Research on CNN-LSTM Brake Pad Wear Condition Monitoring Based on GTO Multi-Objective Optimization

Round 1

Reviewer 1 Report

In this study, the authors proposed a CNN-LSTM brake pad wear condition monitoring method based on the GTO multi-objective optimization. And the proposed method investigated the lab test using an experimental rig. However, there are several issues fully explained and revised before publication.

(1) GTO is described as the artificial Gorilla force algorithm and also as the artificial Gorilla Army Optimization algorithm. The meaning is not quite different, but it is recommended one unified expression.

(2) In Line 429, the authors mentioned, “In this paper, the original data of the above four braking parameters ~.” But in the above paragraph, only three parameters were introduced, including brake disc speed, brake pressure, and brake disc temperature. So it is necessary to explain what the fourth parameter is.

(3) In Line 433, the authors mentioned, “the number of braking times in ∆t time was 300.” It is necessary to provide the unit of time such as 300 m, 300 s, 300 revolutions, or 300 braking.

(4) In Figure 5, the brake pad wear condition is compared with braking speed, braking pressure, and braking temperature. Usually, the cause and effect are compared in the scatter diagram, with the cause on the x-axis and the effect on the y-axis. It is not easy to understand the meaning of Figure 5. Moreover, it is not described the physical parameters and corresponding units for x- and y-axes.

(5) In this paper, the braking pressure has no signification influence on the brake pads' wear. As well known, the braking pressure can be easily converted to the dynamic friction force by multiplying the area and the friction coefficient. And the friction force must have a very significant influence on the brake pads' wear. So it is not appropriate to conclude that the braking pressure is not affecting the wear from the limited experiment with a pressure range between 0.8 MPa and 1.6 MPa.

(6) Figure 7 is not enough to explain the CNN-LSTM model structure. It looks like a combination of functions.

(7) In Equation (28), what is y-bar?

(8) What is the unit of training time in Figure 8? 

(9) In Figure 9, it is unclear that the red line represents the equation y=x.

(10) In Figures 10, 11, 12, and 13, the predicted values are compared with the actual values. However, it is better to use a scatter diagram with actual values on the x-axis and predicted values on the y-axis. It is also recommended to provide the units.

(11) It is not sure whether the same training, validation, and testing data are utilized for multi-objective optimization. It is rational that the optimization is carried out using the same training and validation data. However, the testing data may not properly be utilized during optimization.

(12) There are several parts to be corrected as follows

Line 428: 

Table 1. Selection range of brake parameters.

-> The title of Table 1 is detached from the table.

 

Line 422:

brake pressure in 0.8Mpa to 1.6Mpa,

-> brake pressure in 0.8MPa to 1.6MPa,

 

There are several parts to be revised, including Line 419.

The brake disc speed refers to the relative speed of sliding friction between the brake pads and the brake pads.  

-> The brake disc speed refers to the relative speed of sliding friction between the brake pads and the brake disc.  

Author Response

Reviewer 1

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors present a CNN-LSTM based, brake pad wear state monitoring methodology. They employ the Convolutional Neural Network part to complete the deep mining of brake pad wear characteristics, apply data dimensionality reduction, and uses Long Short- term memory to capture the time dependence of the pads wear sequence. In this way they find a nonlinear mapping relationship between brake pads wear characteristics and brake pads wear values. Next the artificial Gorilla Army Optimization algorithm is applied to perform multi-objective optimization of the network parameters. The results show that the CNN-LSTM model based on GTO multi-objective optimization can effectively monitor the wear state of brake pads.

The authors are congratulated on their meticulous work to attack a subject of significant importance to the automotive industry that assists in the continuous search for more ecological brake pad materials with predictable wear characteristics.

The following improvements are suggested for a revised version:

Literature review needs a little expansion with similar works.

Figure 3 improve resolution

Line 350 you mean the random number function? Please explain

Section 2: please give more details on the implementation of the models, i.e. Python language? Matlab? Which libraries employed, etc.

Line 428 correct table caption position

Lines 638-646 Relying just on a prediction to issue an alarm would not be wise. Better to combine with the usual wear indicator’s noise. Please comment.

 

Author Response

Reviewer 2

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

All my comments on the first submission are well-solved, and no more correction is necessary.

Reviewer 2 Report

The revised version is OK for publication