As a crucial technology for improving the autonomous sensing capability of optical sensors on unmanned aerial vehicles (UAVs), object detection has emerged as a research focus in the UAV patrol inspection. To address the problem of high computational complexity in detection models carried by UAV, a lightweight network model based on YOLO v7 is designed. Based on this model, a smoke detection algorithm is developed. Firstly, a new multi-parameter gated activation function (Better-mish) based on Mish function is constructed. Secondly, CBS and CA modules are improved based on the Better-mish. The Improved-CA is replaced with the Conv in MPConv to build CA-MPConv. The improved modules are added into backbone of the model. Thirdly, the backbone is reconstructed in combination with RepVGG module. An improved YOLOv7 lightweight model is proposed and a smoke detection algorithm based on this model is developed. Finally, several experiments are conducted on Smoke and COCO datasets. Experimental results show three key points as follows. 1) Compared with SiLU, Mish and other activation functions, Better-mish can alleviate the gradient disappearance of neural networks in the saturation region. 2) Compared with YOLO v7, the improved YOLO v7 model reduces computational complexity by 93.24 % and only increases model parameters by less than 2 %. 3) The algorithm can maintain effective recognition of smoke images at low computational complexity. The smoke detection algorithm provides a theoretical basis and an effective algorithm for the early warning of fire hazards in the UAV patrol inspection.

Detecting small objects within extensive remote sensing images captured by unmanned aerial vehicle (UAV) optical sensors is crucial for a variety of military and civilian purposes, but it poses considerable challenges. To address the difficulties associated with the numerous parameters and the high computational requirements of existing object detection models, as well as the inadequate performance in identifying small objects in remote sensing images (RSIs) captured by UAV optical sensors, this paper proposes a lightweight model for small object detection. Firstly, Focal Scalable Intersection over Union Loss (SIoU) loss derived by integrating the concept of Focal loss into the traditional SIoU loss. Secondly, this study utilizes channel attention (CA) and GhostNet to reconstruct the backbone. A layer specifically designed for detecting small objects is incorporated into YOLOv5s utilized a depth residual network neck. Furthermore, this study establishes a small fire dataset and evaluate this model on the small fire, AI-TOD dataset and DOTA dataset. Experimental results show that the proposed model achieves the lightweight goal while maintaining high detection performance for small objects.

A bucket structure optimization method based on compound digging trajectory and limit digging force was proposed for the problem that the real digging resistance of excavators is difficult to measure and difficult to use as bucket load for structure optimization. The uncertainty of the digging resistance in the actual digging process of the excavator was analyzed, and a digging trajectory was obtained by active digging using the compound digging action. Based on the limit digging force model, the digging force and its variation law were calculated, and analyzed for different hydraulic cylinder’s active digging in the trajectory. Based on the bucket shape line equation, a parametric model of the bucket was established using APDL language, and the bucket stress variation under different digging actions was analyzed. The bucket stress and mass were reduced by 4.59% and 5.84% respectively after optimization.

In order to more accurately analyze the dynamic characteristics of the working device of the hydraulic excavator. The load changes on the two digging trajectories were calculated and analyzed by using the limit digging force model and the theoretical digging force model, respectively. The rigid-flexible coupling model of the working device was established in ADAMS. Taking the limit digging force (LDF) and the theoretical digging force (TDF) as the external load of the working device, the dynamic simulation of the hinge force of the working device in the two trajectories was carried out, and the structural strength analysis of the bucket was carried out by using ANSYS. The results show that the tangential force of the LDF is generally larger than that of the TDF, the hinge point force of the working device changes dynamically with the external load of the tooling, and the influence of the LDF on the hinge point force is greater than that of the TDF. When the LDF is taken as the external load, the structural strength of the bucket meets the operational requirements.

The digging resistance in a normal state is the key to excavator design and automated excavation. It is difficult to accurately predict, simulate, or directly measure the digging resistance in a normal state due to uncertainties in the soil properties and excavation parameters. In this paper, a research idea is proposed that uses the working device as the entry point to indirectly calculate the digging resistance in a normal state by measuring the motion parameters and the cylinder pressure intensity. Based on the rule of combination for spatial force systems, a method for combining and projecting the system of the digging resistance is proposed in which the system is projected as six parts, and the tangential force, normal force, and bending moment in the plane of symmetry of the working device are the objects of the solution to avoid redundant equations. Based on kinematics and dynamics models of the excavator and the force and moment equilibrium conditions of the working device, equations for the active‐side calculation of the incomplete digging resistance are derived. Based on these equations, the motion parameters of the working device and data on the cylinder pressure intensity obtained by measurement are used to calculate the incomplete digging resistance. The validation scheme and process proposed use the incomplete digging resistance as the external load to obtain the simulated stress of the working device through transient analysis. The simulated stress and the measured stress corresponding to the position of the measurement point are extracted and compared. The results show that there is a difference in the size of the numerical value between the simulated and measured stress, but the variation law is highly consistent, which validates the calculation method. In this paper, an active‐side calculation method is provided for the incomplete digging resistance in a normal state without considering the soil‐tool interaction relationships, which lays a theoretical foundation for the study of the digging resistance characteristics in a normal state, as well as excavator design and automated excavation.

To improve the adaptability of the robot pan–tilt to the high-temperature environment, a design scheme for a composite thermal protection structure composed of aerogel felt, hollow glass, and skin is proposed. The effects of aerogel felt thickness, glass type, and ambient temperature on the thermal protection performance of the structure are studied, using a fluid–solid–thermal coupling model. Numerical results show that the structure exhibits good protection performance, and that the thermal resistance distribution changes the main path of heat transmission. The optimal thickness of the aerogel felt is approximately 8 mm. Compared to 3 mm, 5 mm, and 10 mm thicknesses, 8 mm reduces the maximum temperature by 15.90%, 8.37%, and 6.22%, and reduces the total entropy by 79.23%, 52.44%, and 12.5%. Lower thermal conductivity of the gas inside the hollow glass results in decreased maximum temperatures and total entropy. Using argon-filled hollow glass at 573.15 K decreases maximum temperature by 33.52% and 8.40%, with a total entropy reduction of 33.46% and 6.04%, compared to the single-layer and air-filled glass. Higher ambient temperatures correlate with increased maximum temperature, total entropy, and average surface-heat-transfer coefficient, indicating that the adaptability of the structure to high-temperature environments is limited.