About

Peiwen "Perry" Li is a Professor of Aerospace and Mechanical Engineering and a member of the Graduate Faculty at the University of Arizona. He holds a PhD in Thermal Science for Energy and Power Engineering from Xi'an Jiaotong University in China. His research interests encompass teaching and research in thermodynamics, heat transfer, fluid mechanics, numerical heat transfer, fuel cell analyses and designs, renewable energy systems, energy conversion systems, and energy storage heating, ventilation, air-conditioning, and refrigeration. His work focuses on fuel cell and electrolysis technologies, hydrogen production, renewable energy technologies, thermal energy storage, and thermal-driven water technology. Professor Li has contributed extensively to the field through experimental and numerical analysis, optimization of solar receivers, desalination systems, and energy storage solutions, with numerous publications in reputable journals and conference proceedings.

Research topics

• Materials science
• Chemistry
• Thermodynamics
• Physics
• Engineering
• Chemical engineering
• Nuclear engineering
• Mechanical engineering
• Inorganic chemistry
• Mechanics
• Metallurgy
• Composite material
• Mathematical optimization
• Mathematics
• Organic chemistry

Selected publications

• High-fidelity three-dimensional reconstruction of musculoskeletal tissues via diffusion based ultrasonic computed tomography

  Medical Image Analysis · 2026-05-01

  article
  PublisherDOI

• Performance analysis of helium gas jet impingement heat transfer enhanced concentrating solar receiver

  Solar Energy · 2026-04-10

  article1st authorCorresponding
  PublisherDOI

• Liberating the expressive capacity of deep hashing for image retrieval

  Information Processing & Management · 2025-07-04 · 1 citations

  article
  PublisherDOI

• Deep Hashing With Walsh Domain for Multi-Label Image Retrieval

  IEEE Signal Processing Letters · 2025-01-01 · 6 citations

  article

  The existing deep hashing methods for image retrieval typically modeling hash-coding layer in real number space. However, these methods frequently overlook the intrinsic information loss that occurs during the hash-coding process, as the hash layer performs two tasks simultaneously: spatial transformation and dimensionality reduction. Especially in multi-label image retrieval, the exponential increase in the number of combinations with the labels further amplifies the information loss in Hamming space. Consequently, the efficiency of the hash-coding is unsatisfactory. To mitigate this limitation, we introduce a novel approach termed WalshHash, which is grounded in the principles of Walsh transformation in signal processing. Unlike conventional techniques, WalshHash formulates the hash-coding layer as a filtering process based on Kolmogorov-Arnold Networks (KANs) in the Walsh domain accompanied by constraint loss functions on multiple domains. It ensures the dimensionality reduction in the Walsh domain can be effectively projected onto the real number domain with minimal information loss, because the Walsh space encapsulates the critical information components. As a result, WalshHash demonstrates superior performance in multi-label image retrieval compared to State-of-the-Art (SOTA) methods.

  PublisherDOI

• Wavelet Swin Transformer Low-Frequency Prediction for Ultrasound Full Waveform Inversion of Bone

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access
  PublisherDOI

• Development of boron-enhanced inconel 718 with superior thermomechanical properties for high-temperature concentrated solar power applications

