Robust Estimation of Deformation from Observation Differences Using Some Evolutionary Optimisation Algorithms

Area of Science:

• Geodesy and Geomatics
• Computational Science

Background:

• Robust estimation is crucial for accurate deformation analysis.
• Traditional Iterative Reweighted Least-Squares (IRLS) struggles with global optima in datum solutions.
• Displaced points in datum networks challenge conventional methods.

Purpose of the Study:

• To present a modified Generalised Robust Estimation Of Deformation from Observation Differences (GREDOD) method.
• To integrate genetic algorithm (GA) and generalised particle swarm optimisation (GPSO) for robust displacement vector estimation.
• To overcome the limitations of IRLS in identifying global optimal datum solutions.

Main Methods:

• Application of GA and GPSO as global optimisation techniques within the GREDOD framework.
• Monte Carlo simulations for experimental analysis.
• Comparative analysis against IRLS and an Iterative Weighted Similarity Transformation (IWST) modification.

Main Results:

• The modified GREDOD method using GA and GPSO demonstrated superior performance.
• Increased overall efficiency by approximately 18% compared to traditional methods.
• Provided more reliable results for deformation analysis.

Conclusions:

• The proposed modification of GREDOD offers a robust and efficient solution for displacement vector estimation.
• GA and GPSO effectively address the global optimisation challenges faced by IRLS.
• The method is practically useful for engineering and geological deformation monitoring.