Fault diagnostics of blast furnace based on NN-DPSO-SVM

Since fault diagnosis of blast furnace is very important in manufacturing, in this paper, a new strategy based on NN-DPSO-SVM is proposed to solve it. Using the nearest neighbor principle deletes the useless samples so that the training set is pruned. A modified discrete particle swarm optimization is applied to optimize the feature selection and the SVM parameters so that the algorithm can improve the performance of the SVM classifier and increase the recognition accuracy. Binary tree classification model is constructed rationally to solve the multi-class fault diagnosis. The new method in this paper can select the best fault features in much shorter time and have fewer errors and better generalization performance in the application of fault diagnosis of the blast furnace.