The goal of this work is to build a novel diagnostic and prognostic system of aircraft, especially for system of aircraft’s engines. Different from classical diagnostic and prognostic system, like Condition-Based Maintenance (CBM) / Prognostic Health Management (PHM), this system can be used on aircraft not only off-line for maintenance, but also at on-line during its flight mission. Machine learning of the system will be done with more powerful computer at Aircraft Ground Center or Maintenance Center. According to the different situations requirement and different equipment’s conditions, two monitoring terminals are built: one is on operation at the aircraft on-line, which only needs to make 2 judgments of the health of aircraft - normal (contains the small faults which fault tolerant system can resolve) and dangerous (-back to airport) for pilot; another one is for the maintenance office, which needs detailed diagnosis results and to forecast the aircraft health. In comparison with some other methods, Support Vector Machines (SVM) is more convenient and stable. Unlike expect system or others methods, it can easily add new faults and new rules into database. In addition, with Principal Component Analysis (PCA), it can also make a visualization of evaluation of flux data with haut dimension and of their boundary, which is used to realize prognosis and useful for engineers to study faults. This system should not be a black box. Indeed, it is designed to illuminate engineering consulting services, and to accumulate the knowledge for re-engineering purposes (including diagnosis operational rules) and the design of new aircrafts. The system has been fully tested by using actual experimental data from Dassault Aviation.