Dynamic Process of Quality Abnormal Pattern Recognition Based on PCA-SVM

Quality abnormal pattern recognition for dynamic process is the key problem to achieve the online quality control and diagnose of automatic production. In the practical applications, there are some existing problems such as computational complexity and low recognition accuracy. A recognition method for quality abnormal pattern of dynamic process with PCA-SVM was proposed. This paper proposes a feature selection technique that employs a principal component analysis, to avoid this information loss. Then, the extracted features were treated as input vector for SVM classifier, following a particle swarm optimization algorithm is proposed to improve the generalization performance of the recognizer. Simulation results show that the proposed algorithm has very high recognition accuracy and high generalization ability. It is significant for quality monitoring and diagnosis in manufacture dynamic process.