I propose an approach that we use in the research work of Mr. Khaled ouni, based on the detection and isolation of faults based on the analysis of nonlinear main components (NNLPCA) and a minimum partial quadrant (PLS ). The analysis of the main components is used for the development of industrial process performance monitoring systems. However, it is a linear technique and, in this respect, it is not necessarily the most appropriate methodology for dealing with industrial problems that exhibit non-linear behavior. PLS-2 is applied to obtain a clear view of the failures in the 2-D space. Finally, faults are diagnosed by contribution plots for the default class. A new process monitoring method is proposed and is applied to the defect detection of a manufacturing process. The performance of the proposed approach is then illustrated and compared with that of classical LPCA

Thanks Mr. nabil. Could you kindly send me links of reference paper artical which could helpful for the above research. 

Maybe the week 9 of Andrew Ng's Coursera Machine Learning Course could be helpful.

https://www.coursera.org/learn/machine-learning#

Week 9 is "anomaly detection", treating the basic concepts of how to approach the fraud detection part of your task..

Dear Hiran, 

I propose the paper containing the approach presented in the research above.

• THANKS Mr. Nabil. Thanks Horn. for your contribution. 

Not at all