A Novel Low-Complexity Fault Diagnosis Algorithm for Energy Internet in Smart Cities

Round 1

Reviewer 1 Report

This paper presents an approach for fault diagnosis in a smart energy system. The paper is interesting, however, it needs to be improved before recommending acceptance.

The language must be improved. There are a lot of language issues, grammatical errors, and typos.

Make sure all the acronyms are defined in first use. The same for the variables, make sure all of them are properly defined in the text. There are a lot of undefined acronyms, for instance in the abstract define SFBT, UKF, do not let the users infer about its meaning. There are several instances in the manuscript.

The literature review must be expanded in order to better frame the contribution of the paper. It would be necessary to mention the different approaches for fault diagnosis, this is that there are also model-based (observer-based) methods for fault diagnosis (e.g. https://doi.org/10.1002/asjc.1913) and data-based methods for fault diagnosis such as PCA (e.g. http://doi.org/10.1016/j.ifacol.2018.09.604).

It is important to define the scope of the study. In this sense, please define which kind of faults are you considering: additive/multiplicative faults and abrupt/incipient/intermittent faults.

Define the term "light-weighted algorithm", do you mean "low complex algorithm"?. In this sense, other methods are not light-weighted? Can you make a comparison between methods in these terms? e.g. the number of basic operations to do the fault diagnosis and isolation, please include also a performance measure to see if by lowering the complexity you are nor sacrificing performance.

There are words that are misleading, such as "status", it should be "state".

Can your methods detect and diagnose simultaneous faults? what would happen if there are simultaneous faults? Include these tests in your results tables.

Please improve the figures, there are figures that need to be resized. Please use Matlab or other specialized software for plotting your results.

The conclusions are rather short, please expand them. Do not put just a summary of the paper. Include future work.

Author Response

Response to Reviewer1

 

This paper presents an approach for fault diagnosis in a smart energy system. The paper is interesting, however, it needs to be improved before recommending acceptance.

 

The language must be improved. There are a lot of language issues, grammatical errors, and typos.

 

Reply: We have checked the grammatical and typo throughout the paper.

 

Make sure all the acronyms are defined in first use. The same for the variables, make sure all of them are properly defined in the text. There are a lot of undefined acronyms, for instance in the abstract define SFBT, UKF, do not let the users infer about its meaning. There are several instances in the manuscript.

 

Reply: We have rewritten the abstract for clarity and give the description of symbols in notion table as well.

 

Abstract: The smart energy system, viewed as an "Energy Internet", consists of the intelligent integration of decentralized sustainable energy sources, efficient distribution, and optimized power consumption. That implies the fault diagnosis for a smart energy system should be of low complexity. In this paper, we propose a strong tracking unscented kalman filter and modified bayes’ classification based modified three sigma test, abbreviated as SFBT, for smart energy networks. The theoretical analysis and simulations indicate that SFBT detects faults with a high accuracy and a low complexity of O(n).

 

Symbol       Description

STUKF       Strong Tracking Unscented Kalman Filter

MB Modified  Bayes’ classification algorithm

MTS Modified Three Sigma test

 

The literature review must be expanded in order to better frame the contribution of the paper. It would be necessary to mention the different approaches for fault diagnosis, this is that there are also model-based (observer-based) methods for fault diagnosis (e.g. https://doi.org/10.1002/asjc.1913) and data-based methods for fault diagnosis such as PCA (e.g. http://doi.org/10.1016/j.ifacol.2018.09.604).

 

Reply: We have added two important references and given the corresponding descriptions.

 

Recently, Samuel et al.[42] presents the design of a H∞sliding mode and an unknown input observer for Takagi-Sugeno systems to deal with the problem of inexact measurements of the premise variables. In [43], a data-driven system based on PCA is designed to detect and quantify fluid leaks in an experimental pipeline.

 

It is important to define the scope of the study. In this sense, please define which kind of faults are you considering: additive/multiplicative faults and abrupt/incipient/intermittent faults.

 

Reply: Thank you for point it out. The proposed strategy SFBT is designed to detect both of attenuation type faults and jump type faults. And we also give the simulation about the performance of SFBT detecting jump type fault of the resistor and the attenuation type fault of the capacitor. To clarify the type of faults the SFBT can detect, we have made the modification in the contribution on line 54-58 as follows.

 

In this paper, we propose a novel low-complexity algorithm named SFBT for fault detection in smart cities, which consists of a Strong Tracking Unscented Kalman Filter (STUKF), a Modified Bayes’ classification algorithm (MB) and a Modified Three Sigma test (MTS), with an aim to detect both of attenuation type faults and jump type faults. Besides, as a special case of the hypothesis test, SFBT can detect Byzantine faults[11].

 

Define the term "light-weighted algorithm", do you mean "low complex algorithm"?. In this sense, other methods are not light-weighted? Can you make a comparison between methods in these terms? e.g. the number of basic operations to do the fault diagnosis and isolation, please include also a performance measure to see if by lowering the complexity you are nor sacrificing performance.

