This paper presents a local multivariate statistical modeling method based on PCA for nonlinear and dynamic processes. It focuses on a popular research topic in the related fields, and there still exist several improvements in order to satisfy the requirements for the acceptance for publications. More details are listed as follows:

1. Several MSPM methods are mentioned in the introduction, but no further analysis on the reason why they choose PCA as the basis of their research topic is provided. The authors should make the advantages of PCA on their research topic clearer.

2. Kernelization is applied to the proposed algorithm, but no further discussion is proposed in the introduction. The authors need to add the related content, such as the advantages of different kernel tricks, the reason why the authors choose Gaussian kernel.

3. The sensitivity-based contribution plot method is adopted in the proposed approach, and the related literature review should be supplemented in the introduction.

4. Some typos in the manuscript need to be corrected. For example, “yield” in the third point in highlights is supposed to be “yields”. “max JLDKPCA (p)” in Equation (14) should be “max JLDKPCA (α)”.

5. In terms of monitoring performance, the authors only compare FARs of DKPCA with that of LDKPCA. What about their superiority to other algorithms mentioned in the introduction such as PCA, PLS, ICA, KPCA? Moreover, other metrics of monitoring performance should also be considered for a comprehensive comparison.

My comments are well addressed.