This paper presents a mechanism for developing knowledge based support in multiagent based control and diagnosis. In particular it presents a way for autonomous agents to utilize a qualitative means-ends based model for reasoning about control situations. The proposed mechanism have been used in different scenarios of electric power distribution system protection and control. Results show that agents can use local models of their environment and coordinate with other agents to analyze and understand a disturbance situation and choose an appropriate control action. The paper also introduces Multi Level Flow Modeling (MFM) which has been used to model agent environment and describes development of multiagent implementation of MFM Workbench.