The implementation of an automated decision support system in the field of structural design and optimization can give a significant advantage to any industry working on mechanical design. Such a system can reduce the project cycle time or allow more time to produce a better design by providing solution ideas to a designer or by upgrading existing design solutions while the designer is not at work. This paper presents an approach to automating the process of designing a gas turbine engine rotor disc using case-based reasoning (CBR), combined with a new genetic algorithm, the Genetic Algorithm with Territorial core Evolution (GATE). GATE was specifically created to solve problems in the mechanical structural design field, and is essentially a real number genetic algorithm that prevents new individuals from being born too close to previously evaluated solutions. The restricted area becomes smaller or larger during optimization to allow global or local searches when necessary. The CBR process uses a databank filled with every known solution to similar design problems. The closest solutions to the current problem in terms of specifications are selected, along with an estimated solution from an artificial neural network. Each solution selected by the CBR is then used to initialize the population of a GATE island. Our results show that CBR may significantly upgrade the performance of an optimization algorithm when sufficient preliminary information is known about the design problem. It provides an average solution 5.0% lighter than the average solution found using random initialization. The results are compared to other results obtained for the same problems by four optimization algorithms from the I-SIGHT 3.5 software: the sequential quadratic programming algorithm (SQP), the insular genetic algorithm (GA), the Hookes & Jeeves generalized pattern search (HJ) and POINTER. Results show that GATE can be a very good candidate for automating and accelerating the structural design of a gas turbine engine rotor disc, providing an average disc 18.9% lighter than SQP, 11.2% lighter than HJ, 23.9% lighter than GA and 4.3% lighter than POINTER, even when starting with the same solution set.
Automatic Case-Based Reasoning Approach for Landslide Detection: Integration of Object-Oriented Image Analysis and a Genetic Algorithm
