A nuclear reactor could be a suitable option in space needs such as planetary and space exploration. A low-enriched uranium fuel (LEU) could be utilized by member states of the Non-Proliferation Treaty (NPT) instead of Highly Enriched Uranium (HEU) fuel for security reasons. This paper deals with a conceptual design and optimization of low-enriched uranium (20 wt% enriched uranium in UO 2 fuel and 100 kWe) integral space nuclear reactor core according to the neutronic and dynamic analysis. The existing HEUfueled reactor is considered for comparison. For neutronic evaluation, different fuel mixtures and reflector thicknesses are considered to derive reactivity of excess, power peaking coefficient in radial direction, reactivity factor of fuel temperature, and reactivity factor of coolant temperature. An Adaptive Neuro-Fuzzy Inference System (ANFIS) is implemented for data estimation. This system is a machine learning method that combines the features of neural networks and fuzzy systems. The initial ANFIS parameters are optimized using optimization algorithms. The optimized ANFIS is coupled with the HGAPSO algorithm to derive an optimally designed fuel mixture and reflector thickness considering a cost function. The simulation results show that the coupled ANFIS-HGAPSO method has better performance and superiority (less error) compared with the ANFIS-GA and ANFIS-PSO methods. In addition, the optimized fuel mixture and core geometry have better performance compared to the present design. In each optimization process (fuel mixture and reflector thickness) the optimally designed model is validated utilizing neutronic computations. The overall optimal design of this reactor can provide 7000 days of full-power operation which is significantly more than the core cycle length of the existing HEU-fueled space nuclear reactor. In addition, this scheme has more reactor safety and stability because of a more negative reactivity factor of fuel temperature.