Anchor bolt corrosion is a complex and dynamic system, and the prediction and identification of its corrosion degree are of significant importance for engineering safety. Currently, non-destructive testing using ultrasonic guided waves can be employed for its detection. Building upon the analysis of anchor bolt corrosion mechanisms, this paper proposes a method for evaluating the corrosion degree of anchor bolts based on multi-scale convolutional neural networks (MS-CNNs) that address the multi-mode propagation and dispersion effects of ultrasonic guided wave signals in non-destructive testing. Electrochemical experiments were conducted to simulate anchor bolt corrosion, and ultrasonic guided wave non-destructive testing was performed every 12 h to obtain waveform data. An MS-CNN was then utilized to accurately diagnose the corrosion degree of the anchor bolts. The test results demonstrate that this method effectively detects and diagnoses the extent of anchor bolt corrosion, facilitating timely troubleshooting and preventing potential safety accidents.

Feedforward neural network (FNN) is an information processing system that simulates human brain function to a certain extent by referring to the structure of biological neural network and the working mechanism of biological neurons. As the most important part, the architecture of FNN essentially influences its application performance. This paper proposed a structure self-adaptive algorithm for feedforward neural network (SSAFNN) based on biological mechanism. Firstly, self-adaptive neuron growing index and pruning index are proposed based on the idea of biological neuron grow factor and neuron competition, respectively. The FNN structure is dynamically adjusted according to the growing and pruning index of hidden neurons. Secondly, the connect weights of FNN are automatically adjusted during the self-organizing process and trained by gradient descent method during the learning process. Thirdly, the theoretical analysis is given that this proposed algorithm not only optimized the network structure, but also ensured the convergence and performance of FNN. The proposed SSAFNN is tested on the benchmark problems in the field of classification and prediction, and applied in the engineering problems of anchor bolt non-destructive testing and wastewater effluent ammonia nitrogen predicting. The experiment results reveal the good performance and potentiality of SSAFNN in industrial applications.

With the rapid development of high-speed railways, the issue of train speed tracking has become increasingly important. A controller based on a Long Short-Term Memory (LSTM) module was designed to meet the high-precision tracking requirements of trains, and its convergence analysis was analyzed. At the same time, in order to further improve the control performance on the basis of the original controller, an adaptive proportional compensation module was designed, and its effectiveness was theoretically proven. Finally, simulation validation was conducted on a multi-particle model with a large number of perturbations for both methods. The results show that the LSTM controller reaches good control results under external interference, and the adaptive proportional compensation module can further improve the control performance of the original controller.

Grout is an important part of the bolt anchorage system, once having the damage, the cohesive force will not reach the requirement, which may affect the support effect. Moreover, as buried in the rock, the defects happened in the grout have unknown characteristics. In order to identify the grout defect, a multi-scale convolutional generative adversarial network (MSCGAN) method is proposed in this paper. The generator based on supervised learning is designed in the MSCGAN framework, which can generate data for enriching the diversity of the system characteristics. Together with adopting the convolutional neural network which can realize end-to-end learning as discriminator, MSCGAN also draws on the idea of multi-scale to improve the ability of feature extraction. The proposed method was experimentally verified on the bolt anchorage system experimental platform. Through the comprehensive analysis for the experimental results of different methods, the effectiveness of the proposed method is demonstrated.

Electrical treeing is a defect of cable aging. Partial discharges within the insulation cause local electrical field anomalies and minor distortions, which act as initiation points for the formation of an electrical tree. As the electrical tree grows, it can eventually lead to insulation breakdown. Therefore, it is of practical importance to study the growth mechanisms of electrical treeing in cable insulation. The growth mechanism of electrical treeing in cross-linked polyethylene is analyzed from the perspective of its aggregation structure. Based on the growth morphology and the rate of electrical tree growth, the development process is divided into three stages: initial stage, rapid growth stage, and stagnant growth stage. The rapid growth stage is further categorized into the early and late stages. COMSOL finite element simulation software is used to model the entire to analyze the electric field distribution at each stage. In addition, the influence of carbon particles, varying in positions, sizes, and densities, on electrical treeing is studied to better understand its growth characteristics. The results showed that the electric field distortion is more serious when carbon particles are close to the tip of the needle and the tip of the electric treeing. At the same growth stage, the electric field intensity was negatively correlated with carbon particle size and positively correlated with density. When size and density are the same, the intensity of electric field increases gradually with the development of the growth stage. The effect of carbon particles on the insulation breakdown characteristics is further revealed.