A Current Spectrum-Based Algorithm for Fault Detection of Electrical Machines Using Low-Power Data Acquisition Devices

Round 1

Reviewer 1 Report

1. The paper presents a mehtod for solving the low sampling frequency problem. However, the method seem unable to be applied when the sampling frequency is lower than the Nyquist rate.

2. Eq. (1) is not right.

3. Averaging operation is adopted to solving the discontinuities of the measured data. But the mehtod would fail when several continus data points are missed in the measurements.

4. Overall, the novelty or application highlights are not clear.

Author Response

Dear Reviewer,

The authors would like to thank you for your valuable comments and suggestions, which helped us improve the manuscript and give us very good ideas related to future research work in this area. The following are the answers to your questions and the changes made in the manuscript.

Reviewer 1:

1. The paper presents a method for solving the low sampling frequency problem. However, the method seems unable to be applied when the sampling frequency is lower than the Nyquist rate.

Yes, you are right. The minimum sampling rate should respect Nyquist criterion. The most significant fault representing frequencies appear near the fundamental component (50Hz). So minimum sampling rate for small power data acquisition devices is not a problem. The problem is the detection of tiny faulty representing frequencies near the fundamental component. If the sampling frequency is very small, then those fault representing frequencies hide themselves under their parent frequency components due to poor resolution and leakage. Hence the proposed algorithm discusses all those issues using less complicated approaches.

The information is included in the introduction portion now.

2. Eq. (1) is not right.

The basic purpose of Eq. (1) is to describe that the acquired signal would have some complete number of cycles and some incomplete number of cycles at the starting and ending of the signal. So, it segregates the input signal into the complete number of cycles and the remaining fractional parts. Moreover it is for near sinusoid signals which is true for induction motors.

In this equation J is the total number of cycles, fin is the frequency of the fundamental component, fs is the sampling frequency, M is the recorded signal’s length, Jint are the integral number of signal cycles and ∆ is the fractional part. The non-zero ∆ leads to the spectral leakage. The information is included now.

3. Averaging operation is adopted to solving the discontinuities of the measured data. But the method would fail when several continuous data points are missed in the measurements.

Thank you for this comment which is a very valid concern. If there are more than one consecutive missing data samples then there are some possible ways of correction. Replace the missing samples with the samples from same lo-cation of the subsequent cycle. The other way is that the samples will be replaced by random values depending upon the amplitude of the available samples at the start and end of the missing segment and the amplitude will be iteratively corrected. The third way is this that if the cycles is worsley effected then it can be totally replaced with the healthy one from the signal. This paper at the moment deals with only one discontinuity between two healthy samples.

The relevant information is included now.

4. Overall, the novelty or application highlights are not clear.

Some modifications are made now.

Reviewer 2 Report

check attach file

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

The authors would like to thank you for your valuable comments and suggestions, which helped us improve the manuscript and give us very good ideas related to future research work in this area. The following are the answers to your questions and the changes made in the manuscript.

Reviewer 2:

• "detecting the integral number of cycles" could be rephrased as "detecting the number of complete cycles" · "deleting the fractional parts of the signal" could be rephrased as "removing any fractional components of the signal" · "signal interpolation for fault diagnostics" could be rephrased as "interpolating the signal for fault diagnosis" · "are increasing due to the increasing trend" could be rephrased as "are becoming more common due to the growing trend" · "spectral leakage of powerful frequency components" could be rephrased as "leakage of powerful frequency components into other parts of the spectrum" · "that are not possible in small signal processing devices" could be rephrased as "that cannot be performed on small signal processing devices" · "an algorithm is proposed which can improve the spectrum resolution" could be rephrased as "this paper proposes an algorithm that can enhance spectrum resolution
• Introduction: The paper proposes a novel algorithm for improving the resolution of the frequency spectrum for fault diagnostics of electrical machines using low-power data acquisition cards. The use of smart sensor-based low-power data acquisition devices such as Arduino cards is increasing due to the trend of IoT, cloud computation, and industry 4.0 standards. However, detecting the fault representing frequencies at an incipient stage is challenging due to their small amplitude and spectral leakage of powerful frequency components. Therefore, the paper proposes an algorithm that improves the spectrum resolution without requiring complex advanced signal processing techniques.
• Methodology: The proposed algorithm detects the integral number of cycles, deletes the fractional parts of the signal, and performs signal interpolation and discontinuity analysis. The algorithm's implementation involves three main stages: signal preprocessing, Fourier analysis, and post-processing. The signal pre-processing stage includes signal filtering and sampling. The Fourier analysis stage involves applying the fast Fourier transform (FFT) algorithm to the filtered and sampled signals. The post-processing stage includes deleting the fractional parts of the spectrum, detecting the integral number of cycles, and performing signal interpolation and discontinuity analysis.
• Results: The paper presents results obtained from both simulation and practical experiments. In the simulation, the proposed algorithm is compared with other state-of-the-art algorithms in terms of spectrum resolution improvement, signal to-noise ratio (SNR), and total harmonic distortion (THD). The results show that the proposed algorithm outperforms other algorithms in terms of spectrum resolution and SNR. In practical experiments, the proposed algorithm is tested on a low-power data acquisition card and compared with a high-performance data acquisition card. The results show that the proposed algorithm achieves similar performance as the high-performance card but with much lower power consumption.
• Discussion: The proposed algorithm offers several advantages for fault diagnostics of electrical machines. First, it improves the spectrum resolution without requiring complex advanced signal processing techniques. Second, it is suitable for low power signal processing devices such as Arduino cards. Third, it achieves similar performance as high-performance data acquisition cards but with much lower power consumption. However, the paper does not discuss the algorithm's limitations, such as its applicability to different types of electrical machines and the accuracy of its results compared to more advanced signal processing techniques.
• Conclusion: The proposed algorithm offers a novel solution for improving the spectrum resolution for fault diagnostics of electrical machines using low-power data acquisition cards. The results obtained from simulation and practical experiments show that the proposed algorithm outperforms other state-of-the-art algorithms in terms of spectrum resolution and SNR while achieving similar performance as high-performance data acquisition cards but with much lower power consumption. However, further research is needed to explore the algorithm's limitations and accuracy compared to more advanced signal processing techniques. Modify introduction with the help of this references. https://doi.org/10.3390/machines10020155;https://doi.org/10.3390/pr10040656doi: 10.1109/TIM.2020.3025396 doi: 10.1109/TCSII.2023.3234609; doi: 10.1109/TPEL.2022.3216513

 

Thank you for your comments. The suggested corrections are made now.

Round 2

Reviewer 1 Report

The information of supplementary materials, author contributions, data availability statement, acknowledgements, as well as the conflicts of interest should be revised.

Author Response

Thanks for review. The suggested revision is performed now.

Reviewer 2 Report

Accept

Author Response

Thanks for review.