绿色矿冶

Copper electrolytic refining is an important part of copper smelting, and electrode short circuit is unavoidable in the process of electrolysis. Therefore, this paper proposed a lightweight short-circuit detection algorithm based on YOLOv8 to effectively solve the problem of short-circuit detection under complex heat distribution. First, advanced technologies of ShuffleNet and SqueezeNet were integrated. A feature extraction module was added to the YOLOv8 framework, reducing model parameters while maintaining feature extraction capability, thereby improving the detection speed of the algorithm. Secondly, to solve the problem of insufficient detection accuracy and poor correlation, according to the characteristics of short-circuit data set, the loss function was optimized and each loss weight was reassigned, which effectively improved the detection accuracy. Finally, the experiment results show that the improved algorithm not only maintains high detection accuracy, but also reduces resource consumption, and the average accuracy (mAP) is increased to 0.854.