<p>Distance relays play an important role in protecting transmission lines from faults. The conventional distance relay calculates the impedance of the line up to the fault point by assuming that the system is homogeneous. However, the integration of inverter-based resources (IBRs) into the grid, assuming a homogeneous system, leads to significant errors, causing the relay to maloperate. The distance calculation error is further influenced by the reduction in the system strength, variability in fault resistance, and fault location. This study proposes a modified distance-relaying algorithm to address the above challenges while considering the non-homogeneity of the system. The proposed scheme works with measurements only at relay locations. This method uses superimposed components and deep learning techniques to calculate impedance accurately up to the fault points. Extensive simulations and hardware implementations were conducted using MicroLabBox with an NVIDIA GPU-based relay. The experimental results demonstrate the efficacy of the GPU-based relay, highlighting the practical applicability of the proposed solution.</p>