FAULT DETECTION AND DIAGNOSIS IN THE PARITY SPACE GENERATED BY MASS AND ENERGY CONSERVATION EQUATIONS

摘要

Processes combining heat and material flow involve non-linear relationships that complicate fault detection and diagnosis (FDD) procedures. The paper proposes a sequential model-based approach which is able to detect and diagnose fault in the process models as well as in the measurements, using only linear relationships. The models consist of stationary mass and energy conservation equations. The chi~2 detection and GLR diagnosis tests can be both analytically calculated. The method is illustrated for an iron oxyde pellet sintering furnace where the FDD procedure is tested for faults present on either mass or energy state equations.