The limitation of conventional homogenization modelling used in the Mori-Tanaka Technique is the difficulty in representing the microstructural interactions between fiber and matrix. This causes differences between the predicted results and experimental testing of the mechanical properties of the composite, especially at high fiber volume fractions. This study is a modification of the Mori-Tanaka model using artificial intelligence (AI) and a correction to the Representative Volume Element (RVE) model to minimise the error between predictions and experimental testing. High-order polynomial regression and least squares optimization are the data processing methods used in this study to maintain computational efficiency while achieving good accuracy. This method offers a solution for characterising composite materials and improving the prediction of composite mechanical properties across various fibre volume fractions. This method is expected to be the basis for future studies on failure prediction, adaptive optimisation with AI, and nonlinear behaviour modelling.