A Strategy for Reliability Evaluation and

Fault Diagnosis of Autonomous Underwater

Gliding Robot based on its Fault Tree

Koorosh Aslansefat

Gholamreza Latif-Shabgahi and Mojtaba Kamarlouei

Email: k_Aslansefat@sbu.ac.ir

Third International Conference on Robotics, Automation and Communication Engineering (ICRACE 2014)

Dubai, UAE, October 24, 2014

Topics

Autonomous Underwater Gliders (AUGs)

Fault Tolerance and its Necessity, Attributes and Solutions

Static Fault Tree, Assumption and Modelling

Faults in AUG’s Components

A Strategy to Construct AUG's Fault Tree

Reliability Evaluation of AUGs

Conclusion

2

Autonomous Underwater Gliders (AUGs) 3

An Autonomous Underwater Glider (AUG) is a special type of

Autonomous Underwater Vehicle (AUV) that uses small changes

in its buoyancy in conjunction with wings to convert vertical

motion to horizontal, and thereby propel itself forward with very

low power consumption.

It can be use for long range survey (thousands of kilometers of range)

Extending ocean sampling missions from hours to weeks or months

Providing data on temporal and spatial scales

Much more costly to sample using traditional shipboard techniques

(typically cost $100,000)

The benefits of using AUGs

Fault Tolerance and Necessity

4

Applications

Safety Critical

Environment Critical

Budget Critical

Fault tolerance is the property that enables a system to

continue operating properly in the occurrence of the failure of

its components

Fault Tolerance and Attributes

Reliability

Availability

Safety

Performability and etc.

5

Generally, in fault tolerance such parameter need to measured:

Failure Rates

Repair Rates

MTTF, MTBF and etc.

Fault tolerant system evaluated by attributes such as:

Fault Tolerance and Modelling

6

Several methods have been developed for Reliability Evaluation

State Space Methods

(Semi) Markov Model

Petri-Nets

Combinatorial Methods

Reliability Block Diagram (RBD)

Fault Tree

Numerical and Simulation-Based Methods

Monte Carlo

Static Fault Tree

A Fault Tree (FT) illustrates the ways

through which a system fails. It states

different ways in which combination of

faulty components (called the "Basic

Events") result in an undesired event in the

system (called the "Top Event"). In this

model, basic events are connected to each

other through logical gates forming upper-

level intermediate events

7

(a)(b)(c)

(d)(e)

Fault Tree Assumption

The occurrence of more than one fault at the same time is not

allowed and the common cause failures (CCFs) are ignored.

On-time repairing of vehicle's components is not allowed.

Dynamic characteristics such as functional dependency, components'

priority and the use of spares are not applied in the model.

The system components failure rate obeys exponential distribution

function

8

Fault Tree Solutions

9

1

AND

i

iQ Q

1

1 1OR i

i

Q Q

From probability theory, the output probability of AND & OR gates is

calculated as follows:

Faults in AUG’s Components

In this paper, 9 subsystems and their faults have been considered.

Power system

Leak detection system

Diving system

Environment detection

Collision avoidance

10

Computer system

Propulsion system

Communication system

Navigation system

Navigation System Faults

In AUGs, the GPS antenna is usually located on

one of the horizontal wings. Whenever the

vehicle comes to the surface, through its 90

degree rolling movement the antenna locates in

the highest place from water surface in order to

receive the data. In some other types of these

vehicles there is a flap behind the main rudder

where the antenna is located. If this system or

the rolling system confronts any problem,

vehicle's location recognition becomes hard and

the probability of vehicle’s loss is increased. It

is obvious that any malfunction in GPS and its

processor cause the vehicle to get lost.

11

A Strategy to Construct AUG's Fault Tree

The tree is constructed in five steps as follows:

1) Determine the level of faults occurrence (sensor, sub-subsystem, and subsystem).

2) Determine the contribution of each component faults in its upper level faults, and

then use appropriate gate (AND and OR gate) to construct subsystem or main

system’s fault tree.

