This research work presents useful framework and mechanism for integrated fault diagnostic system, or FDS. The proposed system is composed of three major subsystems: fault detection, root cause and consequence analyzer, and maintenance analyzer. Learning mechanisms are proposed to extract knowledge about deviations/failure modes from real time process and equipment monitoring data. Fault semantic network is proposed to represent failure modes and fault propagation models as integrated with process and equipment models. Qualitative rules are defined and associated with fault semantic networks for practical Actual maintenance findings are used to tune training data for more accurate fault detection and root cause and consequence analysis. Case study is used to illustrate the proposed idea.