Fault diagnosis of chemical process based on factor analysis and improved K-neighbor algorithm

It has been more and more difficult to just depend on the mechanism models to diagnose the faults of the increasingly complex chemical system. This paper proposes an improved K-neighbor algorithm based on data driven method for process fault diagnosis. For training data, the fault diagnosis accuracy rate is seen as the standard to optimize the value of K. The choice purpose of this improved algorithm is to eliminate the value of K's effect on consequence. In order to improve the fault diagnosis accuracy and reduce the calculation time, this paper integrates factor analysis and improved K-neighbor algorithm. Through TE process, this paper verifies the effectiveness of the proposed method. The fault diagnosis accuracy is increased from 78.7% to 86.7% and the calculation time is reduced 52%.