn short-circuit currents, are difficult to detect and suppress in a timely manner. This article investigates the automatically variable impedance characteristics of the toroidal winding, which restricts the turn-to-turn short-circuit current to an allowable level without terminal short circuit. An analytical model considering the saturation effect of toroidal winding is proposed to reveal the fault tolerance mechanism as well as the mapping relationship between yoke-annular-leakage and fault-tolerant performance. It is found that variable impedance PM machines with toroidal winding have superior fault tolerance performance compared to conventional fractional-slot concentrated-winding and lap winding. Besides, there is a contradictory relationship between its fault tolerance and electromagnetic performance. Furthermore, the laws of phase belt angle and pole number on the electromagnetic and fault-tolerant performance of toroidal winding are first presented in this article. The design criteria and the feasible pole-slot combinations that have inherent fault-tolerant capability are given. Finally, the prototype is built and tested. The experimental results match well with the analytical ones.

This article presents an improved lumped parameter thermal network (LPTN) based online temperature prediction method of permanent magnet (PM) machines by global parameter identification. The proposed global parameter identification incorporates not only the loss but also the magnetic information into a fifth-order LPTN and identifies the multiphysical magneto-thermal parameters as a whole. In the identification process, the nonlinear characteristics, including the magnetic saturation, the cross-saturation, the strand skin effect, and the speed-dependent convective heat transfer, are all considered. It can eliminate the cumbersome magnetic and loss measurement efforts and improve the overall identification accuracy compared with the traditional step by step identification method. Based on the identified multiphysical magneto-thermal model, the critical temperatures of the PM machines, together with the online losses and torque, can then be predicted in real time. The experimental results show that the proposed online temperature prediction method can track the real temperature variation with errors within 5 °C.