Fault Diagnosis of Blast Furnace Based on SVMs

For the complexity and confusion in fault diagnosis of blast furnace, a new artificial intelligent algorithm is presented to solve this problem. The Support Vector Machines (SVMs) are one type of large margin classifier based on statistical methods. With the property of dealing with high dimension data, studying small quality of samples and training large data sets, it is feasible to use it to make classifier. By using one-against-one strategy, construct it with different kernel functions, and select the best one for the classifier. The simulation results show that using polynomial function is superior to the others. Besides, compared with expert system and neural networks, SVMs have a better performance on recognizing patterns, capabilities of fault-tolerance and generalization.