Abstract: In the process of wire electrode discharge grinding (WEDG) of insulating engineering ceramic, there is a close relationship between the technical indexes and process parameters. The operators can only set the process parameters according to their past experience during the practice, and predict the processing results to a certain extent. If the setting of process parameters is unreasonable, it will greatly affect the processing efficiency, accuracy and capacity of machine tools. Therefore, based on BP fuzzy neural network(BPFNN), a prediction model is presented for the effect of technical indicators for WEDG of insulating engineering ceramic. Rough set theory is used to reduce the attributes and rules of training samples and improved particle swarm optimization (PSO) is used to optimize the network. The models before and after optimization are used to simulate the processing of boron carbide (B₄C) ceramics and the results are compared. It is found that the optimized model has the advantages of fast prediction speed, small error and high precision.