I am a Master's student in Software Engineering, focusing on Non-Intrusive Load Monitoring (NILM), Smart Grid, Energy Internet, and time-series deep learning.

My current research explores how synthetic data generation, cross-domain adaptation, and efficient Transformer architectures can improve appliance-level load perception under data-scarce and resource-constrained scenarios.

• Non-Intrusive Load Monitoring (NILM)
• Smart Grid and Energy Internet
• Time-Series Representation Learning
• Synthetic Load Data Generation
• Transformer and Diffusion Models
• Parameter-Efficient Fine-Tuning for Edge-Oriented NILM

• Programming: Python, SQL, Markdown, LaTeX
• Deep Learning: PyTorch, CNN, RNN/LSTM/GRU, Transformer, Diffusion Models
• Data Analysis: Pandas, NumPy, Matplotlib
• Research Tools: Git, GitHub, Overleaf, Anaconda, PyCharm
• Application Areas: NILM, Smart Grid, Energy Internet, Time-Series Modeling

SynerBETA-NILM is a two-stage framework for non-intrusive load monitoring. It combines synthetic appliance-level load generation with parameter-efficient cross-domain load disaggregation.

Keywords: NILM, Smart Grid, Energy Internet, Time-Series Modeling, Diffusion Model, Transformer, PEFT

SynerDiff is a data augmentation model designed for appliance-level power signals. It combines an autoregressive Transformer with diffusion modeling to capture both steady-state trends and transient load variations.

Keywords: NILM, Time-Series Generation, Diffusion Model, Transformer, Data Augmentation

BETA-NILM is a Transformer-based load disaggregation model that integrates adapter-based parameter-efficient fine-tuning and efficient attention for cross-domain NILM.

Keywords: NILM, Adapter, PEFT, Efficient Attention, Domain Adaptation

A practical power load perception project for heterogeneous scenarios such as convenience stores, filling stations, and gyms. I participated in data governance, sample library construction, feature design, and model training.

Keywords: Power Data Processing, Load Perception, Data Governance, Energy Management

• SynerBETA-NILM: A two-stage framework for synthetic load data generation and parameter-efficient cross-domain non-intrusive load monitoring.
  Manuscript in preparation / under submission.

• Name: Peisheng Yang
• GitHub: @AKAYoung
• Research Direction: NILM, Smart Grid, Energy Internet, Time-Series Modeling