Peihong Yang

In the context of resource constrained embedded terminals, the current deep learning-based technology for detecting surface defects in strip steel faces challenges with low detection accuracy and efficiency. This study introduces a lightweight convolution neural network (CNN) model for detecting surface defects of strip steel, which is based on deep learning. Through the integration of non-parameter attention mechanism SimAM, inverse residual structure, and depth-wise convolution module, a lightweight attention mobile backbone network is developed to achieve optimal feature extraction. To address the challenge of detecting small surface defects the context augmentation module is introduced to provide more information for small defects detection by using multi-scale features. To improve the efficiency of feature fusion and reduce parameter redundancy, the GVFPN neck network is proposed. The network aims to represent and deal with multi-scale features objectively while minimizing costs. The experimental results indicate that on the NEU-DET dataset, the proposed network attains an mAP of 78.8% and FPS at 97 frame/s. Moreover, the model requires only 2.39 M parameters and 3.1 G FLOPs. Compared to YOLOv5s, the proposed network reduces parameters by 65.9% and FLOPs by 80.7%, while achieving a 1.3% higher mAP and a 30 frame/s increase in detection speed. These results effectively demonstrate the effectiveness and superiority of the proposed method.

To reduce the carbon emissions of the Integrated Energy System ( IES ) and improve the consumption capacity of renewable energy, a joint mode of operation is proposed, taking into account carbon capture and storage (CCS), power to gas (P2G) and hydrogen fuel cells (HFC). Firstly, the two-stage process of P2G is refined, and a P2G-CCS-HFC coupling model is established to make full use of energy. Secondly, the reward and punishment carbon trading mechanism is introduced to constrain carbon emissions, and the unit output of each period is optimised with the lowest total operating cost as the objective function. The IES optimal scheduling model with P2G-CCS-HFC coupling is constructed and solved by GUROBI in MATLAB. Finally, three scenarios are set up for example analysis. The outcomes indicate a notable enhancement in renewable energy utilization efficiency and a marked reduction in carbon emissions, attributed to the implemented coupling model.

In industrial rare earth electrolysis production, due to the lack of effective prediction ability of cell voltage change, the problem of cell voltage swing cannot be dealt with in time. In order to achieve the prediction of cell voltage change and prevent the occurrence of cell voltage swing, this paper proposes a cell voltage prediction model based on the Improved Pelican Optimization Algorithm (IPOA) optimized Long Short-Term Memory Network (LSTM). Firstly, several improvement strategies are added to the Pelican algorithm to improve the optimization-seeking effect on the hyperparameters of the LSTM model and further improve the prediction accuracy of the neural network model. Secondly, according to the cell voltage, cell voltage change rate, and cumulative slope at different times, a cell voltage prediction model based on IPOA-LSTM is constructed. Finally, the comparison experiments show that the prediction accuracy and prediction error of the IPOA-LSTM model are better than other models, which can accurately predict the cell voltage, prevent the cell voltage swing, and ensure the normal operation of the rare earth electrolysis production.