Reliability-based Design Optimization Of Mechanism And Its Application In Space Tether Considering Hybrid Uncertainty

There exist various uncertainties in mechanical or structural systems and reliabilityis one of the most important indices to evaluate the performance of mechanical systemunder uncertain conditions, as well that optimization design is a effective method formodern design. With the development tendency of high-precision, high-speed,heavy-load and light-weight for complex mechanical systems, the reliabilityrequirements of mechanism reliability is rapidly increasing. In order to address suchproblems as low model versatility, high computational complexity and heavycomputation workload, this paper mainly emphasizes the method of reliabilityoptimization design in such cases. Reliability based design optimization (RBDO) canobtain ideal results under aleatory uncertainty, and this methodology has becomeincreasingly mature. While under epistemic uncertainty, such as lack of knowledge,fuzzy information or insufficient data, possibility based design optimization (PBDO)has presented its own advantages, which is still in the process of development.Considering both aleatory and epistemic uncertainty, there is no a general formulationor framework in reliability design optimization. Based on this, referring to thedeployable space tether as the engineering background, by integrating the probabilitytheory and possibility theory, the thesis aims to investigate and study a methodologyframework that can deal with hybrid uncertainty factors in design optimization, whichcan provide technical support for mechanical equipment design. The main researchcontents are illustrated as follows:(1) Based on probability and possibility theories, the reliability modeling andanalysis methods are introduced respectively. Correspondingly, the RBDO and PBDOmodels and efficient algorithms are also provided. Based on this, MVDO model is thusproposed to deal with both aleatory and epistemic uncertainty, and a efficient methodbased on Sequential Optimization and Reliability Assessment τSORAυ is proposed tosolving the MVDO model. Some typical examples verify the effectiveness andfeasibility of the proposed methods. (2) When the limit state function is nonlinear implicit function, this thesis firstcombines such methods as design of experiment, finite element analysis with BP neuralnetwork, and then presents a high accuracy global surrogate model based on BP neuralnetwork. After considering the error of surrogate model as an uncertain variable, thethesis also shows the process of validation and verification of the model accuracy basedon the hypothesis test theory. Numerical examples of typical linkage mechanism verifythat this method is feasible and has a higher solution accuracy.(3) Considering deployable space tether as the application object in space fields,this thesis firstly analyses the performance and structural characteristics, and uncertainfactors of the deployable mechanism involving loads, parameters and computing models,establishes the neural network surrogate model of the deployable mechanism’s strength,and then provides the reliability design optimization model of the deployablemechanism. In the end, the obtained analysis results can provide some design guidancefor practical engineering application.