Fault detection and diagnosis (FDD) is applied to mechanical-pneumatic systems to perform intelligent diagnosis of various faults in the system by utilizing the sensory information commonly found in typical systems, such as pressures and flow rates. In this paper, we present research results on intelligent FDD and characterization of MEMS flow sensor. Vectorized maps are created and calibrated for the purpose of intelligent FDD. In addition, maps of N-manifold can be used for redundancy in diagnosis to improve the accuracy and reliability of the methodology. Such redundant vectorized maps provide for explanation of physical significance of the behavior of the system and the formation or detection of faults. As a result, both physical-based and signal-based intelligent fault detection and diagnosis techniques and methodology can be applied for various types of applications. Experimental results suggest that intuitive choices of parameters and features, based on the understanding of physics of the mechanical-pneumatic system, can be applied with success to intelligent detection and diagnosis of faults. Furthermore, with miniaturization, sensors can be readily made and integrated for intelligent diagnosis. Characterization and modeling of such innovative sensor designs are presented. Using new smart multi-function, telemetric, and integrated sensors as "intelligent nodes" in systems will provide necessary sensory information (e.g., pressure, flow, and temperature) for the next-generation diagnosis. The characterization and study of MEMS sensor include: correlation of flow and deflection of sensory element, analysis and modeling, vibration characteristics, fatigue tests, backflow characterization,... etc. Specifically, the results of fatigue tests provide information and feedback for the design and fabrication of the MEMS sensors; more importantly, long fatigue life is essential for the flow sensors to sustain as a transducer. Results of the findings are presented.