Multistate Diagnosis and Prognosis of Lubricating Oil Degradation Using Sticky Hierarchical Dirichlet Process–Hidden Markov Model Framework

Round 1

Reviewer 1 Report

My major concern with the paper is the definition of lubricating oil degradation. An increase of wear debris is not oil degradation. It is a contamination of the oil but do not give direct information on the oil condition. For example depletion of anti-wear additives, increase of acidity or the change of viscosity are oil degradation effects.  I propose to rename the title of the paper by replacing Lubrication Oil Condition with “ […] Prognosis of Wear Debris in Lubricating Oil using […]”.

The quality of the paper could be further improved if a brief explanation of the concept of HDP-HMM is added to the introduction section. This would be really helpful for readers who are not fully familiar with such kind of models.

Author Response

Please see the PDF attachment for the author responses to the reviewers' comments. Thank you.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper utilized a model (Sticky (HDP)-HMM) to analysis multiple wear states in the lubrication oil monitoring systems. The multiple wear states are unknown, un-predefined and are affected by the external interferences which are involved in practical or complex oil systems. The author utilized the method to predict residual life of an oil system where the external interferences are oil top-up and oil changes. 

Here are a few points I hope the authors could address: 

1. Please elaborate what is “wear-out phase” in 230, is it equivalent to “abnormal stage” by Fan et al. [33] ? What are other phases in an oil full cycles? Why “wear-out phase” is chosen in this study?

2. Oil change is represented with blue bar in Figure 6 (a), would it possible to explain what the physical meaning of other color bar shown in the Figure 6(a). 

3. In Figure 7 (a), Does the circle represents the state and the line between circle represents the transition between the states? Why there is a state “6” in the Figure 7 (a) and looks like there is no data points reaching state “2” and state “5”.  Does the oil change event being reflected in Figure 7 (a)? 

4. Line 455, “The reliability drops significantly around the 120th, 455 180th, and 240th sampling epochs when the oil is changed. “ did not see them showing in the figure 7(a). 

5. Could you explain why there are only 4  states for the data the Figure 2? Z = {1, 2, 3, 4} as shown in line 432. What are the factors that affect the number of states in a data set? It would be better if the author could provide another simulated data set that has a different sets of states (or more states).

6.  

   In addition to Oil change/ top off,  what are the other intentional or unintentional interferences or what are other examples of complex oil systems that could utilize this proposed method? 

Other small changes: 

1. Please supply the values of key parameters that generates the plot for Figure 2. The values could be supplied at line 231. 
2. Please supply 1-2 reference at line 245. 
3. Line 256 “over. time” -> “over time” 

Author Response

Please see the PDF attachment for the author responses to the reviewers' comments. Thank you.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Thank you for the revision. 

A few more minor suggestions: 

1. For concern #1 in my last comments, please add  "abnormal state in [33]" to make the terminology clear and consistent to wider audience.
2. You explained the motivation to use "wear out state" in your reply, I think it would be good to add that explanation to the paper. I guess some audience may have that "why at wear out state" question as I do. 

Author Response

Please see the PDF attachment for the response letter.

Author Response File: Author Response.pdf