A method may include controlling, by a computing device, a directed energy deposition material addition (DED MA) technique based at least in part on a thermal model. The thermal model may define a plurality of default operating parameters for the DED MA technique. The method also may include detecting, by at least one sensor, at least one parameter related to the DED MA technique. Further, the method may include, responsive to determining, by the computing device, that a value of the at least one detected parameter is different from an expected value of a corresponding parameter predicted by the thermal model, determining, by the computing device and using a neuro-fuzzy algorithm, an updated value for at least one operating parameter for the DED MA technique, and controlling, by the computing device, the DED MA technique based at least in part on the updated value.