Video-based Cargo Fire Verification System with Fuzzy Inference Engine for
Commercial Aircraft

 

Dr. Mokhtar Sadok

Goodrich Corporation

(Fuel and Utility Systems)

Vergennes, Vermont

 

Date: Monday September 20, 2004

Time: 12:20 p.m. - 1:10 p.m.

Location: 367 Votey

 

 

Abstract

 

Conventional smoke detection systems currently installed onboard aircraft often suffer from high rates of false alarms. Under current procedures, whenever the system issues an alarm the pilot is required to release fire extinguishers and divert to the nearest airport. Aircraft diversions are costly and dangerous in some situations. Recent data collected by the FAA and NTSB show that only 10% or less of alarms are genuine. A reliable detection system that minimizes false-alarms and allows the crew to inspect the state of cargo compartments prior and after fire suppression is needed.

Goodrich Corporation recently developed a video-based verification system to address the problem. The Cargo Fire Verification System (CFVS) is a multi
camera system providing live stream digital video to the cockpit and performing hotspot, fire, and smoke detection in aircraft cargo bays.   CFVS uses video information and other sensor information to discriminate between genuine events (e.g. fire, smoke, or hotspot) and nuisance alarms (e.g. fog, dust). A Mamdani-type fuzzy inference engine built around a set of if-then rules is used to provide approximate reasoning for decision making. Gaussian membership functions for frame intensity, relative humidity, and temperature are constructed using experimental data to form the system inference engine.

More than 150 scenarios of fire tests that meet industry standards under various bay configurations and illumination conditions have been collected to form a digital library for algorithm development. Among various technical challenges that the system has to deal with is the limited outlook in aircraft cargo bays. Typically, bays are fully loaded with cargo, leaving only about 2-inch gap between cargo containers and aircraft ceiling.  Other challenges include the apparent behavior of smoke, fog, and dust.

CFVS performed much better than conventional aircraft smoke detectors in all standardized tests. Currently, CFVS is designed to function as a secondary verification system, either confirming or unconfirming smoke alarms issued by the primary smoke detection system. After successful in-service evaluation and FAA certification, the CFVS will possibly replace conventional smoke detection systems in commercial aircraft.

 

Speaker bio:

Dr. Sadok is a senior research scientist at Goodrich Corporation (Fuel and Utility Systems) of Vergennes, Vermont since 1998. He obtained his Ph.D. in Electrical and Computer Engineering from Tennessee Technological University. His area of interest includes image processing, fuzzy logic, and time-frequency analysis.