Multi-anomalies self-potential inversion using enhanced Runge–Kutta optimization

Self-potential (SP) method has been widely used to identify some geological structures related to electrical potential. Multiple sources of anomalies give some difficulty in SP data inversion or interpretation due to a high number of model parameters. This paper proposes enhanced Runge–Kutta optimization (ERUN) using a rank-based mechanism and the chaotic logistic map to balance exploration and exploitation ability for multiple anomalies of SP data inversion in determining SP model parameters. ERUN is compared to the RUN algorithm on SP inversion problem with multiple anomalies. Model parameters are estimated using a posterior distribution model with the applied threshold. The result shows that RUN and ERUN can provide uncertainty analysis using the posterior distribution model. Furthermore, ERUN and RUN are applied to solve synthetic and field data. ERUN and RUN yield good results and show good agreement with geologic information. Model parameters generated by ERUN and RUN are appropriate for validation with prior researchers. LUSI embankment assessment using SP method analyzed using RUN and ERUN algorithm shows that ERUN has more reasonable result than RUN. Overall, ERUN is relatively more reliable in solving SP inversion problems with a relatively high number of anomalies than other algorithms tested in this research. This advantage comes with a relatively high computational cost, which makes ERUN slower.