A stochastic optimization tool for determining spacecraft avionics box placement

In the preliminary design of spacecraft, one particularly difficult task is determining the optimal placement of avionics boxes on the spacecraft panels and decks. This is actually a multi-objective optimization problem, as there are multiple competing constraints that must be satisfied simultaneously. These constraints include minimizing the amount of harness wiring between boxes (and thus the wire harness mass), minimizing the length of RF cable runs (to minimize attenuation), keeping the thermal loading of all panels/decks within prescribed limits, and keeping the mass imbalance of the spacecraft within prescribed limits. This task is generally performed manually, based on prior experience and similarity to previous designs. This type of manual process tends to be highly iterative, wastes valuable time and resources, and is guaranteed to always produce sub-optimal results. As the complexity of the spacecraft increases, this problem becomes increasingly formidable. The avionics box placement problem is shown to be a variant of the classical traveling salesman problem (TSP), which is a well-known problem in combinatorial optimization. The classical TSP and its variants are of the class NP-hard, and thus cannot be solved to optimality in polynomial time due to the vastness of the solution space. Global search techniques that use a stochastic engine to explore diverse regions of the solution space (such as genetic algorithms and simulated annealing) have been employed with great success against such problems. This paper examines the utility of implementing a box placement optimization tool using a stochastic global search algorithm. A candidate algorithm is presented, run against a simplified representative avionics box placement problem, and the results documented. The utility of the algorithm is assessed, and recommendations are made for additional enhancements that would increase algorithm performance and make the algorithm more suitable for actual applications.