|
|
Optical Fibre Alignment Using Fuzzy Techniques
|
|
  |
|
Problem Statement
The problem is to reduce the long assembly time of fibre optic cables caused by the labour-intensive task of coupling the laser with the core of the fibre. |
Improvement Approach
Fuzzy logic and evolutionary adaptation techniques were employed to enable the computer to learn how to perform the alignment then to reproduce the most effective procedures. |
Benefits
The automatic alignment system achieved a 100% success rate in tests and the time for linking components was decreased 10-fold. |
|
|
Contact information
Centre for Advanced Manufacturing Systems
PO Box 925, Newport Road, Cardiff, CF24 0YF, UK
Phone +44 (0)29 20874641 ,
Fax
+44(0)29 20874880
e-mail
manufacturing@cf.ac.uk
http://www.mec.cf.ac.uk
|
|
|
Fibre-optic cable. |
|
Fibre optics has become the optimum technology for long distance telecommunications. Electrical signals are converted to a modulated laser light beam and transmitted down an optical fibre tuned to minimise the attenuation of the signal travelling down its length. A receiver at the other end of the fibre reverses the process to recover the signal. A growing market has greatly increased demand for fibre optic cables and associated electro-optic components such as transmitters and receivers.
During component assembly, Agilent Technologies found that efficient coupling of the laser beam and the 9 mm diameter core of the fibre proved to be difficult, labour intensive and time-consuming. Depending on the product being manufactured, manual or semi-automatic techniques can take up to 5 minutes to achieve an efficient joint.
|
Conventional computer automation cannot match the speed of the best human operator, though it can be more consistent, and part to part variation causes additional difficulties.
Agilent Technologies required an automated solution to the assembly problem which would decrease production time while giving consistent and accurate results, enabling higher throughput and increased productivity.
|
The solution proposed used a mixture of fuzzy logic and evolutionary adaptation techniques developed at the MEC to teach the computer to align the components correctly.
This is because human operators show a combination of intuitive searching and experience of working with similar devices to join components in a relatively small number of moves. For a computer to combine the best features of both human and computer approaches, it has to be configured to ‘learn’ like a human operator and apply its ‘experience’ to the task at hand.
This has been achieved by MEC researchers in collaboration with Agilent using fuzzy logic to reduce the number of moves required to align the fibre and laser beam. Evolutionary adaptation techniques are tehn employed to reproduce the procedures which are found to be the most effective.
|
Additional work is being undertaken to further improve the fuzzy engine and incorporate the fuzzy controller into the architecture of a neural network. |
In initial tests joining pairs of fibre optic components, the automatic alignment system achieved 100% success. The time for linking components was reduced from around 5 minutes to less than 30 seconds on average. This new approach has been transferred directly into Agilent Technologies’ factory operations to improve their quality and efficiency.
Agilent technologies subsequently implemented the fuzzy logic techniques in the manufacture of some of their laser transmitters. This has resulted in significant reduction in assembly times, providing benefits in improved utilisation of the high-cost equipment and higher throughput.
|
|
|
