Search

This paper proposes a highly symmetrical inductive power transfer (IPT) system based on double-sided Z-impedance compensation networks. The double-sided Z-impedance networks are designed with symmetrical configuration and could achieve load-independent CC output with zero-phase angle (ZPA) of input impedance. Compared to the conventional LCC circuit, the proposed Z-network can reduce current stresses on compensation components, which contributes to high-power applications. Meantime, compensation inductors are integrated with main coils to make the whole system compact and eliminate unwanted cross couplings. The proposed IPT system is validated by experiments and 2.3kW power could be transferred with 96.93% dc-dc efficiency across an air gap of 120mm.

This digest proposes a power control strategy using a voltage boosting and frequency detuning approach in an S-S compensated EV IPT system. voltage is increased to increase output power, and frequency is increased from the resonant frequency to decrease output power, reducing the voltage range requirement of the input. The system is designed to maintain full power range control despite fluctuating load and misalignment condition. A prototype is developed that achieves 0.0–3.3kW power control across all battery load and x-y misalignment combinations. The proposed strategy is tested by constructing an experimental prototype, that achieves a peak DC/DC efficiency of 95.7%

This digest proposes a cross-coupled LCC-P coil structure to improve the misalignment tolerance of a Z-class compatible inductive wireless power transfer system for electric vehicles (EVs) described in J2954. The standard specifies the misalignment distances in the x-y-z plane to which the system will be exposed. Misalignment induces large variations in coupling coefficients between the primary and secondary sides of the coupler, resulting in proportionally large power fluctuations. The design uses a reversed coupled filter inductor to improve misalignment performance. Compared to previously proposed designs, the proposed LCC-P topology offers good performance, while reducing the size and complexity of the vehicle side coupler. Additionally, all the coils in the coupler are unipolar, further reducing the overall system complexity. A prototype is designed and constructed to validate the proposed solution. Experiments have achieved 3kW power transfer with 95.8% efficiency.

This paper presents emotion recognition models using facial expression features. By detecting the face in videos and extracting local characteristics (landmarks) to generate the geometric-based features to discriminate between a set of five emotion expressions (amusement, anger, disgust, fear, and sadness) for videos from BioVid Emo database. The classification operation is done using different machine learning models including random forest (RF), support vector machines (SVM), k-nearest neighbors (KNN) and recurrent neural network (RNN), then the evaluation operation is done to generate different discrimination rates that reached up to 82% to discriminate between anger and disgust emotions.