Energy-efficient operation of neural operators for virtual sensing

2026-09-24

AI Summary

This study addresses the high energy consumption caused by redundant physical field reconstructions via neural operators in virtual sensing. To this end, it proposes an energy-efficiency optimization approach based on shared-space computation and graph replay. Explicit backbone reuse is achieved through DeepONet, Fourier Neural Operator (FNO), and compiler-level freezing, revealing how update frequencies and execution lifecycles influence computational reuse benefits while effectively distinguishing arithmetic reuse from launch overhead. Experimental results demonstrate that the proposed method achieves up to 20% energy savings under high request rates in heat exchanger services. Furthermore, in fixed-clock mode, it reduces energy consumption by approximately 22% compared to eager execution.

Abstract

Virtual sensing repeatedly reconstructs physical fields from changing observations, often on a fixed geometry. We investigate how shared spatial computation reduces the energy of these updates while retaining the selected checkpoint and its evaluated predictions. In a heat-exchanger service, standard compiler freezing and explicit trunk reuse give similar operating energy reductions relative to graph replay: approximately 1% at one request per second and 20% at forty requests per second. In 15 W mode with fixed clocks, reuse with graph replay completes the same request sequence with 22.0 to 22.5% less energy than eager execution, including preparation and waiting. DeepONet and Fourier neural operator (FNO) controls distinguish the effects of reusable arithmetic and launch overhead. Preparation, artifact construction, and worker replacement add costs outside repeated inference. These results connect operator structure to operating energy and show how update frequency and execution lifetime govern the benefit of computation reuse in physical-field virtual sensing.

Problem

• virtual sensing
• neural operators
• energy efficiency
• computation reuse
• physical field reconstruction

Innovation

• Neural Operators
• Virtual Sensing
• Energy Efficiency
• Computation Reuse
• DeepONet