A Novel Feature Selection Based on VMD and Information Gain for Pipe Blockages

Round 1

Reviewer 1 Report

I have reviewed the paper “A novel feature selection based on VMD and Information Gain for pipe blockages” by Xuefeng Zhu, Zao Feng, Jiande Wu and Weiquan Deng. The paper needs some revisions to be published in Journal of Applied Sciences. Here are my comments.
1- Please show more figures of the experimental tests.

2- It is needed to improve the quality of Figures 1 and 3.

3- Additional descriptions should be provided on the Figure 4.

4- The literature review in introduction is thorough and it is well written, however six additional references listing below regarding to vibration and signal processing could be added in the introduction. 

"Health monitoring of pedestrian truss bridges using cone-shaped kernel distribution." Smart Structures and Systems 22, no. 6 (2018): 699-709. "Application of power spectral density function for damage diagnosis of bridge piers." Structural Engineering and Mechanics 71.1 (2019): 57-63. "A novel damage identification method based on short time Fourier transform and a new efficient index." Structures, vol. 33, pp. 3605-3614, (2021). "Power quality disturbance recognition using VMD-based feature extraction and heuristic feature selection." Applied Sciences 9, no. 22 (2019): 4901. "Effective crack damage detection using multilayer sparse feature representation and incremental extreme learning machine." Applied Sciences 9, no. 3 (2019): 614. "Validation of Acoustic Emission Waveform Entropy as a Damage Identification Feature." Applied Sciences 9, no. 19 (2019): 4070. 

Author Response

Response to Reviewer #1

Comment 1. Please show more figures of the experimental tests.

Response: Thank you for your comment. In the document, we show more figures of the experimental tests. We supplement the multi classification confusion matrix of four types of classifiers. This part is in 560-631 lines of the manuscript.

The details are as follows：

In order to further investigate the ability of VMD-IG-RMSE algorithm to identify pipe operational conditions and the details of blockage misjudgment, a multi-classification confusion matrix is introduced to quantitatively analyze the results of pipe operational conditions recognition in detail. The confusion matrix comprehensively reflects the recognition accuracy and number of misjudgments of pipelines at different blockage levels, as well as the misjudgment type of real blockage type. The quantization diagram of the confusion matrix of four classifiers KNN, SVM, ELM and RF is shown in Figure 10.

Figure 10. Confusion matrices for four classifiers:(a) KNN; (b) SVM; (c) ELM; (d) Sound pressure level conversion

In Figure 10, the X-axis represents the target class of pipe operational conditions, and the Y-axis represents the output label of pipe operational conditions. The number of test samples in each type of pipe operation conditions was 100, and there were 9 types of operation conditions. The numbers 1-9 represent the nine pipe operational conditions. The numbers on the main diagonal represent the number of samples correctly identified by VMD-IG-RMSE algorithm for the operation conditions of each type of pipeline. As show in Figure 10, each column quantifies the accuracy information, that is, the accuracy of correct classification and the recognition rate of mis-classification, as well as the class blockage misjudged by a certain operating condition. Each line quantifies recall rate information, that is, how many of a given number of real pipeline samples are accurately recalled and diagnosed as blocked, and what category are misdiagnosed as unrecalled. 

It is obvious from Figure 10: class 1 and class 2 respectively represent the normal operation condition of the pipe clean and lateral connection, and the recognition accuracy of both the normal operation of the pipe on the test set reaches 100%. Therefore, the algorithm achieves 100% recognition accuracy between the normal operating conditions and blocked. Class 3, class 4 and class 5 represent a 20 mm blockage, a 40 mm blockage and a 50 mm blockage for a single blockage in the pipe, respectively. By analyzing the types of misjudgment of blockage, it can be seen that the above misjudgment is the error between single blockage categories, which belongs to the misjudgment of different degrees of blockage of single blockage, and there is no misjudgment to multiple blockage. Class 6, class 7, class 8 and class 9 represent a 40 mm blockage and a LC, a 55 mm blockage and a LC of multiple blocked pipes, a 40 mm blockage and a 55 mm blockage and a 40 mm blockage, a 55 mm blockage, and LC respectively. By analyzing the types of multiple blockage misjudgments, it can be seen that the above misjudgments are the misjudgments of multiple blockage categories with different degrees of blockage, and there is no misjudgment to a single blockage.

It can be seen that the recognition rate of comprehensive pipeline operation status can reach 99.56%. Through experimental verification, the improved VMD-IG-RMSE-RF algorithm has superior recognition ability and high diagnosis accuracy for pipeline blockage.

Comment 2.It is needed to improve the quality of Figures 1 and 3.

Response: Thank you for your careful examination of our paper. Figures 1 and 3 has been modified to complete. The results are shown in the figure below.

①We modified the flow diagram in figure 1 of the paper. This part is in 209-210 lines of the manuscript.

Figure 1.The flow diagram of the proposed method

②We modified figure 3 and selected a sample of normal pipeline operation, single blockage and multiple blockage to draw time domain diagram and frequency domain diagram. This part is in 289-291 lines of the manuscript.

Figure 3. Time domain and frequency domain diagrams of pipe conditions(normally, single blockage, multiple blockages): (a) Time-domain; (b) Frequency domain.

Comment 3.Additional descriptions should be provided on the Figure 4.

