Steel surface defect detection is crucial for ensuring steel quality. The traditional detection algorithm has low detection probability. This paper proposes an improved algorithm based on the YOLOv5 model to enhance detection probability. Firstly, deformable convolution is introduced in the backbone network, and a traditional convolution module is replaced by deformable convolution; secondly, the CBAM attention mechanism is added to the backbone network; then, Focal EIOU is used instead of the CIOU loss function in YOLOv5; lastly, the K-means algorithm is used to cluster the Anchor box, and the Anchor box parameters that are more suitable for this paper are obtained. The experimental results show that using deformable convolution instead of traditional convolution can get more feature information, which is more conducive to the learning of the network. This paper uses the CBAM attention mechanism, and the heat map of the attention mechanism shows that the CBAM attention mechanism is beneficial for feature extraction. Focal EIOU is optimized in high and wide loss compared with the CIOU loss function, which accelerates the convergence of the model. The Anchor box is more favorable for feature extraction. The improved algorithm achieved a detection probability of 78.8% in the NEU-DET dataset, which is 4.3% better than the original YOLOv5 network, and the inference time of each image is only increased by 1 ms; therefore, the optimized algorithm proposed in this paper is effective.

The construction hoist drive system is a critical component of the construction hoist, and high speed and low vibration noise are essential development directions. In order to improve the NVH level of the construction hoist drive system, this paper carries out research and analysis of construction hoist drive system excitation, establishes the drive system rigid-flexible coupling dynamics model, and completes the establishment of the vibration and noise model of the drive system, simulation analysis, and optimization work. Ansys Motor CAD 2020 was used to establish the parametric model of the asynchronous motor and it was combined with the virtual work method to calculate Maxwell’s electromagnetic force to arrive at the radial electromagnetic force as the main cause of electromagnetic noise. For the mechanical excitation generated by the gearbox, the time-varying stiffness excitation, mesh shock excitation, and transmission error excitation are considered, and the transmission error of helical gears under different working conditions is calculated by combining it with Romax software 2020. The rigid-flexible coupling model of the construction hoist drive system is established. The load distribution analysis of the unit length of the tooth surface is completed for the first- and third-stage helical gears under different working conditions. The primary source of the drive system excitation is the tooth surface bias load. Based on the dynamic response analysis theory of the vibration superposition method, the maximum vibration speed of the drive system was analyzed by Romax. The maximum noise value of 78.8 dB was calculated from the acoustic power simulation of the drive system using Actran acoustic software 2022 in combination with acoustic theory, and the magnetic density amplitude of the stator teeth of the asynchronous motor was optimized based on the microscopic shaping design of the helical gear by Romax. The vibration and noise simulation of the optimized drive system shows that the vibration value is reduced to 0.75 mm/s, and the maximum noise is reduced to 70.2 dB, which is 10.9% lower than before the optimization. The overall NVH level has been improved. The optimization method to reduce the vibration noise of the drive system is explored, which can be used for vibration noise prediction and control during the development of the construction hoist drive system.