A method is analyzed for the fault detection and isolation (FDI) based on hybrid diagnostic model tosolve the problem of fault detection and isolation. Furthermore, combining with hybrid diagnostic inference, themodified particle swarm algorithm (PSO) is used to improve the disadvantages of the original method. By theexperiment of polybox model, the modified method performs better in FDIand require for lower ability ofcomputation and takes less time for testing. The method has been tested in the system of assistant design fortestability (DFT) and analysis on some equipment.