Robust Flame Video-based Sintering Condition Recognition Method Based on Fractal Theory

The sintering condition of the rotary kiln reflects the characteristics of product quality and energy saving. Due to the harsh environment of coal-fired kilns and complex production processes, it is difficult to detect the sintering condition in industrial site. To solve this problem, we propose a robust flame video-based sintering condition recognition method based on fractal theory. First, we extract the intensity sequence from the flame video and analyze the sequence. We find that some denoising methods may remove some chaotic fractal features instead of noise, resulting in the loss of part of the information in the recognition of the sintering condition. Therefore, we decide to develop a recognition algorithm, which can be directly applied to the original data without denoising. The algorithm uses the wavelet p-leader method and constructs several new indicators to identify the sintering condition. Experimental results on real-world data show the effectiveness of the proposed method. Compared with other methods, the recognition accuracy is improved.