The integration of distributed generation (DG) sources into the electrical grid, particularly renewable energy systems, has introduced a new paradigm in power generation. While DG systems offer numerous benefits, including enhanced grid resilience and reduced environmental impact, they also pose substantial challenges to power quality., the convergence of high capacity Distributed Generation (DG) and transient-heavy loads like Ultra-Fast EV Charging has created unprecedented Power Quality (PQ) instabilities. Traditional mitigation strategies, which once relied on static Instantaneous Symmetrical Component Theory (ISCT), are no longer sufficient for the dynamic, bidirectional flows of modern smart grids. This research work proposes a Unified AI-Driven DSTATCOM framework that operates in tandem with Smart Inverters across a Distributed Generation network. We introduce the 'Deep-Q-Power' algorithm, a reinforcement learning model that predictively adjusts DSTATCOM compensation parameters and DG inverter outputs simultaneously. This synergy reduces harmonic injection by 88% and maintains PCC voltage within a ±0.5% band even during massive load shedding or surge events. Real-world 2026 simulation data demonstrates a shift toward autonomous, self healing grid infrastructures.