M. Anisimov
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
1 records found
1
Efficient unloading of bulk materials at port terminals is essential for reducing demurrage and maritime logistics costs. However, designing high-performing grabs for bulk handling is challenging due to the large number of design parameters, the high computational cost of the grab-material interaction simulation, and the presence of significant uncertainties in real operations.
This work introduces a Bayesian framework for grab design optimisation, which employs probabilistic surrogate modelling to guide candidate selection iteratively. The algorithm selects points with the highest expected improvement in performance, accounting for the surrogate model’s predictive uncertainty. The proposed approach includes complex design constraints and enables efficient exploration across the design space, regardless of the number of parameters or optimisation objectives defined in an optimisation problem.
The algorithm outperformed a conventional offline optimisation method in a two-dimensional benchmark problem (a 20% better-performing design was reached in five iterations) and demonstrated rapid convergence in a high-dimensional optimisation involving nine design variables (a 5.5% improvement to the reference design was reached in 18 iterations). These results underscore its suitability for engineering optimisation problems where an optimal design must be reached with as few trial simulations as possible. ...
This work introduces a Bayesian framework for grab design optimisation, which employs probabilistic surrogate modelling to guide candidate selection iteratively. The algorithm selects points with the highest expected improvement in performance, accounting for the surrogate model’s predictive uncertainty. The proposed approach includes complex design constraints and enables efficient exploration across the design space, regardless of the number of parameters or optimisation objectives defined in an optimisation problem.
The algorithm outperformed a conventional offline optimisation method in a two-dimensional benchmark problem (a 20% better-performing design was reached in five iterations) and demonstrated rapid convergence in a high-dimensional optimisation involving nine design variables (a 5.5% improvement to the reference design was reached in 18 iterations). These results underscore its suitability for engineering optimisation problems where an optimal design must be reached with as few trial simulations as possible. ...
Efficient unloading of bulk materials at port terminals is essential for reducing demurrage and maritime logistics costs. However, designing high-performing grabs for bulk handling is challenging due to the large number of design parameters, the high computational cost of the grab-material interaction simulation, and the presence of significant uncertainties in real operations.
This work introduces a Bayesian framework for grab design optimisation, which employs probabilistic surrogate modelling to guide candidate selection iteratively. The algorithm selects points with the highest expected improvement in performance, accounting for the surrogate model’s predictive uncertainty. The proposed approach includes complex design constraints and enables efficient exploration across the design space, regardless of the number of parameters or optimisation objectives defined in an optimisation problem.
The algorithm outperformed a conventional offline optimisation method in a two-dimensional benchmark problem (a 20% better-performing design was reached in five iterations) and demonstrated rapid convergence in a high-dimensional optimisation involving nine design variables (a 5.5% improvement to the reference design was reached in 18 iterations). These results underscore its suitability for engineering optimisation problems where an optimal design must be reached with as few trial simulations as possible.
This work introduces a Bayesian framework for grab design optimisation, which employs probabilistic surrogate modelling to guide candidate selection iteratively. The algorithm selects points with the highest expected improvement in performance, accounting for the surrogate model’s predictive uncertainty. The proposed approach includes complex design constraints and enables efficient exploration across the design space, regardless of the number of parameters or optimisation objectives defined in an optimisation problem.
The algorithm outperformed a conventional offline optimisation method in a two-dimensional benchmark problem (a 20% better-performing design was reached in five iterations) and demonstrated rapid convergence in a high-dimensional optimisation involving nine design variables (a 5.5% improvement to the reference design was reached in 18 iterations). These results underscore its suitability for engineering optimisation problems where an optimal design must be reached with as few trial simulations as possible.