  Solar Energy Materials and Solar Cells · 2025-06-20 · 2 citations

  article
  PublisherDOI

• A multi-task neural network for full waveform ultrasonic bone imaging

  Computer Methods and Programs in Biomedicine · 2025-04-25 · 3 citations

  articleOpen access1st author

  • The proposed multitask Unet with one encoder and two decoders, designed to reconstruct the bone SOS model and fit the contour of the SOS model simultaneously and outputs more accurate and edge-sharp images of bone SOS models. • An attention mechanism based down-sampling module was adopted to improve the performance of the multi-task model. The Efficient Multi-Scale Attention (EMA) module was introduced to the encoding path for adaptive feature refinement of the input feature map before each max-pooling layer, to refine representative features and enhance the feature representation ability of the network. • The Convolutional Long Short-Term Memory (ConvLSTM) module was adopted to combines the spatial feature extraction capability of Convolutional Neural Network (CNN) and enable the network to better extract multi-scale spatial-temporal joint features. It is a challenging task to use ultrasound for bone imaging, as the bone tissue has a complex structure with high acoustic impedance and speed-of-sound (SOS). Recently, full waveform inversion (FWI) has shown promising imaging for musculoskeletal tissues. However, the FWI showed a limited ability and tended to produce artifacts in bone imaging because the inversion process would be more easily trapped in local minimum for bone tissue with a large discrepancy in SOS distribution between bony and soft tissues. In addition, the application of FWI required a high computational burden and relatively long iterations. The objective of this study was to achieve high-resolution ultrasonic imaging of bone using a deep learning-based FWI approach. In this paper, we proposed a novel network named CEDD-Unet. The CEDD-Unet adopts a Dual-Decoder architecture, with the first decoder tasked with reconstructing the SOS model, and the second decoder tasked with finding the main boundaries between bony and soft tissues. To effectively capture multi-scale spatial-temporal features from ultrasound radio frequency (RF) signals, we integrated a Convolutional LSTM (ConvLSTM) module. Additionally, an Efficient Multi-scale Attention (EMA) module was incorporated into the encoder to enhance feature representation and improve reconstruction accuracy. Using the ultrasonic imaging modality with a ring array transducer, the performance of CEDD-Unet was tested on the SOS model datasets from human bones (noted as Dataset1) and mouse bones (noted as Dataset2), and compared with three classic reconstruction architectures (Unet, Unet++, and Att-Unet), four state-of-the-art architecture (InversionNet, DD-Net, UPFWI, and DEFE-Unet). Experiments showed that CEDD-Unet outperforms all competing methods, achieving the lowest MAE of 23.30 on Dataset1 and 25.29 on Dataset2, the highest SSIM of 0.9702 on Dataset1 and 0.9550 on Dataset2, and the highest PSNR of 30.60 dB on Dataset1 and 32.87 dB on Dataset2. Our method demonstrated superior reconstruction quality, with clearer bone boundaries, reduced artifacts, and improved consistency with ground truth. Moreover, CEDD-Unet surpasses traditional FWI by producing sharper skeletal SOS reconstructions, reducing computational cost, and eliminating the reliance for an initial model. Ablation studies further confirm the effectiveness of each network component. The results suggest that CEDD-Unet is a promising deep learning-based FWI method for high-resolution bone imaging, with the potential to reconstruct accurate and sharp-edged skeletal SOS models.

  PublisherDOI

• Sequential hash representation for deep hashing-based image retrieval

  Knowledge-Based Systems · 2025-08-06

  article
  PublisherDOI

• Cost-Effectiveness Analysis for Solar Energy Harvesting Field With PV Panels at Optimized Tilt Angles Under Year-Round Weather Cycles

