Mitsubishi Heavy Industries Technical Review
    Vol. 56 No. 2 (2019)   M-FET
    Technical Papers

    Advancements of Predictive Diagnosis System for Power Generation Engines

    ICHIRO ISHIDA
    OSAMU NAKAKITA
    TAKAAKI TAKAHASHI
    TAKASHI KADO
    KAZUYUKI WAKASUGI
    JUNNOSUKE ANDOU

    Mitsubishi Heavy Industries Engine & Turbocharger, Ltd. (MHIET) has been engaged in developing predictive diagnosis systems to further help customers realize the stable operation of their power-generating facilities. When it comes to engines for power generation, the application of commonly-used predictive diagnosis techniques has been considered to be difficult, because the power output of these engines is not constant owing to their operating patterns. We developed a new predictive diagnosis system by taking the following three approaches: "the statistical method (i.e., vector quantization clustering or VQC)," "big data analysis" and "the utilization of knowledge from past abnormalities." The developed system is now in operation. By the end of 2018, we obtained solid outcomes indicating a high likelihood of achieving a practically-sufficient correct prediction rate. The predictive diagnosis service will be made available commercially in 2019.