Subcontracting_saristu orig v2 FINAL

Call for Proposal
TITLE
Multilevel Robust Optimisation Platform architecture for sensor position optimisation in a real scale
composite damaged Lower Wing Box panel
START AND END DATE
September 30, 2012 - October 30, 2013
TOPIC DESCRIPTION
Architecture and detail design with sensor position optimisation: ICL will subcontract the architecture for the
Multilevel Robust Optimisation platform to process the Lower Wing Box panel. The integration platform has
to address the panel failure probability for safety and economical design. The subcontractor should have
experience in design, composite structural components and damage tolerance including methodologies for
life cycle control.
Scope of work
To develop the platform architecture for a Multilevel Robust Optimisation algorithm to assess the sensor
positioning for the Lower Wing Box panel. The integration platform has to address the panel failure
probability for safety and economical design.
Background
Due to superior properties of composite materials, they have been vastly adopted in aerospace structures.
Despite their high ratios of strength to mass and stiffness to mass compared to metal structures, they are
more vulnerable to defects occurring during production or damage induced during service.
The presence of damage or the extent of the defects is difficult to predict and when not detected on time can
lead to catastrophic failures. Structural Health Monitoring (SHM) is increasingly being valued as an
alternative method to the conventional Non Destructive Techniques (NDT) for composite structures. The
motivations behind the SHM concept are reduction in the maintenance costs and increase in aircraft safety.
Based on current regulations, undetectable damage may not reduce the residual strength of a structure below
ultimate load capacity. To define undetectable damage, an inspection technique has to be employed in the
early design stage which at present is done by visual inspection. Moreover, detecting the presence of
embedded damage due to impact is uncertain and cannot be detected by visual inspection which may result
in an overly conservative design. Distributed sensor networks are emerging as a critical technical driver in
the application of structural health monitoring for large-scale structures as a result of their excellent abilities
to enhance the reliability and robustness of monitoring systems. Piezoelectric (PZT) transducers are one
of the most attractive types of sensors due to their electro-mechanical properties which allow them
to be utilised as both sensors and actuators. One of the key technical opportunities in the implementation
of a distributed sensor network is the application of information fusion. Not only does this enable the
integration of data from all sensors for the comprehensive assessment of structural conditions, but it also
facilitates the combination of decisions or perceptions from multiple sources or different approaches.
Furthermore, one of the key requirements of a sensor based SHM system is to keep the interference of the
sensor system with the structure as minimum as possible. Therefore an optimisation study must be carried in
order to optimise the number and location of sensors, while keeping the reliability and robustness of the
system at maximum.
The probability of detection (POD) is a reasonable measure to evaluate the accuracy and reliability of an
SHM system (sensor type, number and location) that can be strongly influenced by the choice of structure
configuration, material, number of sensors and their locations.
Due to the level of complexity involved in optimizing the sensor positioning to address the Lower Wing Box
failure probability, this process can be undertaken most efficiently using a multi-level approach.
Activity Description
The main objective of the Multilevel Robust Optimisation Platform is to detect damages beyond the critical
threshold obtained by visual inspection with high reliability.In an aircraft sub-component, the expert designer
can – by using his experience – locate some areas mainly subject to different kind or level of damage due to
different causes. This information can be useful to organize the sensor positioning into a multilevel logic.
Information about the health status of the structure can be obtained from an SHM platform based on
distributed sensor system. The developed SHM platform (in IS12) is capable of detecting impact event
(location and energy) and identifying the presence and characteristic of the consequent damage from sensor
readings. To obtain a reliable and robust SHM system, an optimisation algorithm must be developed and
implemented in the platform to interact with passive and active sensing modules. The optimisation algorithm
must result in the best sensor positioning for impact and damage detection with the minimum number of
sensors and for a required POD.
The optimisation algorthm and the search method have to be based on a two levels of damage classification
for damage severity.
The level-one decision fusion is first implemented by individual active sensors to create their own
perceptions on structural health status. During the fusion process, the active sensors has to interrogate local
physical sensor nodes in the network, and then combine the perceptions of the local sensors in terms of
correlation between features extracted from raw signals and damage scenarios in a knowledge database. The
perceptions of active sensors on structural health status have to be integrated to represent the decision fusion
at level two. As a consequence, the risks of a malfunction of individual sensors or the inappropriateness of
individual assessment procedures will be significantly reduced, and a robust and error-tolerant structural
health monitoring system is developed.
The approach has to be applied to a lower wing panel for damage detection in CF composite structures.
The first level damage is for instance, 100 mm2 delaminations (up to the technological treshold of damage)
where data fusion in a bay is applied and the second scan session can seek for the 500 mm2 delaminations
(up to the inspection service damage treshold) using data fusion from one section of the lower wing panel See Fig.2. Adopting this multilevel strategy optimised sensor positioning for the full lower wing box panel
can be achieved.
The final task in multilevel feature selection for fault classification and prediction is to optimize the
classification accuracy (subcomponent, areas based) that is, to distinguish as unequivocally as possible a
particular fault condition from others and from the healthy state of the lower wing box panel at hand while
providing maximum prediction accuracy. Specific data fusion methods will be able to include accuracy,
reliability sensitivity, robustness. The algorithm has to be written in Matlab.
Fig. 2: Lower Wing Box panel
SECTION
BAY
Fig. 1 Detection Levels
MAJOR DELIVERABLES AND SCHEDULE
Deliverable
D1
D2
Description
Report on Multilevel platform architecture developed to allow
for an effective positioning of sensors on Lower Wing Box
panel to allow maximum sensitivity of damage location and
classification.
Report and software module / code for damage location and
classification.
Due date
January.30.2013
October.30.2013