In this thesis, a progressive refinement approach is proposed for the use of rule-based systems in real time applications. In this approach, the knowledge involved Is constructed in a hierarchy of the levels. Part of this knowledge determines the time available for a decision. Progressive refinement allows the system to take corrective action by forming appropriate solutions within time constraints. An expert system shell written in Prolog is constructed to implement the progressive refinement framework. An English like language is also constructed, which allows the time constrained rules for the system to be specified in a user friendly manner. The framework can be applied to both discrete and continuous systems. An example is provided of the application of the framework to a discrete system. Experiments were also carried out to investigate the application of the framework to a continuous system, and their results are presented here.