Abstract: Because of complex conditions of the burning zone, alumina rotary kiln production control has depended on man-watch operation mode for many years, which causes problems of weak coherence of product quality and big depletion of resources. A recognition method for burning zone conditions based on flame images and process data is put forward in this paper, with data fusion techniques and pattern recognition techniques. The method consists of four parts: burning zone flame image segmentation, feature extraction, process key data fusion, and design and parameters selection for classifier model based on the support vector machine. The industrial experiments prove that the burning zone condition can be recognized correctly with the proposed method, which provide decision basis to kiln temperature controller for product quality indices optimization.