Generalising and re-using knowledge learned while solving one problem\ninstance has been neglected by state-of-the-art answer set solvers. We suggest\na new approach that generalises learned nogoods for re-use to speed-up the\nsolving of future problem instances. Our solution combines well-known ASP\nsolving techniques with deductive logic-based machine learning. Solving\nperformance can be improved by adding learned non-ground constraints to the\noriginal program. We demonstrate the effects of our method by means of\nrealistic examples, showing that our approach requires low computational cost\nto learn constraints that yield significant performance benefits in our test\ncases. These benefits can be seen with ground-and-solve systems as well as\nlazy-grounding systems. However, ground-and-solve systems suffer from\nadditional grounding overheads, induced by the additional constraints in some\ncases. By means of conflict minimization, non-minimal learned constraints can\nbe reduced. This can result in significant reductions of grounding and solving\nefforts, as our experiments show. (Under consideration for acceptance in TPLP.)\n