Traditional Artificial Intelligence (AI) based fault detection approaches need a large amount of data for the model learning. However, in a real-world system, it is very difficult and expensive to obtain massive labeled fault data. In addition, the working conditions of a motor are usually variable, conventional fault diagnosis models with weak generalization ability can only be used for fault detection under constant working condition. The performance of traditional AI based approaches decreases when the working condition changes. To this end, a novel deep Gaussian process (GP) kernel transfer based few-shot learning method (RNGPT) is proposed in this paper for the fault detection of electric machines. First, a deep residual network (ResNet) is used to extract the features of the raw data. Then, the encoded latent feature vector is fed into the GP with kernel transfer ability to make the motor fault detection and classification. The proposed method uses much less data than the traditional AI based method to achieve fault diagnosis under variable working condition, and does not cause an overfitting problem. Experimental results of two case studies demonstrate that the proposed RNGPT model can accurately and effectively detect motor faults with limited labeled data under different working conditions. Experimental results of RNGPT with radial basis function (RBF) kernel model on simulation data present that the fault detection accuracy of the proposed method is about 16% higher than the conventional deep learning methods, 6% higher than other few-shot learning based methods in 5-shot and 4% higher in 1-shot. Finally, experimental on a real-world dataset, the RNGPT-RBF model still has the highest fault diagnosis accuracy in 5-shot (99.39 $ \pm $ 0.09%) and 1-shot (98.55 $ \pm $ 0.16%).

With the steady development of technology, electric motors (EMs) have become one of the most important components in modern industry. To ensure stable industrial production, detecting and classifying the EM faults is crucial. A novel intelligent deep-learning-based multi-fault detection method for EMs under varying working conditions is proposed in this paper. This method involves two steps: first, a 2D convolution network without pooling layer is proposed to extract features from raw EM data. In addition, a long short-term memory (LSTM) network is applied to extract the fault features for comparison. Second, a capsule network (Caps-Net) based on a dynamic routing algorithm is used as a classifier to realize intelligent multi-fault detection and improve the generalization performance of the proposed model. The proposed method is applicable to raw physical signals of EMs, which improves the overall efficiency of the fault detection. Moreover, the proposed method has a strong generalization ability. The simulation results demonstrate that the proposed approach can achieve higher accuracy than various benchmark methods. Moreover, its accuracy is at least 3% and 10% higher than those of other state-of-the-art models under two working conditions, in which the load type and size of the EM are changed, respectively.

In this paper, a new vibration reduction control of the switched reluctance machine (SRM) is proposed that takes into consideration the undesirable effects of the torque ripple. The control approach aims to reduce the variation of the sum of radial forces while optimizing the control parameters of the machine to achieve the torque ripple minimization. Thus, a reference current adapter is used in order to handle the trade-off between the vibration reduction and the torque ripple reduction. It produces an auto-tuning reference current that limits the variation of the torque and the total radial force. The effectiveness of the proposed control strategy is proved by simulation and experimental results. Comparisons with conventional control methods investigating vibrations, noise and torque ripple are provided, which show the advantages of the control approach based on an auto-tuning reference current.

A reactor is an important piece of equipment used for reactive power compensation in power system and has a significant impact on the safe operation of power system. Thermal behavior is one of the main causes of reactor failures. For an accurate analysis of the thermal behavior of reactors, electromagnetic–thermal–fluid multi-physics coupling modeling is chosen. However, there is a huge difference in size between the overall structure of the reactor and its insulating material, which makes it difficult to perform mesh generation, resulting in dense mesh and significantly increased solution degrees of freedom, thus making the solution of the reactor’s multi-physics field model very time-consuming. To address this, this paper proposes a simplified processing method to accelerate the solution calculation of the reactor’s multi-physics model. This method calculates the equivalent turns of each encapsulate with parallel coils in the reactor, simplifying the encapsulate into a single-layer coil, thereby greatly reducing the division and solution degrees of freedom of the multi-physics model, and thus accelerating the simulation calculation. Taking a BKDCKL-20000/35 dry-type air-core shunt reactor as an example, the outer diameter of the coil is nearly 12,000 times bigger than the coil insulation, which is a huge size difference. Both refined models and simplified models are established. Compared to the simulation results of the detailed model, the simplified model demonstrates good accuracy; the maximum relative error of temperature is just 2.19%. Meanwhile, the computational time of the simplified model is reduced by 35.7%, which shows promising effectiveness and significant potential for applying the optimization design and operation prediction of dry-type air-core shunt reactors for enhanced thermal performance.

This paper presents a hardware/software implementation for the vibration reduction control of a switched reluctance machine (SRM). A mixed partitioning-based architecture is proposed to make full use of both microprocessor and Field-Programmable Gate Array (FPGA) resources. The aim of the proposed controller is to control directly the radial forces in the stator teeth allowing a significant vibration reduction compared to regular controller applied to SRM. However, if the control is software-only implemented, it suffers from performance degradation due to the high execution time required by this implementation solution. For this reason, the control system implementation is partitioned. The speed control is implemented using a microprocessor, while the current control and direct force control are implemented in FPGA. This solution provides high dynamic response especially needed to control the radial force and the currents. The effectiveness of the hybrid implementation is verified by experimental results.