 

Reply: Thank you for point it out. Actually a light-weighted algorithm is relative and it is difficult to give a definition, therefore we have changed the tile of our paper into “A Novel Low-complexity Fault Diagnosis Algorithm”. Besides, we have proved that the complexity of SFBT is O(n). Be specific, SFBT requires sampling such that the corresponding message complexity depends on the number of sample, which equals to the data window N1. And the data window can be chosen as a constant such that the number of messages exchanged is less than |MDS|N1. If a small data window is chosen, then detection accuracy is reduced for sure. We take Fig. 1 and Fig. 3 for example. If we choose a small data window, then it will take more time for SFBT to “fit” the real data curve. That suggest we have k>1250 and step>1110.

 

There are words that are misleading, such as "status", it should be "state".

 

Reply: We have changed the status to state throughout the paper.

 

Can your methods detect and diagnose simultaneous faults? what would happen if there are simultaneous faults? Include these tests in your results tables.

 

Reply: Thank you for point it out. The proposed strategy SFBT can detect each single fault. Besides, the Strong Tracking Unscented Kalman Filter (STUKF) and the Modified Bayes’ classification algorithm (MB) of the SFBT is designed to track the change of parameters for fault detection, no matter in a single fault scenario or in the simultaneous faults scenario.

 

Please improve the figures, there are figures that need to be resized. Please use Matlab or other specialized software for plotting your results.

 

Reply: We have resized all figures.

 

The conclusions are rather short, please expand them. Do not put just a summary of the paper. Include future work.

 

Reply: We have expanded the conclusions and given our future work as well.

 

The optimized power consumption for a smart energy system of a smart city is crucial. Although a variety of fault diagnosis algorithms have been developed, how to efficiently detect faults with a low complexity poses a great challenge. To this end, we propose the SFBT that works on a clustered network. It first applies a strong tracking UKF (STUKF) and a modified Bayes’ classification algorithm  (MB) to detect fault-free cluster-heads in a centralized manner. Then, a decentralized modified three Sigma test (MTS) is developed to identify faulty cluster-members to overcome the "masking" problem within the cluster with a faulty cluster-head. Besides, a faulty-free cluster-head can detect any faulty cluster-member by simply comparing the data of its own and the one of the cluster-member. The theoretical analysis and experiment results indicate that the SFBT achieves higher diagnosis accuracy over some contemporary strategies with a complexity of O(n).

 

In fact, each faulty free cluster-head detected by the SFBT is only one hop away from corresponding cluster-members. That suggests data collected by the cluster-head don’t differ much from that of cluster-members which makes three Sigma test work. However, by doing so more cluster-heads are required. Thus, our future work includes how to detect faulty cluster-members at least two hops away with a complexity no more than O(n).

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Optimization of power consumption for a smart energy system within a smart city is an important economic problem to preserve energy. It implies fault diagnosis approaches to efficient detect fault with a low complexity. Some approaches have been developed and have been described in the literature. In this paper, a novel light-weight diagnosis algorithm SFBT is proposed to detect faults within a clustered smart energy network of a smart city. It applies a strong tracking UKF and a modified Bayes’ classification algorithm to detect fault-free cluster-heads in a centralized manner. It is followed by fault statuses detection of 9 corresponding cluster-members with a modified three sigma test. A theoretical analysis and simulations are done which indicate that the SFBT assures higher diagnosis accuracy with a linear complexity.

The sentences i.e. in the Abstract are very long and so it is difficult to follow the content. The Conclusions are too short and sentences about future research work of the authors in this context are missing.Conclusions have to be extended and future work of the authors included.

Author Response

 

Response to Reviewer2

 

 

 

The sentences i.e. in the Abstract are very long and so it is difficult to follow the content. The Conclusions are too short and sentences about future research work of the authors in this context are missing. Conclusions have to be extended and future work of the authors included.

 

Reply: We have rewritten the abstract for clarity.

 

Abstract: The smart energy system, viewed as an "Energy Internet", consists of the intelligent integration of decentralized sustainable energy sources, efficient distribution, and optimized power consumption. That implies the fault diagnosis for a smart energy system should be of low complexity. In this paper, we propose a strong tracking unscented kalman filter and modified bayes’ classification based modified three sigma test, abbreviated as SFBT, for smart energy networks. The theoretical analysis and simulations indicate that SFBT detects faults with a high accuracy and a low complexity of O(n).

 

We have expanded the conclusions and given our future work as well.

 

The optimized power consumption for a smart energy system of a smart city is crucial. Although a variety of fault diagnosis algorithms have been developed, how to efficiently detect faults with a low complexity poses a great challenge. To this end, we propose the SFBT that works on a clustered network. It first applies a strong tracking UKF (STUKF) and a modified Bayes’ classification algorithm  (MB) to detect fault-free cluster-heads in a centralized manner. Then, a decentralized modified three Sigma test (MTS) is developed to identify faulty cluster-members to overcome the "masking" problem within the cluster with a faulty cluster-head. Besides, a faulty-free cluster-head can detect any faulty cluster-member by simply comparing the data of its own and the one of the cluster-member. The theoretical analysis and experiment results indicate that the SFBT achieves higher diagnosis accuracy over some contemporary strategies with a complexity of O(n).

 

In fact, each faulty free cluster-head detected by the SFBT is only one hop away from corresponding cluster-members. That suggests data collected by the cluster-head don’t differ much from that of cluster-members which makes three Sigma test work. However, by doing so morecluster-heads are required. Thus, our future work includes how to detect faulty cluster-members at least two hops away with a complexity no more than O(n).

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

All my concerns were addressed. However, the language still has a lot of issues that need to be corrected.