Non-ferrous metals are very important strategic resources, electrolysis is an essential step in refining non-ferrous metals. In the electrolysis process, plate short circuit is the most common fault, which seriously affects output and energy consumption. The rapid and accurate detection of faulty plates is of great significance to the metal refining process. Given the weak generalization ability and complex feature rule design of traditional object detection algorithms, and the poor detection effect of existing deep learning models in infrared images with many interference factors, an improved Mask R-CNN based fault detection algorithm is--proposed. Improve the generation strategy and the NMS algorithm of proposals to reduce the missed detection; propose the Globally Generalized IoU loss function to characterize better the position and scale relationship between the predicted box and the target box, which is beneficial to the bounding box regression. The experimental results show that the improved model has an accuracy rate of 10.4% higher than the original model, reaching 86.8%. Compared with the common one-stage and two-stage object detection models, the improved model has a stronger detection ability. This algorithm has some reference value for the accurate detection and location of electrolytic cell faults.

To solve the problem of intercepting a moving target by a multirotor unmanned aerial vehicle (UAV) swarm, an optimal guidance strategy is proposed. The proposed guidance law is based on the integration of the classic pure pursuit guidance law and Kuhn-Munkres (KM) optimal matching algorithm, and virtual force potential functions are used to avoid collision. The proposed optimal guidance strategy is demonstrated by simulation experiments. The simulation results indicate that with the proposed optimal guidance strategy, a UAV swarm can intercept a moving target while maintaining the predetermined formation, and during the formation flight, the collisions between UAVs or the target can be avoided. Through a comparative experiment, the proposed optimal matching algorithm is proven to significantly reduce the average per-sampling-period total flight distance of all the UAVs and accelerate the interception process, and the formation completion degree is improved. INDEX TERMS Optimal matching; Unmanned aerial vehicles; Pursuit algorithms; Three-dimensional guidance law; Target interception; Collision avoidance. Xi WANG received his B.E. degree in electrical engineering in 2011 and MA.Eng. degree in control science and engineering in 2014, from Hunan University of Science and Technology. He is currently a doctoral student in the School of Automation, Central South University, China. His research areas are unmanned aerial vehicles, swarm intelligence, and related applications.

In the mobile wireless sensor network (MWSN) field, there exists an important problem—how can we quickly form an MWSN to cover a designated working area on the ground using an unmanned aerial vehicle (UAV) swarm? This problem is of significance in many military and civilian applications. In this paper, inspired by intermolecular forces, a novel molecular force field-based optimal deployment algorithm for a UAV swarm is proposed to solve this problem. A multi-rotor UAV swarm is used to carry sensors and quickly build an MWSN in a designated working area. The necessary minimum number of UAVs is determined according to the principle that the coverage area of any three UAVs has the smallest overlap. Based on the geometric properties of a convex polygon, two initialization methods are proposed to make the initial deployment more uniform, following which, the positions of all UAVs are subsequently optimized by the proposed molecular force field-based deployment algorithm. Simulation experiment results show that the proposed algorithm, when compared with three existing algorithms, can obtain the maximum coverage ratio for the designated working area thanks to the proposed initialization methods. The probability of falling into a local optimum and the computational complexity are reduced, while the convergence rate is improved.

Copper is an important resource of non-ferrous metals. Electrolytic refining is one of the main methods to produce fine copper.  In electrolytic process, plate conditions seriously affects the output and quality of copper.  Timely and accurate prediction of the working condition of the plate is of great significance to the copper electrolytic refining process.  Aiming at the problems of the traditional plate conditions detection algorithm with large lag, poor anti-interference ability and low accuracy, this paper proposes a plate condition prediction model based on LSTM with attention mechanism.  The average gray value of the plate in the infrared image is used to characterize the plate working condition.  In hidden layer, double layer LSTM network is used to improve the effect of model training.  A unique attention mechanism is added to learn the correlation between input and output better and improve the accuracy of model prediction.  Experimental results show that the accuracy of the proposed model for plate condition prediction reaches 95.11%.  Compared with the commonly used LSTM algorithm, the plate condition prediction model proposed in this paper has stronger prediction ability.