Cuckoo Search Algorithm-assisted inversion for estimating petrophysical characteristics using well-logging data

Abstract This paper presents a CSA-DLSQ hybrid inversion that uses Cuckoo Search Algorithm (CSA) which provides better estimation for petrophysical properties. This hybrid technique significantly reduces computational costs while preserving the global convergence properties of CSA and the fast convergence of DLSQ. Inversion is realized of depth-local petrophysical parameters and zone-dependent global variables for facilitating geological consistency in layered formations. A single-layer synthetic model is first used for investigating CSA's hyperparameter sensitivity (population size, discovery rate P a, Lévy exponent β ) and to analyze the algorithm’s dynamic search behavior. An extension to a four-layer model with global cementation exponent is then used to reduce parameter oscillations at layer boundaries. The hybrid algorithm successfully stabilizes the inversion process and converges faster than stand-alone CSA. The method is used for well-log data observed in a Hungarian borehole, resulting in a 56.6% reduction in total misfit and ~ 87 × per-point speedup, generating petrophysical parameter distributions conforming with lithological interpretation. The hybrid inversion is more accurate and efficient than stand-alone CSA, especially for zones of complex geology or noisy measurements. The results confirm that the algorithm developed offers a robust and scalable framework for petrophysical interpretation within conventional hydrocarbon reservoirs.