Effects of using non-uniform magnetic field in a PCM installed 3D vented cavity having triangular-cross section on the phase transition dynamics are numerically assessed. Alumina-copper nanoparticles are used in water as the heat transfer fluid. Numerical study is conducted for different Reynolds number (Re-between 100 and 500), Hartmann number (Ha-between 0 and 80), solid volume fraction of particles (svf-between 0.001 and 0.02), amplitude of non-uniform magnetic field (Amp-between 0.2 and 1) and aspect ratio of the cavity (AR-between 0.4 and 2). Phase transition becomes accelerated with higher Re, Ha, Amp, svf and AR. The reduction in phase transition time (tf) becomes 73% when lowest and highest Re cases are compared while by using nanoparticles at the highest amount, up to 42.9% reduction in complete transition time is obtained. For the highest amplitude of non-uniform magnetic field, 6.9% and 15% reduction amounts are obtained at Ha = 20 and Ha = 80. Faster transition times are observed with higher aspect ratio. Artificial neural networks are used for the dynamic characteristics estimations of phase change process while with 10 neuron in the hidden layer, successful estimations are obtained with network modeling approach.