Exploring advanced diagnostic methods for High Voltage Direct Current (HVDC) systems is essential to swiftly identify and address faults. A robust fault location system that can accurately pinpoint issues is crucial for minimizing downtime and improving overall system reliability. To achieve this, we propose a novel approach which involves leveraging a heterogeneous combination of machine learning and curve-fitting methods. This innovative method addresses the complexity and non-linearity of fault location and detection by utilizing machine learning algorithms to identify patterns and anomalies, while curve-fitting techniques, along with trust region optimization, refine these predictions to enhance precision. The synergy between these techniques not only improves fault location accuracy but also ensures rapid response times, thereby significantly enhancing the reliability and efficiency of HVDC systems. Results from the comparative analysis demonstrate that the proposed method, using peak current signals sampled at 135 kHz, achieves the lowest error rate max average error MAE of 0.316% , and coefficient of determination of 0.99 when the range of maximum average error in existing fault location methods is between 0.78% and 5.69%.