Bayesian Support Vector Regression for Reliability-Based Design Optimization

Ling, Chunyan; Lü, Zhenzhou; Zhang, Wenxin

doi:10.2514/1.j060567

Cited by 11 publications

(8 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These individuals are divided into infeasible individuals (circular points) that do not meet the constraints and feasible individuals (triangular points) that satisfy the constraints. By calculating the objective function values of these feasible individuals, we can obtain the current minimum of the objective function (f Ã (1) ) as well as the current optimal design (z Ã (1) ). Then, based on the information provided by current feasible individuals, the new-generation individuals (green points) are produced via VBI.…”

Section: Proposed Non-directional Optimization Approachmentioning

confidence: 99%

“…This will expand the feasible domain. We can find a better solution (z Ã (2) ) that achieves smaller objective function value (f Ã(2) ), compared with the optimal solution (z Ã (1) ) obtained in the last iteration. In the same vein, the third and fourth-generation individuals (red points and pink points) can be generated by VBI based on the acquired information of the existing Algorithm 1.…”

Section: Proposed Non-directional Optimization Approachmentioning

confidence: 99%

“…Moreover, the ever-harsher work environment together with the increasing awareness of high reliability, creates challenging optimal reliability design for complicated and expensive systems, forcing it to be a hot spot of intensive concern in both academia and industry. 1,2 System reliability can be enhanced by providing redundancy at the component level and/or increasing component reliabilities. [3][4][5] This, without any doubt, increases the system budget.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A sequential two-stage approach based on variational Bayesian inference for reliability-redundancy allocation

Ling

Lei

Kuo

2022

Proceedings of the Institution of Mechanical Engineers, Part O:

Self Cite

View full text Add to dashboard Cite

The reliability-related design has been a crucial chain in complex and reliability-critical engineering systems. It serves as a preventive countermeasure to catastrophic failures caused by uncertainties. To keep up with this rapidly developing field, this paper presents a novel metaheuristic, based on the variational Bayesian inference (VBI) to efficiently solve the optimal reliability design. Specifically, the reliability-redundancy allocation problem (RRAP). The proposed metaheuristic starts from a primary population, then leverages VBI to fully excavate the information of feasible individuals from the previous generation, in order to produce the next-generation population. This process is iterated until the solution converges to the optimal decision scheme. In addition, we set up an automatic stratification strategy, so that the new individuals can approach the optimal solution faster. Furthermore, we divide RRAP into a reliability optimization problem (ROP) and a redundancy allocation problem (RAP). This not only reduces the dimension of decision variables, but also speeds up the convergence. ROP and RAP are solved sequentially and iteratively until the preset stopping condition is satisfied. The case studies showcase that the proposed approach can obtain the optimal or near-optimal solution within a reasonable period.

show abstract

Section: Proposed Non-directional Optimization Approachmentioning

confidence: 99%

Section: Proposed Non-directional Optimization Approachmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A sequential two-stage approach based on variational Bayesian inference for reliability-redundancy allocation

Ling

Lei

Kuo

2022

Proceedings of the Institution of Mechanical Engineers, Part O:

Self Cite

View full text Add to dashboard Cite

show abstract

“…So far, RBDO has turned out to be a prevalent design technique in diverse fields. [1][2][3] Solving RBDO generally involves a nested procedure that the reliability analysis process (inner loop) is repeatedly executed by the optimization process (outer loop). Clearly, when the dimension of the investigated problem increases, or some computer models (e.g., the finite element model) are introduced, such a nested procedure will cause huge computational burden, rendering the brute-force approach useless in practice.…”

Section: Introductionmentioning

confidence: 99%

“…Treating uncertain parameters as random variables in the optimization procedure, the reliability‐based design optimization (RBDO) came into being to obtain the optimal decision scheme that satisfies the reliability requirement. So far, RBDO has turned out to be a prevalent design technique in diverse fields 1–3 . Solving RBDO generally involves a nested procedure that the reliability analysis process (inner loop) is repeatedly executed by the optimization process (outer loop).…”

Section: Introductionmentioning

confidence: 99%

A global strategy based on deep learning for time‐dependent optimal reliability design

Ling

Kuo

2023

Quality & Reliability Eng

Self Cite

View full text Add to dashboard Cite

Time‐dependent reliability‐based design optimization (RBDO) is a computationally tough problem that needs to be addressed urgently. The difficulty of solving the time‐dependent RBDO mainly comes from the time‐dependent reliability analysis involved in probabilistic constraints, which itself is one of the thorny problems in the reliability community and makes the computational cost become much more onerous. In this paper, a deep‐learning‐assisted approach is proposed to solve the time‐dependent RBDO. The proposed approach leverages the classification capability of the deep learning, and constructs the alternative model for the actual probabilistic constraint function in the so‐called augmented reliability space, so as to make the trained alternative model accurate wherever it will be invoked. Moreover, a sequential sampling technique utilizing the classification probability provided by the deep learning is proposed to further reduce the computational cost. Then, the time‐dependent reliability analysis involved in the time‐dependent RBDO is conducted by the cheaper alternative model instead of the original computing‐intensive probabilistic constraint function, which evidently reduces the computational burden. The presented examples showcase the performance of the proposed approach. Especially, for the complicated engineering application, the proposed approach saves about 10% of the computational cost compared with the existing methods.

show abstract

A novel training point selection strategy guided by the maximum reduction of structural state misclassification probability for time-dependent reliability analysis

Lü

Feng

2023

Aerospace Science and Technology

View full text Add to dashboard Cite

Bayesian Support Vector Regression for Reliability-Based Design Optimization

Cited by 11 publications

References 29 publications

A sequential two-stage approach based on variational Bayesian inference for reliability-redundancy allocation

A sequential two-stage approach based on variational Bayesian inference for reliability-redundancy allocation

A global strategy based on deep learning for time‐dependent optimal reliability design

A novel training point selection strategy guided by the maximum reduction of structural state misclassification probability for time-dependent reliability analysis

Contact Info

Product

Resources

About