3) Consider each subsystem fault tree as a module of main fault tree.

4) Find available probability of each component's failure.

5) Construct the main fault tree and evaluate system’s reliability

12

Example: FT construction of navigation subsystems

According to the first step, the level of navigation

subsystem's component is determined. In this

subsystem, four basic events exist each one of

which enables navigation subsystem and this is

the way to use OR gate

13

252222 25

25

22

1 1 1 1 ... 1 1 1 ... 1Nav i

tt

i

P P P P e e

Reliability Evaluation of AUGs

15

Reliability Evaluation of AUGs

16

Cut-Sequence in Fault Tree

17

CsIP(MCs)AUG Sub-Systems

4.452920.5637863Power Sys.

6.018510.7620066Leak Detection Sys.

6.957080.8808393Diving Sys.

6.805150.8616035Environment Detect Sys.

6.832430.8650568Obstacle Avoidance Sys.

6.956660.8807865Computer Sys.

7.616520.9643316Propulsion Sys.

7.213830.9133464Communication Sys.

4.913860.6221463Navigation Sys.

i

i

P MCsMCsI

P TOP

Conclusion

The literature suffer from AUGs fault tree construction and this paper

present a typical fault tree for AUGs

Reliability of two type autonomous underwater robot (AUVs and AUGs)

have been compared and shows the reliability of AUGs is more of than

AUVs

By means of cut-sequence analysis, failure bottleneck of AUGs such as …

and … have been shown.

18

Future Works

The presented fault tree can be developed to use for fault diagnosis of AUGs

This research can be extended for considering a dynamic behaviors of faults

such as priority and sequence dependency, spare, functional dependency and

repair.

The other dependability attributes such as availability and safety can be

evaluated.

It possible to use non-exponential failure distribution.

19

References

20

[1] A. Ortiz, P. Julian, B. Guillem and O. Gabriel, "Imroving the safety of AUVs," in OCEANS '99 MTS/IEEE, Seattle, WA, 1999.

[3] J. Strutt, "Report of the Inquiry Into the Loss of Autosub2 Under the Fimbulisen," National Oceanography Centre, Southampton, 2006.

[4] M. P. Brito, G. Griffiths and A. Trembranis, "Eliciting Expert Judgment on the Probability of Loss of an AUV Operating in Four Environments,"

National Oceanography Centrer, Southampton, 2008.

[5] G. Griffiths and A. Trembanis, "Towards a Risk Management Process for Autonomous Underwater Vehicle," in Masterclass in AUV Technology for

Polar Scienc, Southampton, Society for Underwater Technology, 2007, pp. 103-118.

[6] X. Bian, C. Mou, Z. Yan and J. Xu, "Simulation Model and Fault Tree Analysis for AUV," in International Conference on Mechatronics,

Changchun, 2009.

[7] M. P. Brito, D. Smeed and G. Griffiths, "Underwater Glider Reliability and Implications for Survey Design," Oceanic Technology, 2013.

[8] M. P. Brito, D. Smeed and G. Griffiths, "Analysis of the Operations of 58 Gliders During the Last 2 Years," National Oceanography Center and

Southampton University, Liverpool, UK, 2013.

[9] G. Griffiths, C. Jones, J. Ferguson and N. Bose, "Undersea gliders," Journal of Ocean Technology, vol. 2, no. 2, pp. 64-75, 2007.

[10] H. Xu, G. Li and J. Liu, "Reliability Analysis of an Autonomous Underwater Vehicle Using Fault Tree," in IEEE International Conference on

Information and Automation (ICIA), Yinchuan, 2013.

[11] G. Griffiths, L. Merckelbach and D. Smeed, "On The Performance of Three Deep-diving Underwater Gliders," in OCEANS 2007 - Europe,

Aberdeen, 2007.

[12] K. Aslansefat, G. Latif-Shabgahi and M. Kamarloie, "Faults Taxonomy in Autonomous Underwater Vehicle and Provide their Fault Tree," in 15th

Marine Industries Conference, Kish, Iran, 2013.

Thanks