Real-World Applications - Immunological Computation: Theory and Applications - page 186

used learning vector quantization (LVQ) to determine a correlation between two
sensors from their outputs when they work properly. Each sensor is equated to a B
cell in an immune network, and sensors test one another's outputs to see whether or
not they are normal using the extracted correlations. Here, reliability of the sensor
is used in lieu of the similarity to neighbors.
In the fi eld of diagnosis, there has also been some interest in creating distributed
diagnostic systems. Kayama et al. (1995) initially proposed a parallel-distributed
diagnostic algorithm. h e authors compared their algorithm to that of an immune
network due to its distributed operation, and the systems emergent cooperative
behavior between sensors. h is work was then continued by Ishida (1990, 1996).
Active diagnosis continually monitors for consistency between the current states
of the system with respect to the normal state. Each sensor can be equated with
a B cell, connected through the immune network with each sensor maintaining
a time-variant record of sensory reliability, thus creating a dynamic system. h is
work diff ers from the aforementioned in the way in which the reliability of each
sensor is calculated.
An AIS technique was applied to refrigerated cabinets in supermarkets to
detect the early symptoms of icing up. Taylor and Corne (2003) used in-cabinet
temperature data to predict faults from the pattern of temperature over time.
h is technique used r -bits matching rule in conjugation with a specialized diff eren-
tial encoding of data to spot fault patterns in a time-series temperature data from
supermarket freezer cabinets.
An aircraft fault-detection system, called multilevel immune learning detection
(MILD), was developed (Dasgupta et al., 2004) to detect a broad spectrum of known
as well as unforeseen faults. Empirical study was conducted with datasets collected
through simulated failure conditions using National Aeronautics and Space Admin-
istration (NASA) Ames C-17 fl ight simulator. h ree sets of in-fl ight sensory informa-
tion—namely, body-axes roll rate, pitch rate, and yaw rate were considered to detect
fi ve diff erent simulated faults: one for engine, two for the tails, and two for the wings.
h e MILD implemented a real-valued negative selection (RNS) algorithm, where a
small number of specialized detectors (as signatures of known failure conditions) and
a set of generalized detectors (for unknown or possible faults) are generated. Once
the fault is detected and identifi ed, an adaptive control system would use this detec-
tion information to stabilize the aircraft by utilizing available resources (control sur-
faces). Experiments were performed with datasets collected under normal and various
simulated failure conditions using a piloted motion-based NASA simulation facility.
A snapshot of a running MILD is shown in Figure 7.10.
An artifi cial immune regulation (AIR) scheme was proposed and integrated
into an immune model-based fault detection approach for fault diagnosis (Luh
et al., 2004). h is system generated residuals that contained information about
the faults. However, various disturbances and errors caused residuals to become
nonzero, thus interfering with detection of faults. h e AIR scheme produced a set
of memory B cells whose amount depended on several chemical rate constants.