Model Optimization
}
Do you have any questions about Dlubal products or need assistance in selecting the right one for your project?
I'm here to help. You can easily reach me through the contact options provided below.
Looking forward to hearing from you!
Model Optimization Using Artificial Intelligence (AI)
Model Optimization | Features
- Artificial intelligence technology (AI): Particle swarm optimization (PSO)
- Structure optimization according to the minimum weight or deformation
- Use of any number of optimization parameters
- Specification of variable ranges
- Optimization of cross-sections and materials
- Parameter definition types
- Optimization | Ascending or Optimization | Descending
- Application of parametric models and blocks
- Code-based JavaScript parametrization of blocks
- Optimization taking into account the design results
- Tabular display of the best model mutations
- Real-time display of the model mutations in the optimization process
Model Optimization | Input
Did you know? The structural optimization in the programs RFEM and RSTAB is a completion of the parametric input. It is a parallel process beside the actual model calculation with all its regular calculation and design definitions. The add-on assumes that your model or block is built with a parametric context and is controlled in its entirety by global control parameters of the "optimization" type. Therefore, these control parameters have a lower and upper limit and a step size to delimit the optimization range. If you want to find optimal values for the control parameters, you have to specify an optimization criterion (for example, minimum weight) with the selection of an optimization method (for example, particle swarm optimization).
Model Optimization | Calculation
Two methods are available to you for the optimization process, allowing you to find optimal parameter values based on a weight or deformation criterion.
The most efficient method with the shortest calculation time is nature-inspired particle swarm optimization (PSO). Have you heard or read about it? This artificial intelligence technology (AI) has a strong analogy to the behavior of animal swarms searching for a resting place. In such swarms, you find numerous individuals (cf. optimization solution - e.g., weight) that like to stay in a group and follow the group's movement. Let's assume that each swarm member has the need to rest at an optimal resting place (cf. best solution - e.g., lowest weight). This need increases as they approach the resting place. Thus, swarm behavior is also influenced by the properties of the space (cf. result diagram).
Why the excursion into biology? Quite simply, the PSO process in RFEM or RSTAB proceeds similarly. The calculation run begins with an optimization result from a random allocation of the parameters to be optimized. It continuously determines new optimization results with varied parameter values based on the experience of previously made model mutations. This process continues until the specified number of possible model mutations is reached.
Alternatively to this method, a batch processing method is also available in the program. This method attempts to check all possible model mutations by randomly assigning values to the optimization parameters until the specified number of possible model mutations is reached.
Both variants also check the activated design results of the add-ons after calculating a model mutation. Furthermore, they save the variant where the utilization is < 1 along with the associated optimization result and parameter value allocation.
Model Optimization | Result
Both optimization methods have one thing in common. At the end of the process, they provide you with a list of model mutations from the stored data. Here you can find the details of the controlling optimization result and the associated value assignment of the optimization parameters. This list is organized in descending order. You can find the assumed best solution shown in the first line. For this, the optimization result with its determined value assignment is closest to the optimization criterion. All add-on results have a utilization < 1. Furthermore, once the analysis is completed, the program will adjust the value assignment to that of the optimal solution for the optimization parameters in the global parameter list.
.png?mw=200&hash=ceabdac71454991b172d1a40faa3844b9a756ff1)
Calculate Your Price
