A particle induction heating method to promote electrical heating performance for water is studied in this paper. The method sinks metal particles (we chose Nickel particles) inside water, and uses a RF oscillating magnetic field to heat the particle bed. The heat generated on the surface of particles by electrical eddy current further raises the temperature of water. Our experimental results show that this thermal process has a remarkable high heating rate owning to the huge heat dissipation area of particles (in some experiment cases, water boils in seconds, even though the volume ratio of particle to water is only 1:89). To describe the physical nature of the process, a mathematical model is proposed. And our numerical simulation results generated from the model agrees very well with experimental data. However, some parameters used in the model are not easily obtainable for engineering application. Thereafter, to predict the temperature for the heating process, in the end of this work, an artificial intelligent neural network architecture is further proposed.