Songwei Pei

IGBT Junction Temperature Monitoring and State Identification Based on AWG-NN

Tianqi Li, Songwei Pei, J. Zhang

Precise monitoring of insulated gate bipolar transistors (IGBTs) junction temperature, which impacts device lifespan, is crucial for assessing their health and operational status. However, traditional monitoring methods are limited, and effective IGBT state assessment methods are lacking. This article presents a novel …

cn (code pays fourni par la source)

AP-Net: Attention-fused volume and progressive aggregation for accurate stereo matching

Yansong zhu, Songwei Pei, BingFeng Liu, Jun Gao

Stereo matching plays an important role in advancing substantial applications such as 3D reconstruction, autonomous driving, depth sensing, and has recently made tremendous progress. It still poses a significant challenge, however, in the extraction of disparities from a pair of rectified images …

cn (code pays fourni par la source)

RAD-BNN: Regulating activation distribution for accurate binary neural network

Mingyu Yuan, Songwei Pei

The outstanding performance of deep convolutional neural networks comes from their effective extraction and learning ability. Although binary neural networks (BNNs) have the obvious advantages of low storage and high efficiency over their full-precision counterparts on resource-constrained hardware devices, the accuracy degradation …

cn (code pays fourni par la source)

State evaluation network for UHVDC converter

Songwei Pei

To evaluate the health state or failure of the converter valve in the early stage and realize the goal of condition-based maintenance rather than regular maintenance, in this work, we are the first to propose a state evaluation network for UHVDC converter(SE-Net). …

cn (code pays fourni par la source)

A Method for Evaluating the Status of Converter Valves Based on Siam-LSTM Neural Network

Lei Shi, Tianqi Li, Songwei Pei, Ruopeng Liu et autres

As the key equipment of DC project, the converter valve bears electrical and thermal stress for a long time, and the real-time mastering of its operation status is the key to the safety and stability of DC project. In this paper, a …

cn (code pays fourni par la source)

Rad-Bnn: Regulating Activation Distribution for Accurate Binary Neural Network

Mingyu Yuan, Songwei Pei

cn (code pays fourni par la source)

PO-DARTS: Post-optimizing the Architectures Searched by Differentiable Architecture Search Algorithms

Debei Hao, Songwei Pei

cn (code pays fourni par la source)

Small Sample Hyperspectral Image Classification Method Based on Dual-Channel Spectral Enhancement Network

Songwei Pei, Hong Song, Yinning Lu

Deep learning has achieved significant success in the field of hyperspectral image (HSI) classification, but challenges are still faced when the number of training samples is small. Feature fusing approaches based on multi-channel and multi-scale feature extractions are attractive for HSI classification …

cn (code pays fourni par la source)

A parallel sparse triangular solve algorithm based on dependency elimination of the solution vector

Song Jin, Songwei Pei, Yu Wang, Yincheng Qi

cn (code pays fourni par la source)

On Improving Fault Tolerance of Memristor Crossbar Based Neural Network Designs by Target Sparsifying

Song Jin, Songwei Pei, Yu Wang

Memristor based crossbar (MBC) can execute neural network computations in an extremely energy efficient manner. However, stuck-at faults make memristors cannot represent network weight correctly, thus degrading classification accuracy of the network deployed on the MBC significantly. By carefully analyzing all the …

cn (code pays fourni par la source)

Bending Performance of Steel Fiber Reinforced Concrete Beams Based on Composite-Recycled Aggregate and Matched with 500 MPa Rebars

Xiaoke Li, Songwei Pei, Kunpeng Fan, Haibin Geng et autres

To promote the engineering application of recycled aggregate for concrete production with good adaptability and economic efficiency, this paper performed a campaign to investigate the flexural performance of steel fiber reinforced composite-recycled aggregate concrete (SFR-CRAC) beams matched with 500 MPa longitudinal rebars. …

cn (code pays fourni par la source)

A variation tolerant scheme for memristor crossbar based neural network designs via two-phase weight mapping and memristor programming

Song Jin, Songwei Pei, Yu Wang

cn (code pays fourni par la source)