  2024-07-15 · 1 citations

  article

  Abstract This study introduces the Sunray-Energy Algorithm (SEA), an innovative tool designed to estimate the maximum solar energy received by photovoltaic (PV) solar panels. Evaluating optimal tilt angles under year-round cloud cover, SEA yields annual energy of 2206.36, 2371.58, 2371.89, and 2407.40 for one-time, twice/year, quarterly, and monthly PV tilt angle adjustments. The application of SEA extends to a comprehensive cost analysis for The University of Arizona Tech Park’s Solar Zone in Tucson, AZ, USA. Predicted annual energy harvests range from 39.966 to 43.607 GWh, demonstrating SEA’s capacity estimation. The application of SEA extends to a comprehensive cost analysis of PV Solar Panel Installation, emphasizing a case study within the Solar Zone at The University of Arizona Tech Park. The study assesses the solar field’s theoretical solar energy harvest (AEP) for a 37.3-acre area, considering various tilt adjustment frequencies. SEA predicts annual energy harvests of 39.966 GWh, 42.958 GWh, 42.964 GWh, and 43.607 GWh for one-time, twice/year, quarterly, and monthly adjustments, respectively. The algorithm demonstrates proficiency in predicting potential plant capacity, estimating capacities of 15.68 MW, 16.85 MW, 16.85 MW, and 17.11 MW for the above-mentioned adjustments, respectively. Establishing a 6MW plant capacity aligned with the predicted maximum theoretical solar energy harvest, the study delves into the impact of degradation rates on Annual Energy Production (AEP). Over a 25-year lifespan, the plant is anticipated to produce 90% of its initial electricity, emphasizing long-term sustainability considerations. The Levelized Cost of Energy (LCoE) for the 6MW PV solar field is calculated as $44.05/MWh, $37.35/MWh, and $33.08/MWh for lifespans of 20, 25, and 30 years, respectively, aligning with industry benchmarks. The study highlights the significant influence of orientation and tilt angle on solar energy reception and LCoE. An in-depth exploration of Operations and Maintenance (O&M) costs highlights their significant influence on LCoE. With variations in O&M costs from $13 to $25 per kilowatt-DC per year, the study emphasizes the importance of strategic decision-making in optimizing revenue outcomes. The solar farm’s impact on residential energy needs is assessed, indicating its potential to power approximately 1577 homes in the first year and 1372 homes after 20 years. The 6MW solar plant’s economic viability is supported by a short payback period of 6.78 to 7.63 years. Combining SEA with economic assessments, this research provides a comprehensive understanding of solar energy technologies.

  PublisherDOI

• RealTCD: Temporal Causal Discovery from Interventional Data with Large Language Model

  arXiv (Cornell University) · 2024-04-23 · 2 citations

  preprintOpen access1st authorCorresponding

  In the field of Artificial Intelligence for Information Technology Operations, causal discovery is pivotal for operation and maintenance of graph construction, facilitating downstream industrial tasks such as root cause analysis. Temporal causal discovery, as an emerging method, aims to identify temporal causal relationships between variables directly from observations by utilizing interventional data. However, existing methods mainly focus on synthetic datasets with heavy reliance on intervention targets and ignore the textual information hidden in real-world systems, failing to conduct causal discovery for real industrial scenarios. To tackle this problem, in this paper we propose to investigate temporal causal discovery in industrial scenarios, which faces two critical challenges: 1) how to discover causal relationships without the interventional targets that are costly to obtain in practice, and 2) how to discover causal relations via leveraging the textual information in systems which can be complex yet abundant in industrial contexts. To address these challenges, we propose the RealTCD framework, which is able to leverage domain knowledge to discover temporal causal relationships without interventional targets. Specifically, we first develop a score-based temporal causal discovery method capable of discovering causal relations for root cause analysis without relying on interventional targets through strategic masking and regularization. Furthermore, by employing Large Language Models (LLMs) to handle texts and integrate domain knowledge, we introduce LLM-guided meta-initialization to extract the meta-knowledge from textual information hidden in systems to boost the quality of discovery. We conduct extensive experiments on simulation and real-world datasets to show the superiority of our proposed RealTCD framework over existing baselines in discovering temporal causal structures.

  PublisherOA PDFDOI

Frequent coauthors

• Ben Xu

  Mississippi State University

  40 shared

• Yuesong Shen

  26 shared

• Shemin Zhu

  Nanjing Tech University

  25 shared

• Jianzhi Li

  17 shared

• Shawn Hatcher

  Mississippi State University

  17 shared

• Mathew Farias

  Mississippi State University

  16 shared

• Zhiwei Xue

  Shandong University of Technology

  14 shared

• Cho Lik Chan

  University of Arizona

  13 shared

Education

• Ph.D., Energy and Power Engineering

  Xi'an Jiaotong University

  1995