Response:Thank you for your comment. We have added a description of Figure 4 in the document,This part is in 313-322 lines of the manuscript.

As follows:

The time-frequency map reflects the information that the frequency of acoustic signal changes with time. The time-frequency map reflects the energy carried by each frequency component of the signal through the cold and warm color. The warmer the color, the greater the energy. The common time-frequency transformation methods include short-time Fourier transform, Wigner distribution and wavelet transform. Compared with the first two, wavelet transform has adaptive time-frequency resolution and faster algorithm. Therefore, the sound signal is generated into a time-frequency map by continuous wavelet transform. As shown in Figure 4, the operational condition of a typical pipe is selected for time-frequency map analysis.

Comment 4.The literature review in introduction is thorough and it is well written, however six additional references listing below regarding to vibration and signal processing could be added in the introduction.

Response:Thank you for your comment. The bibliography of other articles in this field have been added to the reference list of this paper according to the reviewer’s comments.

References in the literature:

This part is in 61-62 lines of the manuscript

①-②Some methods, such as power spectral density function(PSD) [19], short time Fourier transform (STFT) [20], empirical mode decomposition(EMD) [21] and local mean decomposition(LMD) [22] are used to decompose signals.

This part is in 65 lines of the manuscript

③VMD decomposes each mode from low frequency to high frequency and rebuilds the original signal by selecting the effective Mode [23,24].

This part is in 72 lines of the manuscript

④Hence, a detailed analysis is necessary regarding the amount of feature information contained in different frequencies of sound signals under the blockage condition [28].

This part is in 175-176 lines of the manuscript

⑤Meanwhile, information entropy represents the system complexity resulting from multiple uncertain factors [35].

References:

This part is in 630,632,641,662 lines of the manuscript

19. Bayat, M.; Ahmadi, H.R.; Mahdavi, N. Application of power spectral density function for damage diagnosis of bridge piers. Structural Engineering & Mechanics 2019, 71, 57-63.
20. Ahmadi, H.R.; Mahdavi, N.; Bayat, M. A novel damage identification method based on short time Fourier transform and a new efficient index. Structures 2021, 33, 3605-3614.
21. Fu, L.; Zhu, T.; Pan, G.; Chen, S.; Wei, Y. Power Quality Disturbance Recognition Using VMD-Based Feature Extraction and Heuristic Feature Selection. Applied Sciences 2019, 9, 4901.
22. Wang, B.; Li, Y.; Zhao, W.; Zhang, Z.; Zhang, Y.; Wang, Z. Effective Crack Damage Detection Using Multilayer Sparse Feature Representation and Incremental Extreme Learning Machine. Applied Sciences 2019, 9.
23. Santo, F.T.; Sattar, T.P.; Edwards, G. Validation of Acoustic Emission Waveform Entropy as a Damage Identification Feature. Applied Sciences 2019, 9, 4070.

Author Response File: Author Response.doc

Reviewer 2 Report

The paper is interesting with the proposed hybrid model, but from scientific point not all factors are evaluated - only the clay pipeline was used for experimental setup, but different materials (Cement Concrete, Cast Iron, Steel, Plastic, etc.) used in the construction of sewerage could have different absorption of sound wave signals. The technical state of pipeline has an impact too - Cement Concrete or Reinforced Concrete could be not only cracked, but deteriorated under the impact of bio-corrosion or sulfate attack. I think an authors of this paper will use these my remarks in the future research. Some of them I saw mentioned at the end of paper in the last sentence of conclusions.

I have noticed some typing mistakes :

lines 239, 240 : "L/s" should be replaced by "l/s"

line 290: "veloc-  ity"should be "veloci-ty"

line 574: "....the following aspects: The sound..." should be"...the following aspects: the sound.."

line 460: title "Figure 8. ERMS distributions of nine pipe operating conditions " should be at the same page as figure

Author Response

Response to Reviewer #2

Comment 1. lines 239, 240 : "L/s" should be replaced by "l/s"

Response: Thank you for your comment. Following your suggestion, all equations are rewritten using the equation editor and the reference of the equation 2, 3 has been added to the reference list of this paper. As shown below

Comment 2.line 290: "veloc- ity"should be "veloci-ty"

Response: Thank you for your careful examination of our paper. Following your suggestion, The word “veloc- ity” has been modified as follows: “veloci-ty”. As shown below

Comment 3.line 574: "....the following aspects: The sound..." should be"...the following aspects: the sound.."

Response:Thank you for your careful examination of our paper. Following your suggestion, "...the following aspects: The sound..." has been modified as follows: "...the following aspects: the sound...".As shown below

Comment 4.line 460: title "Figure 8. ERMS distributions of nine pipe operating conditions " should be at the same page as figure

Response: Thank you for your careful examination of our paper. As shown in the figure below, the figure and title of Figure 8 are already at the same page as figure.

Author Response File: Author Response.doc

Reviewer 3 Report

Good and useful work.

Recommendation for authors: For a better understanding of the usefulness of the research, it is necessary to compare the method of computer processing of information proposed in the work with the method of direct acquisition of images of blockages in pipes using a movable video camera.

Author Response

Thank you for your comment. In the next step, we will add the computer information processing method proposed in this work and compare it with the method of directly obtaining the pipeline blockage image using CCTV.

Author Response File: Author Response.doc

Round 2

Reviewer 1 Report

Thank You.