About Me

I am Ziheng Li, a Ph.D. student at Harbin Institute of Technology, Shenzhen, focusing on data-efficient battery health prognosis under real-world constraints. My research aims to address key challenges arising from practical battery applications, including limited training samples, fragmented charging observations, and distribution discrepancies across different operating conditions and battery types.

My work centers on three methodological directions: few-shot learning, generative modeling, and early-cycle prediction. Specifically, I design meta-learning and domain adaptation strategies for cross-domain SOH/RUL estimation; develop generative models—such as conditional diffusion models—to reconstruct complete battery curves from partial charging segments; and build early prediction frameworks capable of assessing battery health using only partial cycle data.

The overarching goal of my research is to develop robust, data-efficient, and deployable battery prognostic algorithms that support electric vehicles, large-scale energy storage systems, and emerging battery technologies.

Research Interests

Theory

• Few-shot Learning
• Early Prediction
• Generative Learning

Application

• Battery Prognostics & Health Management
• Energy Storage System Control

Current Research Focus

Data-Constrained Battery Health Prognosis

Real-world battery testing often suffers from limited data availability, incomplete charge-discharge cycles, and cross-domain discrepancies. To address these challenges, my research investigates:

• Few-shot learning for rapid adaptation under scarce data
• Generative diffusion models to reconstruct full battery voltage–time curves from fragmented segments
• Early-cycle prediction to achieve SOH/RUL estimation with minimal operational data
• Domain-invariant representation learning to generalize across batteries, protocols, and working conditions

These data-efficient approaches aim to reduce testing cost, enhance reliability, and accelerate battery deployment in EV and energy storage applications.

Contact

• Email: 25b353022@stu.hit.edu.cn
• GitHub: https://github.com/zihenglineu-eng
• Affiliation: Harbin Institute of Technology, Shenzhen