Under partial shading conditions, photovoltaic arrays exhibit multi-peak output characteristics, This phenomenon potentially induces conventional MPPT algorithms to stagnate at local power maxima while exhibiting oscillatory output characteristics, thereby compromising photovoltaic system operational stability. To address this issue, this study proposes an enhanced MPPT control strategy that integrates Q-learning with an improved sand cat swarm optimization algorithm. The proposed approach incorporates three key improvements: First, the Levy flight strategy is employed to prevent premature convergence to local optima. Second, the exploration capability of the algorithm is strengthened by introducing mechanisms from the grey wolf optimizer. Third, Q-learning dynamically adjusts inertia weights during the optimization process, enhancing both convergence speed and accuracy. Comparative experiments with conventional methods—including the sand cat swarm algorithm, particle swarm optimization, cuckoo search combined with perturbation observation, and the whale optimization algorithm—demonstrate that the Q-learning-enhanced sand cat algorithm achieves the fastest global search speed. Additionally, it effectively minimizes power oscillations and improves the overall efficiency of photovoltaic power generation.