MODELLING OF SCLUMBERGER SOUNDING CURVE USED STANDARD AND VARIANTS OF FLOWER POLLINATION ALGORITHM

• Main Author: Raflesia, Farkhan
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/53719
• Institution: Institut Teknologi Bandung

In the last decade, much of the work done on DC resistivity has mainly concentrated on 2D and 3D techniques. However, the results of the 1D inversion are very useful in building initial models for multidimensional interpretation. Therefore the study of the uniqueness of 1D inversion is important. There are inherent problems that can cause parameter estimation errors in complex geological arrangements, namely equivalence and suppression. The purpose of this study was to analyze the performance of FPA standards and FPA variations consisting of Modified Flower Pollination Algorithm (MFPA), elitism Flower Pollination Algorithm (eFPA), Dimension by Dimension Improvement Flower Pollination Algorithm (DDIFPA), and Flower Pollination Algorithm with Bee Pollinator (BPFPA) to interpret a schlumberger sounding curve. Also Vary the FPA to solve equivalence and suppression problems. The data used in this research is synthetic data which contains noise and free noise. Field data is also used to check the program in real data. Compared to the DLSQR inversion programs, PSO and GWO, FPA can obtain more accurate results and provide a better RMS error. Compared to MFPA, DDIFPA, BPFPA, and FPA, eFPA is the only algorithm that can achieve global optimal. Besides having the best level of accuracy, eFPA also has the best stability. In the case of field data, eFPA has the same results as the FPA standard which is better than IPI2Win based on RMS error. The equivalence associated with a thin conductive layer can be solved better than that for a thin resistive layer. Meanwhile, the suppression associated with a model that decreasing resistivity with depth can be solved better than for a model in which resistivities increase with depth.