Vision-Based Fan Speed Control System in the Copper Scraps Smelting Process

This paper presents a vision-based system to control the draught fan speed in the copper scraps smelting process. First, the proposed vision-based system captures exhaust gas images and extracts them. Then, the image features can be classified by the k-means clustering method. For each cluster, a least square support vector regression (LSSVR)-based fan speed estimation model is trained to correlate image features and expected fan speed commands. Sequentially, fan speed commands can be estimated by the models developed in the testing stage. Finally, the vision-based system is applied in situ to generate fan speed control commands. The experiment results indicate that the proposed vision-based control system can automatically regulate the fan speed with good accuracy and real-time performance. The practical application results realize a 30% energy consumption when compared to the artificial control manner.