The job shop scheduling problem is a classical combinatorial optimization problem which has received a lot of attention owing to its wide applicability in the real world. One of the popular solution approaches to this NP-hard problem is to use dispatching rule heuristics. In the real world, uncertainty is ubiquitous in the scheduling environments and dealing with it makes the scheduling problem even harder. Particularly, in developing dispatching rule heuristics for a dynamic and uncertain job shop scheduling environment, following issues arise: (1) a large number of dispatching rules are needed for varied shop scenarios. (2) a dispatching rule which is robust to uncertainty must pay the price of robustness with a drop in the desired quality of scheduling objective. (3) it typically requires long computation time for generating the heuristics. (4) it lacks good approaches which deal with multiple objectives of scheduling simultaneously, under multiple sources of uncertainty. The overall goal of this thesis is to develop efficient hyper-heuristic approaches to generating robust dispatching rules toward solving the dynamic job shop scheduling problem which satisfy multiple objectives under different types of uncertainty.