View full-text article in PMC Sensors (Basel). 2025 Jun 6;25(12):3588. doi: 10.3390/s25123588 Search in PMC Search in PubMed View in NLM Catalog Add to search Copyright and License information © 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). PMC Copyright notice Algorithm 1 DeepONet Require: Dataset v(i)(x),x(i),u(i)i=1N ▹x=(x, y, z, t) Require: Branch Net B:Rm→Rq, Trunk Net T:R4→Rq ▹m=dimension of v(i) 1: Initialize weights θb,θτ 2: for epoch=1 to E do 3: for i=1 to N do 4: B(i)←Bv(i);θb ▹ Branch input from discretized v(x) 5: T(i)←Tx(i);θτ ▹ Trunk input: x=(x,y,z,t) 6: uˆ(i)←∑k=1q Bk(i)⋅Tk(i) ▹ Operator prediction 7: if Physics-informed then 8: RQ(i)← PDE residual at x(i) 9: BQ(i)←B/ IC residual at x(i) 10: L←λ1RQi2+λ2BQi2 11: else 12: L←MSEuˆ(i),u(i) 13: end if 14: Update θb,θτ via backprop on L 15: end for 16: end for