This project aims to develop an intelligent control and energy management system for electric bicycles by integrating Physics-informed Neural Network (PINN) technology, addressing the industry's shift toward digital transformation and high-value-added products. To overcome the limitations of traditional AI models namely their heavy reliance on massive datasets and lack of physical

interpretability. this research embeds physical equations, including motor electrical characteristics, rotational dynamics, and the law of conservation of energy, into a deep learning framework to establish a science-based intelligent model.

The core of the project adopts a hybrid architecture of "PINN-assisted conventional control." At the operational level, the system maintains the real-time safety of traditional FOC-PID controllers while utilizing the PINN model as an intelligent intermediary layer. This layer is responsible for the dynamic identification of motor parameters (such as the torque constant) and load prediction, achieving precise feed-forward compensation and reducing lag errors. Furthermore, the project introduces "energy consistency residuals" as a loss function. By monitoring the balance between electrical energy and mechanical work, the system realizes high-accuracy State-of-Health (SOH) assessments for batteries and reliable cruising range predictions. Through integrated hardware and software validation, this project is expected to shorten the development cycle of motor drive systems and enhance energy efficiency under limited computational resources. The results will not only assist manufacturers in developing differentiated intelligent e-bikes but also support national net-zero emission policies, laying a critical technical foundation for future smart transportation and energy-saving industries.

Start Date: Flexible (to be determined based on the selected candidates)

Eligibility: The project is open to up to 2 undergraduate students

Skills and Knowledge Required: Familiar with Python. PINN and deep learning frameworks, with a strong emphasis on solid theoretical foundations, particularly in mathematics and mechanics.

Financial Support: Up to 15,000 TWD per month (approximately 450 USD).

Application Process: Interested applicants should submit:

1. A resume or CV detailing relevant skills and experience

2. A cover letter expressing interest and motivation for the project

3. Transcripts (for students)

How to Apply: Please send your application to jct@stust.edu.tw with the subject line: “TEEP Project_(Your Name)”. We look forward to receiving your application. (Structural Dynamics and Vibration Laboratory)