A Study of Seating Suspension System Vibration Isolation Using a Hybrid Method of an Artificial Neural Network and Response Surface Modelling

Round 1

Reviewer 1 Report

In this paper, the authors applied the ANN algorithm for studying of seating suspension system. Although the topic seems interesting and relevant to the journal's scope, still the authors should address the following issues:

1. The literature review section is missing, you should review the most recently published relevant papers and clearly need to address the research gaps and contributions of this paper. You may also cite this paper:  https://arxiv.org/ftp/arxiv/papers/2305/2305.13207.pdf

2. The methodology section is missing and you should describe the methodology with a clear block diagram.

3. Dataset descriptions are also missing here, which is very important for the reader. 

Author Response

Editor: Similarity 31% has to be reduced.

Response: Similarity of our manuscript ha sbeen reduced from 31% to 18% (not including references).

Review 1

Q1: The literature review section is missing, you should review the most recently published relevant papers and clearly need to address the research gaps and contributions of this paper. You may also cite this paper:  https://arxiv.org/ftp/arxiv/papers/2305/2305.13207.pdf

A1: I have selected and summarized three relevant studies from 2018 to 2020. The overall Introduction paragraph has been revised to include more detailed research objectives and describes the motivation for the study。The the research gaps and contributions of this paper have been addressed. Unfortunately the suggested paper is not relevant to this manuscript.

Q2: The methodology section is missing and you should describe the methodology with a clear block diagram.

A2: The content has been adjusted, with titles clearly indicating the ANN (Artificial Neural Network) model and RSM (Response Surface Methodology) model. In the second paragraph, the research methodology is illustrated with a clear block diagram.

Q3: Dataset descriptions are also missing here, which is very important for the reader.

A3: Dataset descriptions have been amended. I have revised the data presentation, attaching detailed RSM (Response Surface Methodology) model data for discussion and summary purposes.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper aimed to develop a hybrid method of artificial neural network and response surface modelling to predict the peak seat-to-head transmissibility ratio of a seating suspension system and to evaluate its ride comfort for different seat design parameters. The results showed that the artificial neural network model was able to simulate the dynamic characteristics of the 5-DOF passenger seat suspension system model. In general, this is an interesting and well written manuscript. The specific comments are as follow.

1. Line 148 and 149. The paper mentioned the design parameters and the output response target values were from the 5-DOF model in the previous study, but the mentioned model cannot be found in the referenced paper. Please check the reference and give a brief description of the 5-DOF model.

2. Please explain why the peak vibration transmissibility at 4 Hz was chosen as the predicted output. From the reviewer’s perspective, it’s related to the ride comfort. But why not use the evaluation method described in ISO 2631-1.

3. If the parameters and outputs of the 5-DOF model are used as the input and output of the neural network, there should theoretically be a lot of samples. Why are there only 26 data samples in the paper? Because at Line 240, the paper discussed “if more than 26 data points are used to train the ANN model, the accuracy of the ANN model will become even better”.

Author Response

Review 2

Q4: Line 148 and 149. The paper mentioned the design parameters and the output response target values were from the 5-DOF model in the previous study, but the mentioned model cannot be found in the referenced paper. Please check the reference and give a brief description of the 5-DOF model.

A4: In the method section, detailed descriptions of the models and source citations have been added.

Q5: Please explain why the peak vibration transmissibility at 4 Hz was chosen as the predicted output. From the reviewer’s perspective, it’s related to the ride comfort. But why not use the evaluation method described in ISO 2631-1.

A5: I have explained why the peak vibration transmissibility at 4 Hz was chosen as the predicted output in the introduction. The reason is as follows: Under low-frequency vibrations at 4 Hertz (Hz), resonance of the human body and seat sytem can severely affect the comfort of sitting passenger and may lead to various adverse physiological and psychological effects.

Firstly, prolonged exposure to low-frequency vibrations can cause bodily pain and discomfort, especially in the back, neck, and lumbar regions. This continuous vibration can exacerbate muscle tension, potentially leading to long-term muscle pain and joint issues.

Secondly, sustained vibrations may affect internal organs, particularly in the abdominal and thoracic areas, leading to indigestion, stomach discomfort, and even respiratory problems.

Additionally, being in such a resonant environment for extended periods can trigger or exacerbate headaches, dizziness, and nausea, especially in those who are more sensitive to low-frequency vibrations.

From a mental health perspective, continuous discomfort can lead to anxiety, irritability, and difficulty concentrating, and over time, may cause sleep disorders.

Lastly, for individuals who need to sit for extended periods, such as drivers and office workers, long-term exposure to such environments can lead to a decrease in overall work efficiency, affecting job performance and quality of life.

Therefore, considering these potential adverse effects, reducing or controlling exposure to 4 Hz low-frequency vibrations is an important measure to ensure seating comfort and prevent health issues.

For simplicity, we chose the peak vibration transmissibility at 4 Hz rather than the evaluation method described in ISO 2631-1.

Q6: If the parameters and outputs of the 5-DOF model are used as the input and output of the neural network, there should theoretically be a lot of samples. Why are there only 26 data samples in the paper? Because at Line 240, the paper discussed “if more than 26 data points are used to train the ANN model, the accuracy of the ANN model will become even better”.

A6: Due to the characteristics of neural networks, the accuracy of the model will be enhanced provided that the more input and output data results are used to train the neural network model. In order to verify the neural network model, RSM model has been developed where 26 input parameter combinations were used. I have revised this section of the content and provided an explanation.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I do not have further comments and good to see the authors responses in the revised draft

No major issues, but the authors should carefully check the minor grammar issues