Abstract-A self-adaptive system reacts to the changing environment by modifying its functionality in relation to the encountered state of the environment. In order to adapt to a new situation, such system goes through many decision points during the adaptation process. Knowledge forms the basis of decision making within the adaptation process. There are already many existing self-adaptive system frameworks. However, these frameworks have limitation in the way they represent the rationale for adaptation and the semantics behind the knowledge they use. This paper takes a step forward by proposing a knowledge-intensive adaptation framework to both manage knowledge and support the analytical decision making process. The proposed approach represents the adaptation knowledge by using ontology which helps to organize, analyze and extend knowledge. Ontology is able to represent the semantics behind knowledge and provide the evidence for the adaptation. The proposed approach uses a special ontology named the Adaptation Problem Domain Ontology. It specifies the system goals, features, architectures, and the relationship between them. This ontology is used to answer the problem of adaptation at each decision point and determine the appropriate system structure by reasoning the semantics behind knowledge. Thus, the system can consider the semantics behind knowledge for adaptation, and then the stakeholders can understand the adaptation process. We apply the proposed framework to the smart grid domain and show how the system adapts to a new situation using rationale for adaptation and the semantics behind the knowledge.Index Terms-Self-adaptive system, decision making, ontology, goal model, feature model, role-based architecture