Authors:

Sebastian Hurtado

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Modern power systems require fast and accurate solutions to complex operator equations involving continuous signals, yet classical numerical methods remain computationally prohibitive for real-time applications. To address this challenge, we propose a quantum machine learning framework based on Deep Operator Networks (DeepONets) that leverages quantum computing principles, such as superposition and probabilistic measurement, to improve computational efficiency and uncertainty quantification in operator learning tasks. We explore three classical-to-quantum data encoding strategies: unitary encoding, which encodes classical data into unitary matrices applied as quantum gates; angle encoding, mapping real-valued data into qubit rotation angles; and Fourier encoding, representing input functions via truncated Fourier series with coefficients encoded as quantum states. Furthermore, we develop diversified ensembles of quantum DeepONets formed via random initialization and data shuffling, and introduce a correction term to mitigate variability in quantum circuit outputs when estimating uncertainty intervals. Model performance is assessed using mean squared error, relative L2 error, and uncertainty coverage on complex operator learning problems. This approach demonstrates the potential of conformalized quantum DeepONet ensembles as a scalable and reliable method for transient operator learning in power systems. Keywords: Quantum DeepONet; Uncertainty Quantification; Power System Operator Learning

Source:

Purdue University / 2025

Topics:

No topics listed

Co-authors:

Sebastian Hurtado