Optimization design of a fast neutron imaging collimator by genetic algorithm

Fast neutron imaging, including radiography and computer tomography (CT), has been developed as a useful probe to visualize the inner structure of objects. From large scale neutron facility (like spallation source) to compact neutron source, even to small neutron generator, collimator design is extremely important for achieving high quality images. The main indexes to evaluate the performance of the fast neutron imaging collimator are: collimation ratio (L/D), outlet size (D0), fast neutron intensity (fF), uncollided neutron content (UNC), neutron to gamma ratio (n/γ). At present, the enumeration method is widely used for the collimator design from the reported studies. However, not only one (always several) parameters need to be optimized, as well as multiple optimization objectives should be considered, thus enumeration method can only try an extremely limited number of parameter combinations within the parameter space. In this paper, we proposed an optimization method combining genetic algorithm (GA) with MCNP code for the collimator design. The optimization results show the proposed method can significantly improve its performance compared with those with enumeration method.