The line blocking procedure with vehicle interchange can be controlled via several parameters. The procedure is iterative, by first generating a number of solutions, which are then improved step by step. If there is no improvement within a specified number of iterations, convergence has been reached and the procedure is ended.
As a heuristic procedure, coincidence plays a decisive role, especially as there are many equivalent solutions. By using a random number generator, the procedure is deterministic in the sense, that each calculation with the same data and parameters achieve the same result. However, you can change the procedure by changing the parameter Random seed, and leave the other data unchanged to calculate alternative solutions.
These are the following parameters for controlling the procedure run:
|
Parameters |
Meaning and notes |
|
Use reference solution |
A reference solution is a block version containing blocks. If this option is selected, a solution is searched which can be compared with this reference solution. As reference solution, you should choose another block version than the one for which line blocks are currently being generated. |
|
Maximum number of iterations |
Number of iterations after the procedure is ended if convergence occurs. This value should be a multiple of the number of iterations without improvement. |
|
Number of iterations without improvement |
If for N iterations no improvement of the target function value is determined, the procedure regarded as converged and is ended. Reasonable values depend on the task size, however, they should not be less than 10 - 20. |
|
Number of solutions per iteration |
Number of simultaneous existing solutions per iteration. The more freedom the planning task offers, the greater this value should be. The minimum permissible value is 10, generally however 20 to 100. |
|
Random seed |
By changing this value, you change the random element of the procedure to obtain different results with the otherwise same data and parameters. |
