Transcranial Magnetic Stimulation (TMS) is a non-invasive neuromodulation technique that stimulates the brain using an induced electric field generated by a stimulation coil and pulsed current. The precision of this stimulation largely determines its effectiveness. This paper proposes an intelligent waveform optimization method that utilizes intelligent algorithms combined with real brain structures to optimize stimulation waveforms, innovatively improving stimulation accuracy from the temporal scale while also guiding the construction of TMS circuits. First, a selectivity index is established based on brain structure and neuron morphology. Then, particle swarm optimization is employed to identify waveform parameters that optimize the selectivity index, followed by the construction of an experimental platform to implement the optimized waveforms. The optimized waveforms significantly improve stimulation selectivity, i.e., precision, and the use of intelligent algorithms offers new possibilities for developing personalized treatment plans. Additionally, the insights gained from the waveform effects during the optimization process lay the groundwork for further waveform and circuit optimization.