SMWO/D: a decomposition-based switching multi-objective whale optimiser for structural optimisation of Turbine disk in aero-engines

In this paper, a novel multidisciplinary design optimisation (MDO) algorithm is proposed, which is named as the decomposition-based switching multi-objective whale optimiser (SMWO/D). In particular, a penalty-Tchebycheff value-based decomposition framework is designed to decouple the strongly correlated conflicting objectives, so as to give comprehensive considerations to different disciplinary demands. To overcome the shortcoming of premature in the complicated multi-modal non-linear decision space, two adaptively switchable evolutionary modes are defined to enhance the ability of escaping from local optimum and promote a thorough global search with rich learning strategies. The proposed SMWO/D is evaluated on a series of benchmark functions, and the results show its competitiveness in terms of comprehensive performance as compared with other four popular decomposition-based multi-objective optimisation algorithms (MOAs). In addition, sensitivity analysis is carried out to determine the best parameter configuration of SMWO/D. Finally, in a case study of a real-world turbine disk structural optimisation, the practicality of the proposed SMWO/D is validated, which can effectively handle the multidisciplinary property of this complicated problem, thereby providing valuable experiences in the aero-engine MDO domain.

Automated summary

In this paper, a novel multidisciplinary design optimisation (MDO) algorithm called the decomposition-based switching multi-objective whale optimiser (SMWO/D) is proposed. It utilizes a penalty-Tchebycheff value-based decomposition framework to decouple strongly correlated conflicting objectives and considers different disciplinary demands comprehensively. Two adaptively switchable evolutionary modes are defined to overcome the shortcoming of premature convergence in the complicated multi-modal non-linear decision space and promote a thorough global search with rich learning strategies. The proposed SMWO/D is evaluated on benchmark functions and compared with four popular decomposition-based multi-objective optimisation algorithms (MOAs), demonstrating its competitiveness in terms of comprehensive performance. Additionally, a sensitivity analysis is conducted to determine the best parameter configuration of SMWO/D. Finally, a case study of a real-world turbine disk structural optimisation validates the practicality of SMWO/D in handling multidisciplinary properties and provides valuable experiences in the aero-engine MDO domain.