A novel fault diagnosis method is proposed in this paper which is improved with DRFNN and PSO on basis of PF. There are much more algorithms about fault diagnosis because it has become the spot in intelligent control. As one of status estimation diagnostic methods based on the analytical model, particle filter fault diagnosis has been playing an important role in industrial production. However, it is limited by its shortcoming for its further development, a new particle filter fault diagnosis method based on hybrid algorithm is proposed in this paper, which retains the advantages of the particle filter, and it not only solved the problem of the weight degradation to a certain extent, but also improved particle swarm optimization. A system includes fault detection, prediction and recognized is realized with neural network in the beer fermentation temperature control system.