Artificial Immune Systems: An Overview for Faulting Actuators
Round 1
Reviewer 1 Report
This paper reviews Artificial Immune Systems (AIS) that can be implemented to compensate for 8 actuators that are in a faulted state or operating abnormally.
I am concerned with the fact that how the distributed and cloud-based computing support the development fault diagnosis and fault tolerant control algorithms?
The main concern of the reviewer is on the novelty of using Artificial Immune Systems (AIS) to solve the fault tolerant control problem. In fact, many effective fault tolerant methods have been developed in recent years, such as the adaptive control methods, robust methods, and data-driven methods. What is the advantage of the AIS compared with the existing methods?
In Section 2.3, the introduction of the AIS is quite theoretical, can you provide an illustrative diagram to show how the AIS is used to solve fault tolerant control problems? This is benefit to highlight the novelty of the paper.
In addition, as a review paper, some related references are missing. The fault diagnosis and fault tolerant control results with or without including the Artificial Immune
doi.org/10.1080/23307706.2014.960554;
DOI:10.1109/TCYB.2013.2286209;
DOI:10.1109/TNNLS.2015.2507183
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper deals with the usability of the Artificial Immune System (AIS), which are one of the genetic algorithm variants, for fault tolerant control. This work presents research for the fundamental operating principles of AIS, their applications, and a brief case-study on their applicability to fault diagnosis by considering an over-actuated rover with four independent drive wheels and independent front and rear steering. The binary genetic algorithm is described and then illustrated using an example of a piping system with ten valves that control the output flow of a tank, and where a binary genetic algorithm could be used to determine the optimal valve combination. For application cases that cannot consider binary inputs, the real valuated genetic algorithm is proposed and illustrated using the same example. This application support allows the authors to highlight the limitation of these algorithms. Then the AIS is introduced. To demonstrate how an AIS can be implemented for fault tolerant control, an example was built in MATLAB. For simplicity in the setup, the Optimization Toolbox was used to test many of the functions required. Indeed, with improvements to the computational speeds available as well as improvements in the speed of the algorithms, the AIS could become a feasible alternative for real-time fault tolerant control.
The issue of the FTC is old and the proposed solutions are diverse. This paper offers an original alternative, and presents in an unbiased way the benefits and limitations of using genetic algorithms for the FTC. This paper will be of interest to readers, and will encourage researchers to explore these tools from operational research in greater depth.
However some points can be improved:
1) It would be nice to briefly discuss the other approaches in the introduction, citing for example papers: A survey of fault diagnosis and fault-tolerant techniques; part i: fault diagnosis with model-based and signal-based approaches, IEEE Trans. Ind. Electron. 62 (6) (2015) 3757e3767. Or works based on the physical model like for example exemple : Fault diagnosis and fault tolerent control of an electric Vehicle over-actuated", IEEE Transaction on Vehicular Technology; Vol. 62 (3), pp. 986 - 994, fev 2013.
2) One of the keys of the FTC is the anticipation of defects, can the failure prognosis be associated with algorithms derived from the OR as genetic algorithms to have reliable FTC algorithms.
3) In FTC, there are physical constraints in terms of actuating systems, can genetic algorithms be used in just-actuated or under-actuated system cases.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The paper can be accepted now
