Gasifiers operating in high-temperature, high-pressure, and highly corrosive working environments are prone to instrument measurement failures. The failure affects production processes such as coal-to-liquids and coal-to-methanol, and even endangers personnel safety. In order to solve the above problems, a global local preserving projection and geodesic distance gasifier based fault detection method were proposed in this paper. Firstly, the GLPP algorithm was adopted to extract the local features of the data determined by the sample neighborhood. Then, a geodesic distance measurement sample's non-neighbor relationship-based data global feature extraction method was proposed. Further, the extracted global features were used to construct corresponding statistics for fault detection. Finally, the effectiveness and feasibility of the proposed method were verified through two cases which were Tennessee Eastman (TE) and a real gasifier.