A Fault Detection Method Based on Oriental Hyper-bounding Box

Most of the traditional fault detection methods based on MSPM assume that the data obey multivariate Gaussian distribution when designing statistics and determining control limits. Aiming at this limitation, this paper proposes a fault detection method based on oriental hyper-bounding box(OHBB). Firstly, the definition of OHBB is given, and it is pointed out that the minimum volume OHBB can be obtained by using the directions of all principal components of random vector as the directions of OHBB. Then, an optimization problem is proposed to solve the size and location of soft boundary OHBB. After the off-line modeling is completed, the on-line fault detection can be realized by judging the position relationship between the new sample point and the soft boundary OHBB. Finally, the effectiveness of the proposed method is verified by numerical simulation and TEP.