Aiming at the problem of low recognition efficiency in the extreme learning machine (ELM) recognition model caused by the random setting of input weights and bias vectors in the quality abnormal pattern recognition of the manufacturing process,the network structure of the ELM recognition model was optimized by using particle swarm optimization algorithm,a manufacturing process quality monitoring model was proposed based on improved ELM.Feature extraction of process quality data was carried out by using principal component analysis.The recognition model was trained by using principal component features.Particle swarm optimization algorithm was used to optimize the network structure of the recognition model.The simulation experiments and measured data indicate that the recognition efficiency of the proposed abnormal identification model in the manufacturing process based on the improved ELM is significantly higher than other similar models,this improved model can be used for real-time monitoring of the manufacturing process.